Derivatization-free determination of estrogens in human plasma and serum by Liquid Chromatography/Tandem Mass Spectrometry

(1)

MSc Chemistry

Track Analytical Sciences

Master Thesis

Quantification of estrogens in human plasma by Liquid

Chromatography/Tandem Mass Spectrometry

by

Suying Wu

11425830

August 2019

54 ECTS-credits

Period: December 1

th

_{2018 to August 31}

th

₂₀₁₉

Supervisor/Examiner:

Examiner:

prof. dr. Govert Somsen

dr. Rob Haselberg

Department of Clinical Chemistry

Erasmus MC

(2)

ii

Foreword

This work described in this thesis constituted the end phase of my study (chemistry) at the University of Amsterdam, The Netherlands and was performed at the Erasmus Medical Center, Rotterdam, The Netherlands. The aim of this research was to develop an LC-MS/MS method for the analysis of estrogen in plasma/serum. The research was funded by the Erasmus Medical Center, Department of Clinical Chemistry, The Netherlands.

I would like to thank dr. T. Klein for supervising me throughout my internship. Further, many thanks are due to dr. R. Haselberg, to Prof. G. Somsen for giving me valuable advice and support always when needed.

(3)

iii

List of abbreviations

ACN: Acetonitrile

AEBSF: 4‐(2‐aminoethyl)benzenesulfonyl fluoride APCI: Atmospheric pressure chemical ionization APPI: Atmospheric pressure photoionization BEH: Ethylene Bridged Hybrid Technology CE: Collision energy

CI: Chemical ionization CXP: Collision cell potential DCl: Dansyl chloride

DMIS: 1,2-dimethyl-1H-imidazole-5-sulphonyl chloride DP: Declustering potential

E1: Estrone E2: Estradiol E3: Estriol

EI: Electron impact

ELISA: Enzyme-Linked immunoassays ESI: Electrospray ionization

FMP-TS: 2-fluoro-1-methylpuridinium-p-toluenesulfonate GC-MS: Gas chromatography-mass spectrometry

HSS: High Strength Silica Technology

LC-MS/MS: Liquid chromatography-tandem mass spectrometry LLE: Liquid/liquid extraction

LOD: Limits of detection LOQ: Limits of quantitation 2-ME2: 2-methoxyestradiol 4-ME2: 4-methoxyestradiol MeOH: Methanol MTBE: Methyl-tert-butylether NH4F: Ammonium fluoride 2-OHE2: 2-hydroxyestradiol 4-OHE2: 4-hydroxyestradiol PFP: Pentafluoropropionyl PP: Protein precipitation R2: Coefficient of determination RIA: Radioimmunoassays Rs: Resolution

SLE: Supported liquid extraction SPE: Solid-phase extraction TEA: Triethylamine

TMS: Trimethylsilyl UP: Ultrapure water

(6)

vi

Abstract

Estrogens, include estrone (E1), estradiol (E2), and estriol (E3), are well-known sex hormones related to both health and disease. In postmenopausal women, men, and children present extremely low concentrations of estrogen in serum or plasma, which could be problematic for proper clinical monitoring and diagnosing. The main goal of this work was to develop a liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for the analysis of estrogen in serum/plasma. Measuring of estrogen in picomolar was not achieved using the non-derivatization method. An additional derivatization step with dansyl chloride was used in sample preparation to enhance the ionization efficiency of estrogens. Unfortunately, the derivatization method is still unfinished and further investigation is needed, involving improvements in sample preparation, derivatization, and chromatographic conditions without significant problems.

(7)

1

1. Introduction

The human steroid hormones are the group of hormones derived from cholesterol that perform as chemical messengers that tell specific tissues to behave in a certain way1_{. Estrogens,}

androgens, and progestogens are the main steroids that play an essential role in biological processes, including development, hypothalamic programming, sexual differentiation, reproductive physiology, behavior, osmoregulation, and metabolism. The estrogen family, including estrone (E1), estradiol (E2) and, estriol (E3), functions through estrogen receptors (ERα and ERβ) in both gene regulation (genomic effects) and cell signaling (epigenomic effects)2_.

