Journal of The American Society for Mass Spectrometry

, Volume 22, Issue 9, pp 1501–1507

Quantitative Analysis of Therapeutic Drugs in Dried Blood Spot Samples by Paper Spray Mass Spectrometry: An Avenue to Therapeutic Drug Monitoring

Research Article

Abstract

A method is presented for the direct quantitative analysis of therapeutic drugs from dried blood spot samples by mass spectrometry. The method, paper spray mass spectrometry, generates gas phase ions directly from the blood card paper used to store dried blood samples without the need for complex sample preparation and separation; the entire time for preparation and analysis of blood samples is around 30 s. Limits of detection were investigated for a chemically diverse set of some 15 therapeutic drugs; hydrophobic and weakly basic drugs, such as sunitinib, citalopram, and verapamil, were found to be routinely detectable at approximately 1 ng/mL. Samples were prepared by addition of the drug to whole blood. Drug concentrations were measured quantitatively over several orders of magnitude, with accuracies within 10% of the expected value and relative standard deviation (RSD) of around 10% by prespotting an internal standard solution onto the paper prior to application of the blood sample. We have demonstrated that paper spray mass spectrometry can be used to quantitatively measure drug concentrations over the entire therapeutic range for a wide variety of drugs. The high quality analytical data obtained indicate that the technique may be a viable option for therapeutic drug monitoring.

Key words

DBS API Ambient ionization TDM Point of care POC Small molecule Pharmaceuticals Quantitation Quantification 

1 Introduction

Even when dosing guidelines are strictly adhered to, the drug exposure, or the amount of drug circulating at any particular time, can vary significantly from patient to patient due to factors such as genetic variation, renal or hepatic impairment, and the presence of interfering drugs [1]. Direct measurement of therapeutic drug levels post-dose is known to be of significant value as a means of dose optimization, particularly with those drugs that have a narrow therapeutic window and high inter-patient variability. The use of therapeutic drug monitoring (TDM) to prevent over- or under-dosing is limited, however, by a lack of available drug assays with adequate performance that can also be implemented rapidly in a clinical setting.

The methods most commonly used for TDM currently are immunoassays and high performance liquid chromatography (HPLC) with spectroscopic or mass spectrometric (MS) detection. The application of HPLC-MS to the analysis of clinical specimens has been limited due to the cost of the equipment and the training required for its operation. Immunoassays, on the other hand, have been widely implemented in clinical settings for the quantitation of small molecule therapeutic drugs. While widely used, the shortcomings of immunoassays have been well-documented, particularly issues related to antibody selectivity [2], antibody saturation [3], limited dynamic range, and long development time.

Paper spray is a new ionization method that allows the quantitative analysis of pharmaceutical drugs by mass spectrometry directly from dried blood or other biofluid samples without prior sample preparation or separation [4, 5, 6]. Briefly, blood or another biofluid is first deposited on paper and allowed to dry, a process that is similar to that used in the screening of newborns for inborn errors of metabolism. After the blood is dry, the paper is cut to a sharp point, and a small volume (ca. 25 μL) of solvent (selected to effectively extract the drug) is applied to the paper so that it flows through the dried blood spot (DBS) sample by capillary action. A high voltage (3–5 kV) is applied to the moist paper, inducing an electrospray at the sharp tip of the paper; the solvent evaporates from the droplets generating gas phase ions of the analyte molecules, which can then be detected by a mass spectrometer [7]. The entire analysis requires about 30 s and, aside from an unmodified mass spectrometer, requires nothing more than the paper substrate on which the blood is already stored, approximately 30 μL of solvent, and an electrical connection to a low-current high voltage power supply.

Cellulose-based paper is an effective medium for biological fluid collection and storage because a number of drugs have been found to be more stable in DBS samples than in refrigerated or frozen liquid blood [8, 9]. Also, a small volume of blood (typically <50 μL) is sufficient for several replicate analyses by paper spray, compared with 0.5 mL or more used for typical plasma analysis. This facilitates sampling of small children and critically ill patients. Further, these small blood volumes may enable simplified blood sampling, i.e., finger sticks rather than collection of venous blood. This in turn may enable samples to be collected at home or at outpatient clinics. The use of DBS sampling for TDM has been previously explored [10]; the typical approach is a complex multi-step procedure that involves taking a punch from the center of the blood spot, extracting the punch using an internal standard-spiked solvent, transferring the extract to a fresh tube, and centrifuging the sample to remove any particulates, again transferring the supernatant to a fresh tube, drying the sample and resuspending it in a new solvent, and performing HPLC-MS/MS analysis.

