Correlation between the quantification results of DBSs and plasma sampling

CHAPTER 3 SIMULTANEOUS QUANTIFICATION OF 6 HIV MEDICINES ON

3.3.4 Correlation between the quantification results of DBSs and plasma sampling

Current data on therapeutic concentrations of AR drugs is obtained from regular plasma sampling. In order to use DBS sampling alternative to plasma it is essential to correlate analyte-specific DBS concentrations to plasma concentrations and translate DBS concentrations to predicted plasma concentrations using the conversion factors (CF).

The CFs are dependent on the plasma to whole blood drug distribution ratio, therefore it is important to know the conversion factors for each drug ^89-90.

To establish the correlations between DBS and plasma concentrations, we collected paired DBS and plasma samples from patients undergoing ART treatment. ART is the combination of multiple drugs and TFV and FTC are the most commonly used NRTIs in ART. This study collected paired samples containing TFV (n=26) and FTC (n=20), and they are used to study the correlation between DBS and plasma concentrations. Because DRV, RTV, EVG and COBI were recently approved for ART,

doi:10.6342/NTU201804215

and they are still in clinical trials at National Taiwan University Hospital, only very limited blood samples could be obtained for plasma/DBS concentration comparison.

Therefore, we manually generated the paired DBS and plasma samples using a healthy volunteer whole blood with Hct 45. We created 20 paired samples at different concentrations with different Hct from 20% to 56% to estimate the CFs of COBI, DRV, RTV and EVG. Linear correlations between DBS and plasma concentrations were established for all the drugs (Figure 3.6.5). Bland−Altman plots were generated using the mean predicted plasma concentrations from DBS samples and plasma concentrations for all the drugs (Figure 3.6.6). The slope of the regression equations were 1.83 for TFV, 1.11 for FTC, 0.82 for COBI, 1.02 for DRV, 0.82 for RTV and 0.94 for EVG. DBS concentrations were utilized to predict the plasma concentrations using the CFs obtained from the average ratio of plasma and DBS concentrations ^91-92. The average CFs of each drug was 2.2 for TFV, and 1.37 for FTC, 0.64 for COBI, 0.93 for DRV, 0.89 for RTV and 0.88 for EVG. The prediction accuracy of TFV, FTC, COBI, DRV, RTV and EVG was evaluated by Bland-Altman plots. The results of Bland−Altman plots revealed above 95% of the predicted values for all the drugs were within upper and lower LoA with 95%

CI.

Although the number of samples used to generate regression equations for all the drugs were limited, the calculated CF or slopes were comparable to the previously published results ^93-95. Through our results, we found that not all the drugs DBS concentrations were similar to the plasma concentrations. This could be due to the influence of Hct in DBS samples. Therefore, future studies to evaluate the Hct variation effect is essential to generate the translation equations for DBS and plasma concentrations.

doi:10.6342/NTU201804215

3.4 Conclusions

HIV is a major health problem worldwide. cART is the standard treatment to improve the life expectancy in HIV patients. But maintaining the treatment efficacy is another important consideration in cART regimens as the HIV infection attacks the immune system and causes multiple infections. Therefore, monitoring the drug concentrations in patient blood samples and related dose adjustments are important to improve the treatment efficacy. DBSs sampling simplifies the blood sampling process for routine drug monitoring in patients with self-sampling. In this study we developed an accurate, precise and reproducible PCI-IS combined with LC-ESI-MS method for blood volume estimation and accurate quantification of 6 anti-HIV drugs in DBS samples. We successfully applied our developed method to quantify AR drugs in paired DBS and plasma HIV samples and correlated their concentrations using conversion factors. Bland-Altman plot revealed that predicted plasma concentrations from DBS concentrations showed good agreements with the measured plasma concentrations after translated using CFs. Since our study included the newly approved drugs, the developed method could be a feasible and reliable drug quantification method for clinical trials, pharmacokinetic/pharmacodynamics studies, and also for TDM when small sample volumes are available. The method can also be applied to support the investigation of the patient compliance of cART in rural areas where clinical professionals and facilities are very limited.