Recently, alteration of estrogen levels is frequently related to several pathological disorders such as postmenopausal breast cancer, coronary artery disease, and cognitive dysfunction3–5_.

Estrogens were primarily analyzed by immunoassay-based methods followed by gas

chromatography-mass spectrometry (GC-MS or GC-MS/MS) with derivatization techniques. Some of these assays were found to have a high level of inaccuracy, lack of sensitivity, and reproducibility and to be imprecise over a wide concentration range due to the low

concentration of estradiol observed in men and postmenopausal women. Due to the

derivatization process and long cycle time of the GC-MS/MS method, LC-MS/MS has been widely investigated because of its higher sample throughput and sensitivity. Most of the LC-MS/MS methods include a derivatization procedure to enhance the sensitivity results, time-consuming sample preparation, which is less proper for large-scale studies and obstructs its clinical

practicality setting. Therefore, in recent times, considerable efforts have been devoted to investigating an LC-MS/MS method without a derivatization procedure to simplify sample preparation. The existing LC-MS/MS methods without derivatization could exclusively measure one or two estrogens, and measuring all three types of estrogen in the plasma sample will require further investigation3,4,6–10_{. In clinical settings, an improved specific, sensitive,}

high-throughput non-derivatization LC-MS/MS method for simultaneous quantification of estrone, estradiol and, estriol in plasma and serum is dominantly required.

This project focuses on the development of a simple, accurate, and highly sensitive LC-MS/MS method without the derivatization process for the quantitation of estrogens in human and murine plasma and serum. The aim is to reach a limit of quantification of 10 pg/mL in 50 L sample matrix. The method will be employed in ongoing (murine) studies within a collaboration with the Dept. of Endocrinology & Reproduction to achieve a better understanding of the relationship between sex steroid hormones and disorders to human biology.

(8)

2

2. Theoretical background

In this Chapter, background information concerning function, metabolism, and quantification methods of estrogen are described.

2.1. Estrogens

Estrogens, including estrone (E1), estriol (E3),17α- and 17β-estradiol (17α- and 17β-E2), are a group of sex steroids with the structures as shown in Figure 1, indicating that these estrogens differ in their oxidation state through the number of hydroxyl (OH) groups. Steroid hormones typically contain 18–21 carbon atoms comprising a general structural scaffold formed by four bonded rings: a benzene ring labeled as ring A, two cyclohexane rings designated as rings B and C, and one cyclopentane ring designated as ring D. Estrogen have a common steroidal backbone with a phenolic hydroxyl group at position 3 of the carbon atom of the A-ring, illustrated in Figure 1. The variances among them are the substituents at position 15C and 17C in the D-ring. Estrone has a carbonyl group at 15C, β-estradiol has an OH group at the β-side of 15C, and estriol has two OH groups. Estrogen exerts various physiological activities on multiple organs such as regulating differentiation, growth, homeostasis in tissues of the human body, and ovaries in pre-menopausal females11,12_.

Nevertheless, estrogens show a negative effect as well as a positive effect on the human body, and studies show increased estrogen levels in the blood may raise the risk of inducing breast and endometrial cancer11,12_{. Endogenous and exogenous estrogen has been studied by scientists to}

reveal the origin of the biological activities, and these studies express that the physiological effect of estrogen is initiated upon binding with receptors, in most cases the estrogen receptor. The hydrogen-bond originated from OH at the A-ring, and the hydrogen-bond originated from OH or C=O at the D-ring of estrogen react with the DNA-bound receptors to form the ligand-binding domains (LBDs), leading to regulation of the physiological activities11_.