In this report, we demonstrate the application of paper spray to the quantitative analysis of drugs directly from DBS samples. This single step analysis, which can be applied to a wide range of therapeutic drugs with minimal additional development, is a simple, effective, and relatively inexpensive alternative to currently available options for TDM.

2 Methods

2.1 Materials

Most drug standards were purchased from Sigma-Aldrich (St. Louis, MO, USA) with the exception of the following: [2H10]sunitinib: Toronto Research Chemicals (Ontario, Canada), [2H6]amitriptyline and [2H4]citalopram: CDN isotopes (Pointe-Claire, Quebec, Canada), paclitaxel and sunitinib: Axxora (San Diego, CA, USA), acetaminophen, [2H4]acetaminophen, sitamaquine, and [2H10]sitamaquine: GSK (Stevenage, UK), proguanil: Molekula (Dorset, UK), simvastatin: Calbiochem (Nottingham, UK), benzethonium chloride: Alfa Aesar (Lancs, UK). Blood card paper (grade 31 ETF) was obtained as a sample from Whatman (Piscataway, NJ, USA). Bovine and human blood stabilized with K2EDTA was purchased from Innovative Research (Novi, MI, USA). Control rat blood was obtained from Harlan (Hull, UK, USA). All control blood was shipped in insulated containers containing fridge packs. In all cases, blood was collected into containers coated with EDTA to prevent coagulation.

2.2 Mass Spectrometry and Paper Spray Ionization

All experiments were performed using a TSQ Quantum Access Max (Thermo Scientific, San Jose, CA, USA) in the selected reaction monitoring (SRM) mode. The SRM parameters are shown in Supplementary Table 1.

For paper spray ionization, the paper was cut into a triangle of 7.2 mm base width × 11 mm height. A volume of solvent, typically 25 μL of 90% methanol:10% water with 100 ppm acetic acid, was added to the paper using a pipette after the blood spot had thoroughly dried. For compounds detected as sodium adducts, 90% methanol with 200 ppm sodium acetate was used as the spray solvent. After the solvent was added, the spray voltage was then set to 3500 V, initiating the electrospray plume; analyte ions were detected essentially immediately. After 35 cycles, the voltage was set to zero, halting the spray of ions into the instrument. At this stage, the paper was still wet with the spray solvent. The SRM signal intensity obtained for the analyte(s) and the internal standard was integrated over the entire analysis time and used to measure the amount of analyte in the blood sample. No background subtraction was performed.

2.3 Preparation of Samples

Unless otherwise noted, primary stock solutions for each test compound and internal standard were prepared in dimethylformamide (DMF, 10 mg/mL for acetaminophen, 1 mg/mL for all other compounds). For each assay, working standards at suitable concentrations were made up in methanol:water (1:1, vol/vol). Analytical samples were then prepared by diluting the appropriate working solutions with blank drug free blood. In all cases, the non-matrix solvent spiked into the blood did not exceed 5% of the total volume.

For the sitamaquine assay, calibration standards and quality control (QC) samples were prepared over a concentration range relevant for the physiologic exposure of these drugs by spiking sitamaquine and [2H10]sitamaquine into liquid rat blood. For sitamaquine, the concentrations of calibrators were 5, 10, 20, 50, 100, 200, 500, 800, and 1000 ng/mL and the separately prepared QCs were 5, 20, 100, 800, and 1000 ng/mL. The concentration of the internal standard was 100 ng/mL.

The standards for the experiments involving citalopram, amitriptyline, sunitinib, and telmisartan were prepared as follows: drug solutions at 20× concentration were prepared by serial dilution of the stock solutions into 1:9 methanol:water. The 20× standards were then spiked into blood by pipetting 50 μL of the standard into 950 μL of blood. Both human and animal blood (either bovine or rat) were used in this study; animal blood was employed for some of the experiments for convenience and experimental/handling safety reasons. The type of blood used for each experiment is noted in the appropriate experimental section.

DBS samples were prepared by spotting a fixed volume (15 μL) of blood onto the paper substrate and drying for at least 2 h at room temperature. Samples were stored at room temperature in a sealed plastic bag containing desiccant.