doi:10.6342/NTU201804215

3.5 Tables

Table 3.5.1. Mass parameters, retention time and chemical structures of the target compounds

Compound name

Chemical structures Parent ion (m/z)

Product ion (m/z)

Dwell time

Fragmentor (V)

Collision energy

(V)

MS acquisition rate (Spectra s^-1)

Retention time (min)

Tenofovir 288.04 175.9 175 130 30 1.30 4.16

Emitricitabine 248.1 129.8 125 90 15 1.30 4.44

doi:10.6342/NTU201804215

Cobicistat 776.4 606.2 50 182 24 1.02 5.69

Darunavir 548.2 392.1 50 106 12 1.02 6.15

Ritonavir 721.3 140 50 159 76 1.02 6.55

Elvitegravir 448.1 344 50 129 44 1.02 6.77

doi:10.6342/NTU201804215

Table 3.5.2. Selection of PCI-IS by evaluating the quantification accuracy and precision using calibration curve generated with five different isotope labelled internal standards as PCI-IS (n=30).

Concentration

doi:10.6342/NTU201804215

Darunavir^d

LLOQ 78.0 12.7 75.9 29.0 86.4 4.7 116.8 8.2 121.0 6.3

LQC 66.7 0.3 51.6 2.2 84.9 0.8 85.2 1.8 88.6 2.1

MQC 82.3 1.0 69.0 1.0 89.5 1.0 86.7 2.3 93.9 0.3

HQC 94.5 0.6 82.3 0.8 99.9 0.9 83.3 0.8 94.3 0.7

Ritonavir^e

LLOQ 67.3 8.1 60.1 25.2 111.1 1.9 79.0 4.9 77.2 5.3

LQC 66.6 6.3 47.6 7.8 84.3 7.6 77.2 5.9 83.3 5.8

MQC 84.6 6.0 79.3 5.4 94.0 7.0 90.6 4.4 100.1 4.4

HQC 95.8 4.6 95.7 2.0 105.2 5.0 99.0 3.0 113.1 3.0

Elvitegravir^f

LLOQ 78.5 8.9 91.2 25.5 90.3 1.5 98.9 5.0 89.2 6.3

LQC 79.8 3.6 64.9 0.8 87.9 3.1 90.9 1.7 92.6 2.1

MQC 92.6 1.9 86.0 2.5 104.4 1.7 100.6 1.6 114.9 0.3

HQC 98.0 2.8 85.7 1.6 108.2 2.6 94.4 1.5 114.3 0.7

aTenofovir concentrations at LLOQ/LQC/MQC/HQCs were 2.5/7.5/300/750 ng/mL.

bEmtricitabine concentrations at LLOQ/LQC/MQC/HQCs were 5/15/750/1875 ng/mL.

cCobicistat concentrations at LLOQ/LQC/MQC/HQCs were 5/15/600/1500 ng/mL.

dDarunavir concentrations at LLOQ/LQC/MQC/HQCs were 10/30/750/1875 ng/mL.

eRitonavir concentrations at LLOQ/LQC/MQC/HQCs were 10/30/750/1875 ng/mL.

fElvitegravir concentrations at LLOQ/LQC/MQC/HQCs were 20/60/750/1875 ng/mL.

doi:10.6342/NTU201804215

Table 3.5.3. Blood volume estimation accuracy of the PCI-IS method.

No of

Table 3.5.4. Calibration curve, limit of detection (LOD) and limit of quantification (LOQ) of AR drugs in DBS samples.