Estrogens are the primary female sex hormones, and they differ between reproductive and non-reproductive women. The regulation of the non-reproductive system in women, such as pubertal onset, fertility, menstrual cycle, and childbearing age are controlled by the ovary-produced estrogen, and naturally, a lower amount of estrogen was found in men. Non-reproductive tissues, including kidney, liver, heart, muscle, skin, bone, and brain, are the primary sources for the production of estrogen in non-reproductive females, as females before puberty or women after menopause. Besides, estrogens are synthesized primarily in the ovaries, corpus luteum, and placenta in premenopausal women. β-estradiol (E2) has the highest physiological activity, which is followed by estrone (E1) and estriol (E3). E2 is the most common type of estrogen in women and becomes potent during the premenopausal period, while E1 plays a more significant role after menopause. E3, produced mostly by the placenta, is the primary estrogen during pregnancy13–16_.

17α-E2, a stereoisomer of the more common 17β-E2, has been proven to have a low estrogenic function in humans due to its weak binding affinity for classical estrogen receptors, and in the brain, a novel estrogen ER receptor ER‐X shows a higher binding affinity with 17α-E2. Additionally, recent studies proposed that 17α-E2 may exert a higher estrogen activity in fish species than in mammals16–19_.

(9)

3 10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H O CH₃ 18 H H H

A

_B

C

D

Estrone (E1)

10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H

A

_B

C

D

17β-Estradiol (E2)

10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H

A

_B

C

D

17α-Estradiol (17α-E2)

10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H OH

A

_B

C

D

Estriol (E3)

Figure 1 Chemical structures of estrogen, consist of a steroidal backbone with four bonded rings and they vary in the number of OH-group

2.2. Estrogen metabolism and association with cancer

Pregnenolone originates from cholesterol and performs as a precursor intermediate for the biosynthesis of the steroid hormones. It is metabolized to progesterone and androstendione by the CYP17A1 gene. Androstendione is in equilibrium with testosterone, and both are further metabolized to estrogen by aromatase. In postmenopausal women, aromatase inhibitors are used to lessen the production of estrogen and to prevent the growth of breast cancer cells. Estrone (E1) is reversibly transformed to estradiol (E2) through the biocatalyst called 17β-hydroxysteroid dehydrogenase enzyme, and they are further metabolized by three pathways: 2, and 4-hydroxylation, and 16α-hydroxylation20–22_{, illustrated in Figure 2. Two and}

four-hydroxylated forms of estrogen are known as catechol estrogens. 2-hydroxylation pathway

The cytochrome P-450 enzyme CYP1A1, mainly expressed in liver tissue, catalyzes the hydroxylation of E1 and E2 to their catechol estrogens (hydroxyestrone and

2-hydroxyestradiol) at position C2 of the A-ring. The low binding affinity of 2-hydroxy estrogen with estrogen receptor (ER) was found, and a particular indication from the studies of breast cancer cell culture submitting that 2-hydroxyestrone and 2-hydroxyesradiol reduce cell growth and cell

(10)

4 proliferation. The 2-hydroxylated estrogen is either methylated by Catechol Ortho Methyl

Transferase (COMT) or oxidized to 2-estrogen quinones. The methylated metabolites of 2-hydroxy estrogen play a role in normal cell differentiation and cell death, which suppress tumor cell

proliferation and forming of new blood vessels. When COMT is inhibited and excessive reactive oxygen species (ROS), such as oxidized lipids, and heavy metals, are present, the 2-catechol estrogens are further oxidized to highly reactive 2-estrogen quinones. The 2-quinones bind covalently to DNA, forming adducts, and they are less harmful than 4-quinone estrogens, while the methylation of 2-hydroxyestradiol is at a more rapid rate than 4-hydroxyestradiol, therefore the generation of 2-quinones is limited20–23_.

4-hydroxylation pathway

The 4-hydroxylation pathway of E1 and E2 is similar to the 2-hydroxylation pathways. E1 and E2 are metabolized to 4-hydroxyestrone and 4-hydroxyestradiol by the cytochrome P450 enzymes CYP1B1 at position C4 of the A-ring. These catechol estrogens are further transformed into 4-methoxyestrogen (4-methoxyestrone and 4-methoxyestradiol) or 4-quinone estrogens. The electrophilic quinone estrogen reacts with glutathione, a tripeptide, under ideal conditions. When low glutathione levels occur, these highly reactive quinones may react with DNA creating adducts, causing DNA damage and DNA mutation, which increase the risk of cancer in estrogen target tissue such as breasts, uterus, ovaries, and prostate20,21,24,25_.