2.4 Addition of the Internal Standard to the Paper

The blood card paper was first cut into a triangle with a 6.5 mm base and 9.5 mm height. A 2 μL aliquot of an internal standard solution in water was added to the middle of the paper triangle using a pipette; for the amitriptyline assay, the concentration of [2H6]amitriptyline was 1250 ng/mL. For the sunitinib assay, the concentration of [2H10]sunitinib was 500 ng/mL. After the paper had dried thoroughly, 10 μL of amitriptyline or sunitinib spiked bovine blood was added directly on top the dried internal standard. After the blood spot dried, analysis was carried out with 25 μL of 95% methanol:5% water.

2.5 Data Analysis

For quantitative analysis, the area under the curve (AUC) for the analyte SRM transition was normalized against the AUC for the corresponding IS-SRM transition. See Supplementary Table 1 for the SRM transitions used. Calibration curves were generated from calibration samples by plotting this ratio against analyte concentration in Microsoft Excel and subjecting the data to 1/× weighted linear least squares. The lower limit of detection was calculated using the formula 3*sB/m, where sB is the standard deviation of the blank and m is the analytical sensitivity (i.e., slope of the standard curve) [11]. All limits of detection are reported in nanograms of compound per milliliter of whole blood (ng/mL). The lower limit of quantitation was defined to be the lowest analyte concentration that gave a signal 5× greater than drug-free blank blood, had a relative standard deviation (RSD) of <20%, and was within 20% of the expected value. Curve fitting was done using 1/× weighted least squares in Mathsoft Mathcad 13.

2.6 Assessment of Relative Matrix Effects

Stock solutions of citalopram hydrobromide with concentrations ranging from 25 ng/mL to 12,500 ng/mL were prepared by serial dilution in 1:9 methanol:water containing 2500 ng/mL [2H4]citalopram as the internal standard. These solutions were diluted 1:20 in human blood from five different donors (Innovative Research, Novi, MI, USA) by mixing 10 μL of the working solution with 190 μL of blood. One sample was analyzed per concentration point per blood sample by paper spray for a total n of 5 at each concentration. The standard curve obtained for each blood sample were fit to a straight line using 1/× weighted least squares.

2.7 Selectivity

Three drugs, telmisartan, sunitinib, and amitriptyline, were tested for adequate selectivity at a concentration of 1 ng/mL. A stock solution with 20 ng/mL of each of the drugs was prepared in 1:9 methanol:water with 50 ng/mL droperidol as the internal standard. This stock solution was spiked into human blood from five different donors by mixing 50 μL of the stock solution with 950 μL of blood. Drug spiked blood samples and blank (presumed drug free) blood from each donor were spotted in triplicate onto dried blood card paper and analyzed by paper spray after drying.

3 Results

3.1 Accuracy and Precision

The analysis of a set of calibration standards and separately prepared quality control samples for the oral drug sitamaquine is presented as an example of the results typically obtainable for small molecule quantitation from DBS samples using paper spray when an isotopically labeled internal standard is spiked into the liquid blood prior to drying the blood on paper. The mass chronograms for the selected reaction monitoring (SRM) transitions for sitamaquine and the internal standard, [2H10]sitamaquine, during one of the calibration runs are shown in Figure 1. Signal for both the analyte and internal standard was detected immediately after application of the high voltage. If the high voltage were to remain on, signal would be detected for approximately 90 s until the solvent was depleted due to the spray process and evaporation. In this experiment, the analysis was halted after a specific number of cycles (typically 35) by setting the voltage to zero. As can be seen from the chronograms in Figure 2, the total ion current is somewhat unstable, commonly varying by 10% to 20% over the course of a single sample analysis. The ion currents obtained from the SRM of sitamaquine and the [2H10]sitamaquine track each other closely, however, and the ratio between the analyte signal and internal standard signal was ultimately found to be reproducible and linear as a function of analyte concentration, as will be described below.
Figure 1

Total ion current chronograms (signal versus time) for the sequential analysis of sitamaquine calibrators in dried rat blood with [2H10]sitamaquine as the internal standard. Each peak in the chronogram is from the analysis of a different dried blood spot. The concentration of the internal standard was a constant 100 ng/mL in each sample. The variability in the signal obtained for the isotopically labeled internal standard (bottom panel) was 17% (standard deviation/mean*100% of the area under the curve for the nine samples shown)