Compound name Calibration range

(ng/mL) r² LOD

Tenofovir 2.5-75 ≥0.9992

0.5 1.5 0.30±0.09

75-1000 ≥0.9999

Emitricitabine 5-100 ≥0.9987

0.5 1 1.80±0.70

100-2500 ≥0.9999

Cobicistat 5-100 ≥0.9999

0.5 1 0.99

100-2000 ≥0.9999

Darunavir 10-250 ≥0.9997

0.5 1 7.74±1.69

250-2500 ≥0.9999

Ritonavir 10-250 ≥0.9985

0.5 1 11.20±3.60

doi:10.6342/NTU201804215

Table 3.5.5. Intra-day, inter-day precision and accuracy, extraction recovery (ER) and matrix effect (ME) of all drugs.^a

Tenofovir^b Emtricitabine^c Cobicistat^d Darunavir^e Ritonavir^e Elvitegravir^g Accuracy

aIntra-day and Inter-day precision, accuracy, recovery and matrix effects were analyzed as triplicates with their corresponding coefficients of variation (CV; range includes triplicate of each drug response ratio).

bTenofovir concentrations at LLOQ/LQC/MQC/HQCs were 2.5/7.5/300/750 ng/mL.

cEmtricitabine concentrations at LLOQ/LQC/MQC/HQCs were 5/15/750/1875 ng/mL.

dCobicistat concentrations at LLOQ/LQC/MQC/HQCs were 5/15/600/1500 ng/mL.

eDarunavir concentrations at LLOQ/LQC/MQC/HQCs were 10/30/750/1875 ng/mL.

f Ritonavir concentrations at LLOQ/LQC/MQC/HQCs were 10/30/750/1875 ng/mL.

g Elvitegravir concentrations at LLOQ/LQC/MQC/HQCs were 20/60/750/1875 ng/mL.

doi:10.6342/NTU201804215

Table 3.5.6. Stability of 6 AR drugs at different conditions. (A) Stability of 6 AR drugs in whole blood samples at room temperature (RT) for 2 and 4 hours and at 4^oC for 24 and 48 hours. (B) Short-term and long-term stability of 6 AR drugs in DBS samples stored at RT, 4 ^oC and -20 ^oC for 1, 7 and 30 days. (C) Post preparation stability of 6 AR drugs stored at 4 ^oC in auto sampler for 48 hours and at -20 ^oC for 1 week. All the conditions were set according to the sample collection and analysis.

Stability (%) LQC HQC

RT 4^oC RT 4^oC

2 hours 4 hours 24 hours 48 hours 2 hours 4 hours 24 hours 48 hours Tenofovir^a 105.04 94.34 106.20 99.36 92.24 95.64 93.54 99.36 Emitricitabine^b 110.77 102.33 102.07 98.37 92.61 100.19 96.25 93.74 Cobicistat^c 91.68 107.55 92.29 99.29 103.43 89.20 85.70 95.50 Darunavir^d 112.40 111.23 105.17 113.29 95.92 107.31 94.89 108.64 Ritonavir^e 109.16 109.92 101.91 103.91 99.97 104.18 100.21 97.93 Elvitegravir^f 110.15 101.63 98.22 96.56 97.56 105.07 91.09 95.62 (A)

doi:10.6342/NTU201804215

Stability (%) Storage condition

1 day 7 days 30 days

LQC HQC LQC HQC LQC HQC

Tenofovir^a RT 108.34 107.50 111.96 111.60 105.48 92.75 4^oC 114.46 117.43 105.54 107.28 113.06 115.19 -20^oC 109.51 118.28 114.09 115.67 114.11 110.47 Emitricitabine^b RT 107.51 107.49 105.95 107.76 108.09 102.72 4^oC 106.46 110.72 102.72 104.87 101.46 115.54 -20^oC 105.87 109.12 105.64 115.33 104.89 112.39

Cobicistat^c RT 92.32 97.83 87.59 75.96 81.03 72.90

4^oC 95.49 103.30 92.98 75.90 90.23 75.57 -20^oC 101.72 112.97 93.19 83.62 97.61 97.57 Darunavir^d RT 111.67 106.59 102.51 113.11 114.07 115.72

4^oC 97.08 96.52 97.98 105.77 118.66 117.06 -20^oC 117.69 117.72 101.50 116.80 102.84 104.71

Ritonavir^e RT 94.61 95.41 85.47 94.58 99.18 104.98

4^oC 92.37 98.38 87.40 92.42 103.91 102.23 -20^oC 99.91 103.65 87.21 95.49 104.26 110.50 Elvitegravir^f RT 97.00 99.53 105.22 99.50 99.57 98.01