16-hydroxylation pathway

Another pathway of E1 is the 16α-Hydroxylation at position C16 of D ring by cytochrome P450 3A4 enzyme to produce hydroxyestrone and metabolized to estriol (E3). The

16α-hydroxyestrone metabolite is a potent estrogen, which can strongly bind to the estrogen receptor and stimulate cell proliferation. Widespread research over the last decades evaluated the

relationship between the ratio of 2-hydroxyestrone to 16α-hydroxyestrone and breast cancer risk in serum, plasma, or urine samples using enzyme immunoassay (EIA). The results of these studies were not consistent in either premenopausal or postmenopausal women. Cohort studies of estrogen metabolism using LC-MS/MS method shown that the 2-hydroxylation pathway: parent estrogens ratio and the 2-hydroxylation pathway:16-hydroxylation pathway ratio significantly related with a decrease in breast cancer risk, whereas an increase of breast cancer risk was associated with the ratio of 4-hydroxylation pathway to their methylated metabolites5,21_{. Black et}

al. observed a positive association between the ratio of 2:16α-hydroxyestrone in serum and aggressive prostate cancer, pathway studies in prostate cancer model systems are still lacking25_.

(11)

5

Figure 2 Estrogen metabolism and harmless (2-hydroxy-E1, 2-hydroxy-E2, 16-hydroxy-E1, and E3) or potentially toxic (4-hydroxy-E1 and 4-hydroxy-E2) estrogen metabolites

(12)

6

2.3. Analytical methodologies for estrogen

Steroids were commonly measured using immunoassays. Recently, GC-MS/MS and LC-MS/MS have become popular methods for estrogen analysis.

2.3.1. Immunoassays

Enzyme-Linked immunoassays (ELISA) and radioimmunoassays (RIA) were frequently used for the monitoring of estradiol and estrone in clinical diagnosis and research due to their low cost and routine nature. Both techniques are based on the action of an antigen (estrogen) that binds to specific antibodies. The detection of the antigen-antibody binding of ELISA is performed via incubation with a substrate to produce a measurable signal corresponding to the amount of estrogen, and radioactive scintillation counting is applied for RIA. Since immunoassays are dependent on antibody characteristics and possible cross-reactivity with structurally similar compounds, they lack selectivity, biomarker sensitivity, specificity, and accuracy are problems when measuring at low concentration in postmenopausal women, men, and children. Estrogen analysis is necessary for monitoring the health of postmenopausal women, men, and children. Several studies for the abundant steroids, such as testosterone and vitamin D, demonstrate an imprecision between reported concentrations and a bias for false positives using immunoassays over several analytical methods. The Endocrine Society is issuing a consensus announcement mentioning the avoidance of immunoassays for the steroid. In addition, most steroid

immunoassays are not certified by the Centers for Disease Control and Prevention

(CDC) Hormone Standardization (HoSt) Program. Therefore, to develop a robust, sensitive, and accurate method for quantification of estrogen at low concentration using mass spectrometry became popular in recent years3,26_.

2.3.2. Gas chromatography-Tandem mass spectrometry

The mobile phase of the gas chromatograph (GC) is an insert gas, usually helium, and the stationary phase is in the capillary column as a coating on the wall. Estrogen has low volatility, therefore chemical derivatization is required for the gas chromatography-tandem mass

spectrometry (GC-MS/MS) analysis. The derivatized estrogens are separated through the capillary column based on their affinities for the stationary phase and the boiling point. A temperature gradient is applied to the GC column through an oven. Electron ionization (EI) and chemical ionization (CI) have been used for MS detection. In chemical ionization, ions are produced by collisions between the analyte and the ions of the reagent gas (e.g., ammonia and methane). GC-MS/MS working in negative chemical ionization is preferred in the literature due to improved sensitivity. In electron ionization (EI), the capacity of producing a stable negative ion for some analytes is limited, and the negative chemical ionization will be complementary to the electron ionization. Therefore, most derivatization agents used for estrogen analysis contain a

