97
4.1. Introduction
Emerging applications of metabolomics to translational research highlights the dysregulation of metabolites in disease states. [1-4] Quantification of endogenous metabolites enables the discovery of biomarkers for diagnosis or understanding disease etiology. [5-9] Liquid chromatography-electrospray ionization mass spectrometry (LC-ESI-MS) is one of the most frequently used techniques to quantify endogenous metabolites in bio-samples. [10-15] The matrix effects (MEs) which come along with the complicated sample matrices not only cause sensitivity loss, but also significantly decrease quantification accuracy. [16, 17]
To mimic the MEs encountered in real samples, a general approach to generate calibration curves for bio-sample quantification is to spike target analytes in blank matrix. [18, 19] However, “true blank matricies” are impossible to obtain when
measuring endogenous metabolites. [20-22] Due to the lack of a true blank sample, some studies use “mimic matricies” such as bovine spiked saline to establish calibration
curves. [23, 24] Even so, the “mimic matrices” are not able to reflect true MEs encountered in bio-samples. Besides, when using the standard spiked calibration curve to mimic the encountered matrix effect, the ME differences between individuals still is
98
not resolved. Therefore, the standard spiking in blank matrix method is not a feasible approach for quantifying endogenous metabolites.
To accurately quantify endogenous metabolites, two commonly used approaches include standard addition method (SAM) and stable isotope labeled-internal standard (SIL-IS) method. [25-27] SAM works by serial addition of target analytes into the specific sample to establish a standard addition curve, then the metabolite concentrations in the sample can be calculated by interpolation. SAM is effective to calibrate ME-caused quantification errors. [26-28] The major disadvantage of SAM method is that each sample has to be tested for several runs to obtain its standard addition curve, which makes the SAM method impractical for clinical analysis where large batches of samples need to be analyzed. SIL-IS is currently the gold standard for the quantification of endogenous metabolites. Since the physio-chemical properties of SIL-IS are almost the same with the target analyte, the signal changes that are caused by the ME will be similar, and can be effectively calibrated by SIL-IS. [29-31] Although this method could provide accurate quantification and efficient analysis, SIL-IS is not always available. The cost of SIL-ISs would be extremely high if multiple metabolites needed to be quantified.
99
An alternative, more economic approach to adjust MEs in LC-ESI-MS is the post-column infused-internal standard (PCI-IS) method. The PCI-IS method compensates MEs by correcting the signal intensity with the intensity of a post-column infused-internal standard. [32, 33] In the beginning, this method was developed to resolve the individual ME differences for same type of matrices. We previously proposed the use of PCI-IS in combination with matrix normalization factor (MNF) to adjust MEs between different biofluids with diverse matrix composition. The MNF between different biofluids can be acquired by dividing the analyte to PCI-IS response ratios obtained from different biofluids. MNFs were used to normalize the encountered MEs in various biofluids to the matrix effect encountered in standard solutions. PCI-IS can additionally tailor the correction of the matrix effect for individual samples. When using the PCI-IS method in combination with MNFs, the calibration curve generated from standard solutions can be applied to quantify the target analytes in various biofluids. This special character facilitates the use of standard solution generated calibration curve for quantifying endogenous metabolites in bio-samples.
Hormones perform many biological functions in the human body, for example, the development of male reproductive tissues, and the regulation of cognitive and physical energy. [34-40] The correlation of plasma hormone level to disease had been studied
100
comprehensively, and some of these studies show the relationship between different cancers and plasma hormone levels. [36, 39] The serum androgen level has also been reported to be lower in female rheumatoid arthritis patients at early disease stages. [41, 42] Furthermore, testosterone levels in the serum may be a predictive marker for cardiovascular disease. [43-45] For this reason, an accurate and simple hormone quantification method is helpful for related studies and disease prediction.
The growing importance of quantifying endogenous metabolites in clinical measurement indicates a strong need to have an effective method for accurate quantification. This study propose using PCI-IS in combination MNF to quantify endogenous metabolites. In order to acquire the MNF values in the specific matrix without interference from the endogenous metabolites, excessive amount of analyte was spiked into the specific matrix. We successfully applied this method to quantify androstenedione and testosterone in 50 plasma samples. The quantification results were compared with the results obtained by SIL-IS method using Pearson correlation test. We demonstrated that this PCI-IS in combination MNF method is an accurate and economic approach for the quantification of endogenous metabolites which shows great potential for applications in clinical measurement.