4^oC 101.37 101.55 90.91 97.75 110.48 102.04 -20^oC 102.60 95.50 97.12 96.16 106.57 110.75 (B)

doi:10.6342/NTU201804215

Stability (%) Auto sampler at 4^oC -20^oC

48 hours 1 week

LQC HQC LQC HQC

Tenofovir^a 91.84 107.65 115.07 93.84

Emitricitabine^b 93.46 96.88 112.99 92.00 Cobicistat^c 111.00 104.80 92.81 93.80

Darunavir^d 94.65 100.65 99.23 95.20

Ritonavir^e 100.54 104.44 90.47 101.53 Elvitegravir^f 113.90 101.34 98.83 89.79

aTenofovir concentrations at LQC/HQCs were 7.5/750 ng/mL. ^bEmtricitabine concentrations at LQC/HQCs were 15/1875 ng/mL,

cCobicistat concentrations at LQC/HQCs were 15/1500 ng/mL. ^dDarunavir concentrations at LQC/HQCs were 30/1875 ng/mL, ^eRitonavir concentrations at LQC/HQCs were 30/1875 ng/mL, ^fElvitegravir concentrations at LQC/HQCs were 60/1875 ng/mL.

(C)

doi:10.6342/NTU201804215

3.6 Figures

Figure 3.6.1 Concept of blood volume estimation by PCI-IS. (Modified from ref 27)

doi:10.6342/NTU201804215

Figure 3.6.2 Schematic representation of the study design.

doi:10.6342/NTU201804215

Figure 3.6. 3 MRM Chromatograms of (A) tenofovir, (B) emtricitabine , (C) cobicistat, (D) darunavir, (E) ritonavir and (F) elvitegravir from DBS sample obtained under optimal LC-MS/MS conditions. The concentration of all the drugs including TFV/FTC/COBI/DRV/RTV/EVG were 2.5/5/5/10/10/20 ng/mL

doi:10.6342/NTU201804215

Figure 3.6.4 Calibration curve of for estimation of the blood volume on the DBS card by using blood volume and reciprocal PCI-IS intensity.

y = 4E-06x - 1E-05 R² = 0.9948

0 0.00001 0.00002 0.00003 0.00004 0.00005 0.00006 0.00007 0.00008 0.00009

0 5 10 15 20 25 30

R ecip ro cal in ten sity o f PC I- IS

Blood volume (µL)

Calibration curve

doi:10.6342/NTU201804215

Figure 3.6.5 correlation between paired DBS and plasma samples for (A) tenofovir (TFV) (B) emtricitabine (FTC) (C) cobicistat (COBI) (D) darunavir (DRV) (E) ritonavir (RTV) and (F) elvitegravir (EVG).

doi:10.6342/NTU201804215

Figure 3.6. 6 The Bland-Altman plots for the predicted plasma concentrations for (A) tenofovir (TFV) (B) emtricitabine (FTC) (C) cobicistat (COBI) (D) darunavir (DRV) (E) ritonavir (RTV) and (F) elvitegravir (EVG) with the measured plasma concentrations. The mean (blue line), the upper and lower LoAs (95% CI, dashed redlines) are also indicated.

doi:10.6342/NTU201804215

Chapter 4

An Improved DBS-Based Metabolomics Analysis

by Post Column Infused-Internal Standard

Assisted Liquid Chromatography-Electrospray

Ionization Mass Spectrometry Method

doi:10.6342/NTU201804215

4.1 Introduction

Over the past decades, metabolomics has been developed as one of the important omics sciences, as it allows the estimation of a wide range of metabolites to generate new insights in disease diagnosis, investigate physiological status, interpret the pathways of a disease state, and find new biomarkers ^2-4. The common samples of biofluids used for metabolomics studies are plasma or serum, urine, and recently, dried blood spots (DBS)

6-7, 96-97. Metabolomics studies developed using plasma or serum may lack information about whole-blood metabolites, such as metabolites involved in red blood cell (RBC) energy metabolism. Moreover, their more invasive sample collection procedure and stringent storage conditions emphasize the growing importance of DBS sampling techniques in metabolomics studies.