halogenated group. The negative chemical ionization is capable of analyzing the halogenated analytes due to their high electronegativity. Pentafluoropropionyl (PFP) and trimethylsilyl (TMS) are generally used as derivatization agents for GC-MS/MS, the derivatization occurs at the C3 position of the A ring, whereas reaction with the D ring does not increase sensitivity3,7_{. However,}

GC-MS/MS is time-consuming as it routinely runs from 30 min to 1 h and requires extensive sample preparation (2-steps-extraction)3,7_{. Recently, one study reached a limit of quantification of}

0.5 pg/mL in 250 µL rodent serum using the GC method, though this approach needs to be verified in human plasma and extensive sample preparation, both liquid-liquid extraction and solid-phase extraction, was performed27_{. LC-MS/MS combining softer ionization techniques, APCI}

(13)

7 and ESI, are preferred to generate charged ions. There is a growing interest in bringing LC-MS/MS to the forefront in the clinical field26_.

2.3.3. Liquid chromatography-Tandem mass spectrometry

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) became a favored method for steroid analysis in the clinical world. Reversed-phase chromatography is exclusively applied, using a non-polar stationary phase, consisting of supporting material (silica, polymer) and bonded layer (C18, C8, phenyl), in conjunction with an aqueous, moderately polar mobile phase (water and organic solvent). The separation is based on the polarity of the analytes; less polar compounds retain in the stationary phase and elute later, and polar compounds elute earlier. C18 columns with their improved retention abilities and robustness are majorly selected for estrogen separation28_.

Limits of quantitation (LOQ) by LC-MS/MS vary over a wide range of 0.14 - 3000 pg/mL for estrone and estradiol. Generally, 0.1 - 2 mL of serum or plasma is tested, while 0.5 mL or lower volume is preferred for routine analysis without unnecessary blood consumption. The used sample amounts with different ionization techniques present different limits of detection. Mass spectrometers ionize molecules in the ion source and identify the generated ions according to their mass-to-charge (m/z) ratios in the mass analyzer. Three types of soft ionization techniques are frequently applied in estrogen analysis. Electrospray ionization (ESI) with positive mode analyses are the majority, moreover, alternative soft ionization modes are reported, i.e., atmospheric pressure chemical ionization (APCI) and atmospheric pressure photoionization (APPI). Estrogen analyses are performed in both positive and negative ionization modes related to the charge as a result of (de)protonation, and mobile phase additives such as formic acid and ammonium formate are commonly added26,29_.

2.3.3.1. Ionization techniques

Electrospray ionization (ESI) transfers ions from solution into the gas phase before the analyte reaches the mass spectrometer in three steps: (1) dispersion of charged droplets from a fine spray, followed by (2) solvent evaporation and (3) ion ejection into the gaseous phase. The LC eluent is sprayed (nebulized) into a chamber at atmospheric pressure by an electrostatic field and the nebulizing gas (e.g., nitrogen). The electrostatic field leads to the generation of a fine mist. Elevated temperature and another stream of heated nitrogen provide gas-assisted drying and the evaporation of the solvent in the droplets resulting in the charged droplets shrinking in size and an increase in the charge concentration in the droplets. Eventually, the electric field strength within the charged droplet reaches a critical point, which is kinetically and energetically possible for the ejection of ions into the gas phase. These emitted ions flow through a capillary sampling orifice into the mass analyzer, illustrated in Figure 3. According to the nature of the analyte of interest, positive or negative ion mode can be selected. In positive ion mode (ESI+), the capillary is the positive electrode, and the aperture plate of the entrance of the mass analyzer is the negative electrode, the reverse condition appears in negative ion mode (ESI−)30_.