101
4.2. Experimental
4.2.1. Chemicals
Androstenedione, testosterone, progesterone,
4-androsten-3,17-dione-2,2,4,6,6,16,16-d7 (androstenedione-d7), and 4-androsten-17-OL-3-one-16,16,17-d3 (testosterone-d3) were purchased from Steraloids (Newport, RI). MS-grade methanol was purchased from Scharlau Chemie (Sentmenat, Barcelona, Spain). Acetonitrile (ACN) was obtained from J.T. Baker (Phillipsburg, NJ). MS-grade methanol was purchased from Scharlau Chemie (Sentmenat, Barcelona, Spain). Formic acid solution (99%) was purchased from Sigma (St. Louis, MO, USA).
4.2.2. UPLC-ESI-MS system
The LC system used in this study is an Agilent 1290 UHPLC system equipped with a binary solvent pump, an autosampler, a sample reservoir, and a column oven (Agilent Technologies, Waldbronn, Germany). The coupled mass spectrometer was an Agilent 6460 triple quadrupole system (Agilent Technologies, Waldbronn, Germany). Another Agilent 1260 quaternary solvent pump was applied for postcolumn infusion of the
102
PCI-IS. The PCI-IS (progesterone) was dissolved in ACN at 1 ng mL-1 and introduced into the ESI interface at a flow rate of 0.1 mL min-1.
A Kinetex C18 2.1×50 mm (2.6 m) column (Phenomenex, Torrance, USA) was employed for separations. The mobile phase consisted of 0.1% aqueous formic acid (solvent A) and 0.1% formic acid in ACN (solvent B). A 0.4 mL min-1 linear gradient elution was used: 0-2 min, 25-100% B, and column re-equilibration with 25% B for 1.2 min. The sample reservoir and column oven were maintained at 4 °C and 40 °C, respectively. The injection volume was 5 µL. Positive electrospray ionization mode was utilized with the following parameters: a 350 °C dry gas temperature, a 8 L min-1 dry gas flow rate, a 35 psi nebulizer pressure, a 400 °C sheath gas temperature, an 11 L min-1 sheath gas flow rate, a 4000 V capillary voltage, and a 2000 V nozzle voltage.
MS acquisition was executed in multiple reaction monitoring (MRM) mode. The transitions for androstenedione, testosterone, progesterone, 4-androsten-3,17-dione-2,2,4,6,6,16,16-d7 (androstenedione-d7), and 4-androsten-17-OL-3-one-16,16,17-d3 (testosterone-d3) were m/z 287.2→97, 289.2→97, 315.2→97, 294.2→100.1, and 292.2→97, respectively.
103
4.2.3. Sample preparation procedures
Androstenedione, testosterone, progesterone,
4-androsten-3,17-dione-2,2,4,6,6,16,16-d7 (androstenedione-d7), and 4-androsten-17-OL-3-one-16,16,17-d3 (testosterone-d3) were prepared separately in methanol at concentrations of 100 mg mL-1. The working solution was prepared by spiking an appropriate amount of each analyte stock solution into deionized (DI) water to obtain 100 ng mL-1 of diluted working solution.
The 3 blank plasma samples for testing the feasibility of this method were obtained from healthy volunteers. Protein precipitation was performed by mixing 40 μL plasma sample with 160 μL methanol. The deproteinized sample was centrifuged at 10,000 g for 15 min, and the supernatant was then filtered through a 0.22-μm PP membrane (RC-4, Sartorius, Göttingen, Germany) before UPLC-ESI-MS analysis. The 50 real plasma samples used to test the quantification accuracy of this developed adjustment method were collected from 50 healthy volunteers at different time points.
104
4.2.4. PCI-IS method combined with MNF method
The PCI-IS method assumes that the response ratio of target analytes to PCI-IS are proportional to the target analyte concentration. The analyte signal intensities at every time point in the chromatogram were divided by the PCI-IS responses at the respective retention times, and the ratios were used to generate the new adjusted chromatogram. To obtain the analyte to PCI-IS ratio of each time point, all of the data obtained from the Agilent triple quadruple was converted into comma separated values (csv) format and processed by Microsoft Excel 2007 (Albuquerque, NE). The information in the csv file included mass transition, retention time, and signal intensity. The MS acquisition rate was set to 1 spectra s-1.