Figure 2

Quantitative analysis of sitamaquine from dried rat blood spots by paper spray by spiking the internal standard into the blood sample. (A) Standard curves (ratio of instrument response for the analyte to that of the internal standard versus analyte concentration) for sitamaquine generated from two different runs, demonstrating linearity and repeatability of the assay. (B) Accuracy and precision (n = 6 replicates at each concentration) for sitamaquine quality control samples. The error bars show ± one standard deviation. These data are shown in tabular form in Supplementary Table 2

The results for quality control samples and two sets of calibration samples are shown in Figure 2. The peak area ratio of the analyte to the internal standard plotted against sitamaquine concentration was linear over the concentration range examined (5–1000 ng/mL) with a correlation coefficient greater than 0.99 (Figure 2a). The accuracy and precision of this assay were well within internationally accepted standards for bioanalytical methods [12]. The variation of replicate measurements was less than 5% (relative standard deviation) for all of the quality control levels (n = 6 per concentration point), and the measured concentration was within 2% of the actual value, except at the lowest concentration level, which was within 10% of the actual value (Figure 2b and Supplementary Table 2). The sitamaquine sample with the lowest concentration analyzed here, 5 ng/mL, is sufficient to quantitate clinically relevant levels of sitamaquine in blood. In phase II clinical trials, the lowest dose administered gave mean plasma trough concentrations (i.e., concentration immediately before administration of the next dose) of 21 ng/mL after 28 d of administration [13].

The standards set by the Food and Drug Administration for bioanalytical method validation require that the inaccuracy and the imprecision to be less than 15% except at the lower limit of quantitation (LLOQ), where 20% is acceptable. For TDM, a number of different standards have been proposed that take into account various factors such as the therapeutic window, elimination rate, and with-in subject variability rather than setting an absolute standard for all analyses. For the assays presented here for sitamaquine, in which the internal standard was mixed into the liquid blood, the precision of the quality control samples over the therapeutic range was < 2%. This performance meets the absolute standards generally adopted for bioanalytical method validation [12] and is adequately reproducible at the upper and lower decision limits using criteria similar to that adopted by the Clinical Laboratory Improvements Amendments (CLIA), in which acceptable performance is defined to be a particular percent of the target value, typically 20% or 25%.

3.2 Limits of Detection

The detection limits obtained for a chemically diverse set of small molecule drugs examined by paper spray MS of DBS were found to vary depending on the chemical and physical properties of the drugs, particularly their acid/base properties and their solubilities. The detection limits ranged from <100 pg/mL for proguanil and benzothenium to >100 ng/mL for acetaminophen and ibuprofen (Supplementary Table 3). Most drugs are weakly basic and hydrophobic; these compounds, including sunitinib, amitriptyline, verapamil, citalopram, and dextrorphan, generally have lower limits of detection around 0.25 to 0.75 ng/mL. It should be noted that these concentrations represent only the lower limit of detection; the achievable lower limit of quantitation will be somewhat higher.

The upper and lower limits of quantitation should be sufficient to cover at least the entire therapeutic window of the drug. For paper spray mass spectrometry, the lower limit of quantitation obtainable was the most important single factor for determining its applicability to a particular drug, as the upper limit of quantitation, the accuracy, and the precision were generally sufficient provided the lower limits of quantitation were adequate. Ultimately, adequate lower limits of quantitation will need to be established for each drug assayed by paper spray. In this study, the lower limits of quantitation by paper spray were below the therapeutic range for sunitinib, paclitaxel, and citalopram, for example. Based on current data, it is reasonable to expect that paper spray MS will have adequate sensitivity to detect therapeutic levels of many drugs, but not all. In its present form, the method will likely not have sufficient lower detection limits for pharmacokinetic studies or for the detection of therapeutic levels of highly potent drugs or non-small-molecule therapeutics such as recombinant therapeutic proteins.

3.3 Selectivity

The selectivity of paper spray MS was established for amitriptyline, sunitinib, and telmisartan after spiking the drugs into blank control human blood from five different individuals. No unacceptable levels of interference were present in unspiked blank samples compared with the same blood samples spiked with drug at a concentration of 1 ng/mL (Supplementary Figure 1). Acceptable levels for the blank signals were defined to be less than 20% of the analyte response at the lower limit of quantitation [14].