The DBS sampling technique was introduced in 1960s to screen the inborn errors of metabolism in neonates ⁹⁸. It has been used extensively in newborn screening as it has numerous advantages including a minimally invasive collection procedure using a finger or heel prick, low sample volume, easy sample handling, and shipping. Newborn screening tests in developed countries have been performed via DBS sampling techniques

99. Other additional advantages, including feasibility of patient self-sampling with moderate training, stability of photosensitive compounds, and low biohazard risks ^{5, 73,}

100-102, increased the applicability of the DBS sampling technique in adult care in various fields such as therapeutic drug monitoring,⁵ pharmacokinetics,⁷⁴ genomics,¹⁰³ proteomics,^104-105 lipidomics,^106-107 and metabolomics 7, 75, 108-110. Although the DBS sampling technique has recently gained importance in metabolomics, few studies assessed the challenges associated with the DBS sampling technique in metabolomics studies ⁷⁵ .

doi:10.6342/NTU201804215

The most widely discussed challenges associated with DBS sampling techniques are the effects of hematocrit (Hct) variation and spotted blood volumes. Hct represents the volume percentage of RBC, which affects DBS analytical results as it varies largely by individual ¹¹¹. The individual difference in Hct values influences the blood viscosity, plasma and blood distribution and results in variations in spot formation, spot size, drying time, and homogeneity leading to analytical uncertainty ⁷⁷. Several methods including potassium concentration measurements, and reflectance-based methods have been proposed to estimate Hct levels 75, 112-113. Additionally, using fixed-diameter punches for DBS samples, whole spot analysis could also mitigate the analytical variation caused by Hct variation and could improve the method sensitivity ¹¹⁴. Hct variation may affect the measurement of endogenous metabolite concentrations depending on the intracellular and extracellular distribution, which may result in biased comparisons in DBS-based metabolomics studies. However, Hct effects in DBS-based metabolomics studies have not been discussed in previous studies.

Blood volume variation is another frequently discussed challenge associated with DBS sampling techniques. In addition to use Mitra or volumetric absorptive microsampling to control sampling volume,¹¹⁵ a post column infused-internal standard (PCI-IS) method was recently developed to estimate the blood volume of DBS spots ²⁷. The developed method estimated the blood volume by measuring the total salts in blood samples by the PCI-IS. The developed method could facilitate the whole spot analysis.

Moreover, the PCI-IS strategy could additionally correct matrix effect caused errors of liquid chromatography-electrospray ionization mass spectrometry (LC-ESI-MS). The efficiency of the PCI-IS coupled with LC-ESI-MS for matrix effect correction for amino acids, hormones, and small molecules in biological samples has been demonstrated previously 27, 84, 116-117. Therefore, the PCI-IS assisted LC-ESI-MS method could enable

doi:10.6342/NTU201804215

the simultaneous estimation of the blood volume on DBS cards and correction of the matrix effect for DBS-based metabolomics studies as matrix effect differences between each test sample would be more severe due to spot volume and Hct variations.

Although DBS-based metabolomics has become increasingly important, previous studies have not specifically discussed the potential bias of this sampling method in metabolomics studies and provided solutions to overcome the sampling related bias. The present study proposed a PCI-IS assisted LC-ESI-MS method to improve the data quality of DBS-based metabolomics studies. We tried 6 extraction solvents to compare their efficiency in the extraction of the maximum reproducible metabolite features from DBS samples for metabolomics profiling. The most efficient extraction solvent was further used in targeted metabolite analysis. To examine the critical factors including the effects of Hct variation and spotted blood volumes, which hamper the wider application of DBS-based metabolomics, we selected 20 target metabolites with varied characteristics. We used whole spot analysis to ameliorate the analytical variation caused by Hct variation in DBS samples and proposed a PCI-IS method to simultaneously estimate the blood volume and correct the matrix effect to evaluate the spot volume effect and Hct variation effect on target metabolites. The proposed PCI-IS strategy could improve the data quality of DBS-based metabolomics studies and benefit various clinical research applications.