The MNF was used for correcting the MEs encountered in the different matrixes.
With the aim of calculating the MNFSTD-plasma between standard solution and plasma matrix, a plasma sample was selected and spiked with the same amount analyte as standard solution. The response ratio of analyte and PCI-IS in the plasma was divided by in the standard solution to obtained MNFSTD-plasma. Finally, the response ratio of the analyte and PCI-IS in every plasma sample at each time point was divided by MNFSTD-plasma at the respective retention time point, and the corrected ratios were used to generate the corrected chromatogram.
105
4.2.5. SIL-IS method
With the intention of comparing the accuracy of this new correction method to SIL-IS method, a constant level of 1 ng mL-1 androstenedione-d7 and testosterone-d3 was spiked in the standard solution samples to establish calibration curves. Lastly, the calibration curves were applied to quantify the 50 plasma samples.
4.2.6. SAM method
In terms of comparing the accuracy of this method to the SAM method, 3 plasma samples were chosen to be analyzed using the SAM method. Each plasma sample was spiked with 500, 1,000, 3,000 pg mL-1 androstenedione and testosterone to plot standard addition curves.
4.2.7. Validation
4.2.7.1. Linearity, limit of detections (LODs), and limit of quantifications (LOQs)
Aliquots of the androstenedione and testosterone stock solution were added to deionized water to obtain 100, 500, 750, 1000, 3000, and 5000 pg mL-1 androstenedione
106
and testosterone standard solutions to build the calibration curve used for quantification.
Linear regressions were established by plotting the corrected ratios of androstenedione and testosterone with progesterone (PCI-IS) against androstenedione or testosterone concentrations. The limit of detection of each analyte was determined as the concentration at which the signal to noise ratio equals 3 (S/N = 3). The limit of quantification of each analyte was determined as the concentration at which the signal to noise ratio equals 10 (S/N = 10).
4.2.7.2. Precision and accuracy
For testing the precision, androstenedione and testosterone standard solutions were tested at low, medium, and high concentrations of 100, 1000, and 5000 pg mL-1 four runs a day for 3 days. The accuracy was also tested at low, medium, and high concentrations of 100, 1000, and 5000 pg mL-1 for 3 runs.
107 different types of sample matrices that exhibited distinct MEs. We previously demonstrated that the response ratios of analyte to PCI-IS in different types of biofluids can be calibrated by introducing MNF [ref]. MNF was obtained by spiking identical amounts of analyte and PCI-IS in study matrices. The calculation of MNF is based on
the following equation:
𝐶𝑎𝑛𝑎𝑙𝑦𝑡𝑒,𝑥
𝐶𝑃𝐶𝐼−𝐼𝑆,𝑥 =𝑅𝑎𝑛𝑎𝑙𝑦𝑡𝑒,𝑆𝑇𝐷,𝑥
𝑅𝑃𝐶𝐼−𝐼𝑆,𝑆𝑇𝐷,𝑥 =𝑅𝑎𝑛𝑎𝑙𝑦𝑡𝑒,𝑀,𝑥
𝑅𝑃𝐶𝐼−𝐼𝑆,𝑀,𝑥 × 𝑀𝑁𝐹𝑆𝑇𝐷−𝑀 (Eq. 1)
Where Canalyte,x represents the concentration of analyte, and Ranalyte,STD, x represents the analyte’s signal response in standard solution at time point x.
MNFSTD-M in Eq. 1 is used to correct the ratio differences between standard solution (STD) and testing matrix (M). The application of MNF in PCI-IS method corrects the response ratios of analyte and PCI-IS between different biofluids which facilitates quantifying target analytes in various biofluids by using calibration curve generated from the standard solution.
108
When using Eq. 1 to quantify endogenous metabolites in the M matrix, the existence of target analyte in the study matrix (Rendogenous analyte,m,x) will affect the calculation of MNFSTD-M values. This effect can be described using the following calculation of MNFSTD-M,we therefore propose to add excessive amounts of analyte into the studying matrix (M) to minimize the contribution of Rendogenous analyte,m,x in the calculation of MNFSTD-M. If the Ranalyte,m,x is much higher than the Rendogenous analyte,m,x , the contribution of Rendogenous analyte,m,x can be ignored. The concentrations of most endogenous metabolites in biofluids can be obtained from literature survey, and this information facilitates the design of spiking concentrations.