3.4 Incorporation of the Internal Standard

For the sitamaquine assay presented previously the 2H4 isotopically labeled analog of sitamaquine was spiked into the liquid blood prior to deposition of the blood onto the paper. For the collection of blood samples as DBS at the point of care, however, spiking a known quantity of internal standard into the liquid blood may not be feasible. An alternative approach is to add the internal standard to the paper prior to spotting the blood, or to add the internal standard to the blood spot after it has dried but prior to analysis. In this study, we pretreated the paper with an internal standard solution prior to blood application. Later, the blood sample was spotted directly on top of the internal standard spot and dried prior to paper spray analysis. A stable isotopically labeled internal standard was employed to eliminate the possibility of separation of the analyte and the internal standard during blood deposition or spray solvent application.

The accuracy obtained for amitriptyline by prespotting [2H6]amitriptyline onto the paper prior to depositing the blood sample was within normally acceptable levels, with absolute errors within 5%–10% of the actual value (Figure 3b and Supplementary Table 4). The imprecision of the method was approximately 10% for all of the concentration levels with the exception of the lowest concentration, 0.9 ng/mL, which varied by 22%. This method has acceptable analytical performance over the normal therapeutic range of amitriptyline, which is approximately 50–200 ng/mL in plasma [15]. Similar results were obtained for sunitinib by prespotting the internal standard [2H10]sunitinib (Supplementary Figure 2 and Supplementary Table 5). Sunitinib was measured with acceptable accuracy and precision over a concentration range of 1–500 ng/mL; the therapeutic range of this drug is around 50–100 ng/mL [16].
Figure 3

Quantitative analysis of amitriptyline from dried bovine blood by paper spray by pretreating the paper with the internal standard [2H6]amitriptyline. (A) Standard curve (ratio of instrument response to that of the internal standard versus analyte concentration), demonstrating linearity of the assay. (B) Accuracy and precision (n = 6 replicates at each concentration) for amitriptyline samples. The error bars show ± one standard deviation. These data are shown in tabular form in Supplementary Table 4

3.5 Assessment of Relative Matrix Effects

In any analytical assay, the potential exists for relative matrix effects arising from the use of biofluids from different individuals. One accepted method to assess relative matrix effects is to construct standard lines (i.e., instrument response versus drug concentration) in five different lots of biofluid. In this approach, for the assay to be considered free of relative matrix effects, the slopes of the standard lines obtained for each lot of biofluid should vary less that 3% (relative standard deviation) and each individual concentration point should vary less than 15% across all five lots of biofluid, except at the lower limit of quantitation where 20% is acceptable. In the validation performed here, the antidepressant drug citalopram, which has a therapeutic concentration of ca. 10–200 ng/mL [15], and the internal standard [2H4]citalopram were spiked into blood collected from five different human donors over a concentration range of 1–500 ng/mL. The blood was spotted onto paper and analyzed directly by paper spray. The slopes of the standard lines determined for the five different blood samples varied by 1.3% (Supplementary Figure 3), which compares favorably with validated HPLC-MS methods [17]. The precision of the individual concentration points was <15% except at the lower limit of quantitation (Supplementary Table 6). Based on generally accepted criteria, this assay can be considered free of relative matrix effects. All assays will need to be assessed for relative matrix effects, but it is anticipated that the use of isotopically labeled internal standards will prevent the occurrence of relative matrix effects in paper spray, as has generally been the case with HPLC-MS methods.

4 Conclusion

We have presented paper spray, a mass spectrometry based method for the quantitative analysis of small molecules directly from DBS samples. Analysis was carried out on the same dried blood card paper used for sample storage and required only the addition of solvent and high voltage to generate gas phase ions of the analyte molecules from the tip of the paper. By incorporating an appropriate internal standard into the liquid blood, or preferably onto the paper itself prior to blood deposition, quantitatively accurate and precise results are routinely obtainable down to single ng/mL levels for small molecule therapeutics.