4.2 Experimental section 4.2.1 Chemicals and materials

Hexakis (1H,1H,3H-perfluoropropoxy2,2-difluoroethoxy)-phosphazene (HKP) was purchased from Apollo (Apollo, Graham, NC, U.S.A.). Acetonitrile (ACN) (LC/MS grade) and ethanol (EtOH) (LC/MS grade) were purchased from J.T. Baker (J.T. Baker, Phillipsburg, NJ, U.S.A.). Water (MS grade) and methanol (MeOH) (MS grade) were

doi:10.6342/NTU201804215

bought from Scharlau (Scharlau, Sentmenat, Barcelona, Spain). Formic acid and tert-butyl methyl ether (MTBE) were purchased from Sigma (Sigma, St. Louis, MO, U.S.A.).

Acetone was obtained from Avantor Performance Materials (Center Valley, PA, U.S.A.).

Whatman 903 Protein Saver cards (Whatman, Maidstone, UK) were used for spotting the blood samples (DBS samples). Hybrid solid phase extraction (SPE) cartridges were purchased from Sigma (Sigma, St. Louis, MO, U.S.A.). The manual puncher was purchased from a local store, and the diameter of the hole was 6 mm.

4.2.2 PCI-IS assisted LC-ESI-MS analysis

PCI-IS assisted LC-ESI-MS analysis was performed on an Agilent 1260 quaternary solvent pump (Agilent Technologies, Santa Clara, CA, U.S.A.) and Agilent 1290 UHPLC system (Agilent Technologies, Santa Clara, CA, U.S.A.) coupled with Agilent 6460 triple quadrupole system (Agilent Technologies, Waldbronn, Germany). An Agilent 1260 quaternary solvent pump was applied for the post-column infusion of the PCI-IS. D8-phenylalanine as the PCI-IS was dissolved in ACN at 1 μg/mL and was introduced into the ESI interface at a flow rate of 0.1 mL/min.

An Acquity HSS T3 column (2.1 × 100 mm, 1.8 μm, Waters, Milford, MA, U.S.A.) was used for metabolite separation. The compositions of mobile phases A and B were 0.1% formic acid (FA) in deionized water (DI), and 0.1% FA in ACN, respectively. A gradient with a flow rate of 300 μL/min was used to separate the metabolites and consisted of 0−1.5 min, 2% B; 1.5−9 min, 2% to 50% B; 9−14 min, 50% to 95% B; 14−17 min, and 95% B. The column re-equilibration time was 3 min. The sampler and column oven were maintained at 4 and 40 °C, respectively. The sample injection volume was set as 5 μL. The ESI parameters for both positive and negative ionization modes were set as

doi:10.6342/NTU201804215

follows: 300 °C dry gas temperature, 6 L/min dry gas flow rate, 45 psi nebulizer pressure, 325 °C sheath gas temperature, 11 L/min sheath gas flow rate, 3500 V capillary voltage, and 500 V nozzle voltage. The MS acquisition was performed using the multiple reaction monitoring (MRM) mode. D8-phenylalanine MRM transition in positive mode was m/z 174.1 -> 128.2, and negative mode was m/z 172.1 -> 154.0 respectively. The MRM transitions and other parameters of all the metabolites and their pathways were summarized in Table 4.5.1.

4.2.3 Metabolomic profiling

Metabolic profiling was performed on Agilent 1290 UHPLC system (Agilent Technologies, Santa Clara, CA, U.S.A.) combined with Bruker maXis QTOF (Bruker Daltonics, Bremen, Germany). LC separation conditions were similar with the PCI-IS assisted LC-ESI-MS analysis. ESI mass parameters for both positive and negative ionization detection modes were 500 V end plate offset, 4500 V capillary voltage, 200 °C drying gas temperature, 12 L/mindrying gas flow, and 30 psi nebulizer flow. The mass spectrometer was calibrated with 10 mM sodium formate before daily use. During analysis, HKP was used as reference mass to correct the mass accuracy in both positive and negative ionization modes.