4.3.2. Using PCI-IS combined with MNF method for quantifying androstenedione and
testosterone in human plasma
In this study, we chose to quantify the endogenous metabolites androstenedione and testosterone in plasma to demonstrate the use of this method. Based on previous
109
reports, the concentration of androstenedione and testosterone in human plasma is generally less than 5 ng mL-1. [46-48] Accordingly, we spiked 100 ng mL-1 of androstenedione and testosterone into a representative plasma sample to calculate the MNFSTD-plasma values for both metabolites. The obtained MNFSTD-plasma values was used to adjust the response ratios of analyte and PCI-IS in plasma and standard solutions. The individual differences in MEs of test plasma samples can be additionally calibrated by the PCI-IS method. Progesterone was selected as the PCI-IS in this case due to its structure similarity with the test metabolites. We used a human plasma sample to test the correction efficiency of the proposed method. Before correction, the response of testosterone is smaller in plasma because of the MEs. After adjusting the signal intensity by PCI-IS combined with MNF method, the signal intensities were almost the same in the two matrices (Figure 4.1).
We additionally verified the quantification accuracy of the proposed method by comparing the quantification results with SAM and SIL-IS methods using three human plasma samples. Table 4.1 shows the quantitative results of the three methods. The quantification differences between the PCI-IS in combination with MNF method and the other two methods were less than 20%.
110
4.3.3. Validation of PCI-IS combined with MNF method
To verify the PCI-IS combined with MNF method for routine bio-analysis, this method was validated in terms of precision, accuracy, linearly and sensitivity.
Androstenedione and testosterone in human plasma was again used as the test metabolites. Considering the endogenous concentration, we spiked 100, 1000 and 5000 pg mL-1 of androstenedione and testosterone in plasma samples for method validation.
4.3.3.1. Precision and accuracy
The intra-day and inter-day precision of the PCI-IS combined with MNF method were tested four times a day for 3 days. The intra-day precision (n = 12) and inter-day precision (n = 3) of androstenedione at all tested concentrations were all less than 9%
relative standard deviation (RSD). The intra-day precision (n = 12) and inter-day precision (n = 3) of testosterone at all tested concentrations were less than 11% RSD (Table 4.2).
The accuracy of the PCI-IS combined with MNF method was tested by spiking 100, 1000, and 5000 pg mL-1 of androstenedione and testosterone in four plasma samples,
111
and was evaluated by % recoveries. The results showed that both androstenedione and testosterone have recoveries between 95.72 to 113.46% (Table 4.3).
4.3.3.2. Linearity, limit of quantifications (LOQs) and limit of detections (LODs)
One of the main advantages of the PCI-IS combined with MNF method is the ability to quantify target analytes in various sample matrices with calibration curves constructed using standard solutions. Therefore, the linearity was tested using androstenedione and testosterone standard solution at concentrations of 100, 500, 750, 1000, 3000, and 5000 pg mL-1 (Supplement Figure 4.2). The calibration curves were y = 1E-07x2 + 0.0021x - 0.0551 (r2 = 0.9999) for androstenedione, and y = 8E-08x2 + 0.0022x + 0.0114 (r2 = 0.9998) for testosterone. The LODs and LOQs were tested using the standard solution. The LOQs of androstenedione and testosterone were both 500 ng mL-1, and the LODs were 100 ng mL-1.
4.3.4. Comparison with SIL-IS method
To ensure the feasibility of the proposed new method, quantification results were compared with the well established clinical routine SIL-IS method. Plasma
112
samples from 50 healthy volunteers were both quantified by the PCI-IS in combination with MNF method and the SIL-IS method to obtain parallel data pairs. Figure 4.3 shows the correlation of the results obtained from the PCI-IS combined with MNF method and the SIL-IS method. The correlation coefficient of androstenedione and testosterone were 0.9802 and 0.9841 respectively.
The high quantification accuracy was attributed from the good calibration performance of the PCI-IS in combination with MNF method. If the PCI-IS in combination with MNF method was not applied to adjust the MEs in plasma samples, serious MEs caused significant ion suppression which gave negative bias of quantification results when using the calibration curves generated by the standard solution to quantify androstenedione and testosterone (Figure 4.3 a, c). Though adjusting the MEs caused signal changes by MNF and PCI-IS, the quantification errors were greatly reduced. In this case, all of the 50 tested samples showed quantification errors less than 20% (Figure 4.3 b, d).