One issue that has not yet been tested is the effect of fragile metabolites and pro-drugs, such as acyl glucuronide and N-oxide metabolites and esters that could convert back to the tested drug during MS analysis, thereby leading to over estimation of circulating drug concentrations. For paper spray MS analysis of drugs known to be metabolized into fragile compounds, methods will need to be developed to avoid overestimation of the parent drug. One possibility is the use of selective on-paper chemical reactions, which have been previously demonstrated for paper spray [7], to convert the metabolite into a stable form that will not interfere with parent drug analysis [18]. Another is to use a simple ion mobility separation in conjunction with the mass spectrometry (IMS/MS) provided that the fragmentation does not occur during the ionization process [19].

MS has the potential to significantly increase the availability of clinical tests that can be used to aid in the guidance of patient care. A significant barrier to the application of MS in clinical laboratories, however, has been the expertise required to design and trouble shoot methods along with operation of the instrument and the HPLC apparatus. Paper spray mass spectrometry shortens the workflow and decreases the complexity of MS quantitation. The tradeoff for this simplicity is decreased performance relative to traditional HPLC-MS assays. However, the performance of paper spray seems to be adequate for the quantitative analysis of a large number of pharmaceuticals over their therapeutic range for the purpose of TDM.

Acknowledgments

The authors acknowledge funding and other support from the Alfred Mann Institute for Biomedical Development at Purdue University (AMIPurdue) for development of the technology presented in this paper. They particularly thank Rizaldi Sistiabudi for his assistance.

Supplementary material

13361_2011_177_MOESM1_ESM.doc (44 kb)
Supplementary Table 1SRM conditions (DOC 43 kb)
13361_2011_177_MOESM2_ESM.doc (34 kb)
Supplementary Table 2Assay performance data for quality control samples of sitamaquine in dried blood spots analyzed directly by paper spray. The internal standard [2H10]sitamaquine was mixed into the liquid blood prior to spotting at a concentration of 100 ng/mL. These data are shown graphically in Figure 3 (main text). (DOC 34 kb)
13361_2011_177_MOESM3_ESM.doc (128 kb)
Supplementary Table 3Lower limits of detection in dried bovine blood for several small drug or drug-like molecules with a wide variety of chemical and physical properties. (DOC 127 kb)
13361_2011_177_MOESM4_ESM.doc (34 kb)
Supplementary Table 4Analysis of amitriptyline from dried blood spots. The internal standard [2H6]amitriptyline was prespotted on the paper prior to blood deposition. (DOC 34 kb)
13361_2011_177_MOESM5_ESM.doc (33 kb)
Supplementary Table 5Results for the quantitative analysis of quality control samples of sunitinib in dried blood spots at ~3× the limit of quantitation and at the middle of the standard curve. The internal standard sunitinib-d10 was spotted onto the paper and dried prior to deposition of the blood. (DOC 33 kb)
13361_2011_177_MOESM6_ESM.doc (35 kb)
Supplementary Table 6Analysis of citalopram by paper spray in dried blood spot samples derived from five different sources of human blood. The precision and bias data shown were obtained across all five different blood samples. (DOC 35 kb)
13361_2011_177_MOESM7_ESM.doc (50 kb)
Supplementary Figure 1Demonstration of selectivity for telmisartan, sunitinib, and amitriptyline at a concentration of 1 ng/mL in five different human blood samples. (DOC 50 kb)
13361_2011_177_MOESM8_ESM.doc (46 kb)
Supplementary Figure 2Analysis of standards (n = 2 per concentration point) and quality control samples (n = 8 per concentration) of sunitinib in dried blood spots; the internal standard was prespotted onto the paper prior to blood deposition. (DOC 46 kb)
13361_2011_177_MOESM9_ESM.doc (61 kb)
Supplementary Figure 3Paper spray analysis of blood from five different human donors spiked with citalopram and [2H4]citalopram (internal standard). The slopes of the standard lines generated from the five different blood samples varied by 1.3%, indicating a lack of relative matrix effects arising from the use of different blood lots. (DOC 61 kb)

Copyright information

© American Society for Mass Spectrometry 2011

Authors and Affiliations

  1. 1.Department of ChemistryPurdue UniversityWest LafayetteUSA
  2. 2.Drug Metabolism and Pharmacokinetics, GlaxoSmithKline Research and Development Ltd.WareUK
  3. 3.School of ScienceUniversity of GreenwichKentUK
  4. 4.Weldon School of Biomedical EngineeringPurdue UniversityWest LafayetteUSA
  5. 5.Center for Analytical Instrumentation DevelopmentPurdue UniversityWest LafayetteUSA

Personalised recommendations