4.2.4 DBS sample preparation procedure

The blood samples collected in EDTA tubes were used to spot on DBS cards within 15 min of blood collection. The protocol has been approved by the Institutional Review Board at the National Taiwan University Hospital (201412115RINB). Twenty microliters of the blood sample were spotted onto a DBS card, and the spotted cards were

doi:10.6342/NTU201804215

air-dried for 2 h. After drying, the whole blood spot was cut into a clean Eppendorf tube using a 6-mm manual puncher by punching several times. Then, 300 μL of 80% ACN was added to extract the cut blood spot using ultra sonication for 30 min. After sonication, the samples were centrifuged for 5 min at 15000xg. Two hundred and seventy microliters of the supernatant from each sample were evaporated under nitrogen gas. The dried samples were stored at −20 °C till the analysis. Before sample analysis the stored dried samples were reconstituted with 150 μL of 50% MeOH using 10 min sonication. The samples were vortexed and then filtered through 0.22 μm regenerated cellulose membrane (RC-4, Sartorius, Göttingen, Germany) and analyzed by UHPLC-ESI-MS/MS system.

4.2.5 PCI-IS method for the estimation of blood volume on DBS cards and matrix effect correction

The concept of the PCI-IS for blood volume estimation and matrix effect correction was described by Liao et al.^{27, 83, 85}. For estimation of the blood volume, blood samples from 3 volunteers were used to generate the DBS with varying volumes ranging from 10 to 50 μL. After air drying for two hours, the samples were prepared by using the sample preparation protocol described previously. After sample analysis, the PCI-IS signal intensity at the first ion suppression zone was used to construct the calibration curve at different volumes. The generated calibration curve was used to estimate the blood volume of the DBS, and the estimated blood volume was used to correct the intensity difference of target metabolites. For correction of the matrix effect caused errors and lipid accumulation caused signal change, analyte signal intensities at each time point were divided by the PCI-IS signal intensity to generate the adjusted chromatogram.

doi:10.6342/NTU201804215

4.2.6 Preparation of DBS sample for evaluation of Hct variation effect

Fresh blood samples collected in EDTA tubes were used to create the blood models with different Hct levels. The collected blood samples were transferred into clean Eppendorf tubes and centrifuged at 4 °C for 5 min at 3000 rcf to obtain an upper plasma layer and a lower cellular layer. By adding or removing plasma, we generated blood models with different Hct levels. In the present study, we created 6 blood models with different Hct levels ranging from 20 to 75% using each blood sample (n=3×6). Twenty microliters from each blood model with a specific Hct level were spotted onto the DBS card and air-dried for 2 h. After drying, all the DBS samples with different Hct levels were prepared using the sample preparation method described in the previous section.

After sample analyses by PCI-IS assisted LC-ESI-MS, Hct variation effect was evaluated by comparing the intensity of each target metabolite obtained from samples with different Hct levels (low to high).

4.2.7 Data analysis

Untargeted metabolomics data analysis was done by using Bruker Data Analysis software (Version 4.1 (build 359)). Molecular features were extracted from the raw data files with set criteria including a signal-to-noise ratio of 3 (S/N=3), inclusion of adducts and lock mass calibration. The extracted molecular features were exported to the comma-separated values (csv) format using Bruker Data Analysis software. To select the best extraction solvent for metabolomics analysis using DBS samples, we further filtered the molecular features from csv files using Microsoft Excel 2007 (Albuquerque, NE). The filtration of reproducible molecular features from each extraction solvent was done in a step-by-step manner. In step 1, reproducible features from 3 replicate runs were extracted

doi:10.6342/NTU201804215

using the parameters including retention time (RT) < 0.2, mass accuracy tolerance < 0.05 Da, and peak area rsd < 25%; in step 2, the reproducible features from 3 DBS spots were

在文檔中以液相層析質譜儀開發藥物與代謝體於採血卡之分析方法 (頁 71-0)