Due to the high sensitivity and selectivity of LC-ESI-MS, it has been one of the major tools used to quantify endogenous metabolites. The effective application of the PCI-IS in combination MNF method for the adjustment of ME-caused-quantification-errors gives more reason to use this platform for measuring
113
endogenous metabolites. The complicated compositions of matrix components hinder the use of standard solution constructed calibration curves to quantify endogenous metabolites. As the true blank matrix is unavailable for the generation of analyte spiked calibration curves, the only solution for accurate quantification of endogenous metabolites are the SAM and SIL-IS methods in current practice. However, the SAM method is impracticable in routine clinical measurements, and the SIL-IS method is very costly, especially when multiple analytes need to be quantified. The PCI-IS in combination with MNF method used structure analogs to calibrate ME-caused-signal-changes. Single PCI-IS is capable to calibrate multiple analytes if the target analytes share somewhat structure similarity, our demonstration case showed that progesterone was effective to calibrate both androstenedione and testosterone. The proposed method overcomes the inherited difficulties in measuring endogenous metabolites by LC-ESI-MS, which open a new window for various bio-analysis.
4.4. Conclusions
This study proposed a novel approach to use the PCI-IS combined with MNF method for quantifying endogenous metabolites. The MNF was used to bridge the analyte response in standard solution to other biofluids, and PCI-IS additionally
114
calibrated MEs differences for each sample. This method allows the quantification of analyte concentrations in various biofluids using standard solution generated calibration curves which overcome the previous difficulties related to measuring endogenous metabolites where true blank matrix is unavailable. The application of this new approach on the measurement of androstenedione and testosterone in plasma samples revealed that the correlation between this method and SIL-IS method is higher than 0.98.
PCI-IS combined with MNF method was demonstrated to be an effective and economic approach, which could speed up targeted metabolomic studies for the discovery of clinical biomarkers and investigation of disease etiology.
4.5. References:
[1] A.H. Zhang, H. Sun, X.J. Wang, Serum metabolomics as a novel diagnostic approach for disease: a systematic review. Analytical and Bioanalytical Chemistry, 404 (2012) 1239-1245.
[2] A.H. Zhang, H. Sun, X.J. Wang, Recent advances in metabolomics in neurological disease, and future perspectives. Analytical and Bioanalytical Chemistry, 405 (2013) 8143-8150.
115
[3] Y.Y. Zhao, Metabolomics in chronic kidney disease. Clinica Chimica Acta, 422 (2013) 59-69.
[4] M. Ciborowski, J. Teul, J.L. Martin-Ventura, J. Egido, C. Barbas, Metabolomics with LC-QTOF-MS Permits the Prediction of Disease Stage in Aortic Abdominal Aneurysm Based on Plasma Metabolic Fingerprint. Plos One, 7 (2012).
[5] P. Voehringer, R. Fuertig, B. Ferger, A novel liquid chromatography/tandem mass spectrometry method for the quantification of glycine as biomarker in brain microdialysis and cerebrospinal fluid samples within 5 min. Journal of Chromatography B-Analytical Technologies in the Biomedical and Life Sciences, 939 (2013) 92-97.
[6] B. Warth, M. Sulyok, F. Berthiller, R. Schuhmacher, P. Fruhmann, C. Hametner, G.
Adam, J. Frohlich, R. Krska, Direct quantification of deoxynivalenol glucuronide in human urine as biomarker of exposure to the Fusarium mycotoxin deoxynivalenol.
Analytical and Bioanalytical Chemistry, 401 (2011) 195-200.
[7] M. Berna, L. Ott, S. Engle, D. Watson, P. Solter, B. Ackermann, Quantification of NTproBNP in rat serum using immunciprecipitation and LC/MS/MS: a biomarker of drug-induced cardiac hypertrophy. Analytical Chemistry, 80 (2008) 561-566.
116
[8] G.L. Milne, S.C. Sanchez, E.S. Musiek, J.D. Morrow, Quantification of F-2-isoprostanes as a biomarker of oxidative stress. Nature Protocols, 2 (2007) 221-226.
[8] G.L. Milne, S.C. Sanchez, E.S. Musiek, J.D. Morrow, Quantification of F-2-isoprostanes as a biomarker of oxidative stress. Nature Protocols, 2 (2007) 221-226.