Understanding of the ESI mechanism is still limited, despite its extensive application. Positive ion mode is generally preferred as more biologically relevant compounds are expected to ionize in this mode. However, the major advantage of the negative ion mode in the lower background signal. Diverse compounds have different ionization efficiency in both positive and negative

(14)

8 ionization modes. Ionization efficiency (IE) refers to the number of ions created from the analyte of interest in the ionization source. Higher IE allows better sensitivity and lower detection limits. The ionization efficiency of a compound is determined by the physicochemical properties (e.g. acid‐base properties, hydrophobicity, molecular volume), the solvent used for analysis, sample matrix31–33_{. Hermans et al. have shown that the molecular volume of amino acids affected IE the}

most, but pH and proton affinity may also affect ESI efficiency31_{. Several research groups have}

developed different models to predict the ionization efficiency for compounds in different solvent mixtures in both modes, while these approaches are still inapplicable for biological sample

matrix34–37_{. A complex sample matrix may cause matrix effect, which can decrease or increase the}

ESI/MS signal of the compound of interest. Matrix effect is usually linked with ion suppression and is expected to occur from the competition of compounds for the surface charge in the ESI droplet. It was observed that increased hydrophobicity of the matrix compound increases both sensitivities as well as the matrix effect in ESI- mode. If non-volatile matrix components are present in the sample, the ion signal response will be reduced via precipitation of the analyte on the ESI interface. Matrix effects must be reduced or compensated to obtain accurate results, and this can be reached by an extensive sample clean-up procedure. This is time-consuming, though, and the risk of the loss of analyte of interest is commonly increased during numerous consecutive clean-up steps. The use of two-dimensional LC provided an improvement of separation efficiency, which may decrease the amount of coeluted compounds with the target analyte37,38_.

Figure 3 Electrospray ion source involved the three steps: (1) spraying of a charged droplet, (2) solvent evaporation, and (3) ion ejection into the gaseous phase.

In atmospheric pressure chemical ionization (APCI), the LC eluent is sprayed through a heated vaporizer (250 – 550°C) at atmospheric pressure, and the samples are vaporized with the help of nitrogen gas and by heating. The gaseous solvent (S) and sample (M) molecules are ionized by a corona discharge needle to form a radical cation (M+●_{and S}+●_{). The corona needle is a charged}

electrode that produces an electric field to ionize nearby molecules. Frequent collisions between the ions and molecules enable to transfer charge from an ion to another neutral. The collision of an ionized solvent ion with an analyte molecule generates a direct charge transfer to form a radical cation analyte ion (S+●_{+ M → M}+●_{+ S).}

Additionally, a collision of solvent ions with a neutral produce a proton donation from the

molecule (S+●_{+ S → [S+H]}+_{+ S[-H]). The generated ionized solvent is able to ionize the analyte via}

proton transfer ([S+H]+_{+ M → [M+H]}+_{+ S). The resulting analyte ions (M}+●_{or [M+H]}+_{) pass}

through a capillary sampling orifice into the mass analyzer39_{, shown in Figure 4. APCI is less liable}

to ion suppression, and this technique reached a low limit of 1 pg/mL in saliva samples for estradiol analysis40_.

(15)

9

Figure 4 Atmospheric pressure chemical ionization (APCI) ion source involving vaporize the sample following by collision between the ions resulting in charge transfer to create radical cation analyte ions.

Atmospheric pressure photoionization (APPI) for LC-MS/MS is relatively the newest among the methods of soft ionization in mass spectrometry. In APPI, the LC eluent is transferred in the gas phase by nebulizing gas and vaporizer at atmospheric pressure. The solvent and analyte

molecules are exposed to ultraviolet light from a discharge lamp, which generates photons with a specific ionization energy level. The range of energies is carefully selected to ionize as many analyte molecules as possible, whereas reducing the ionization of solvent and unwanted molecules41_.

The photons (hν) excite both the analyte molecule (M), which loses an electron (e-_{) to creating}

a radical cation (M+•_{) and the solvent molecules, which are present at a considerably higher}

amount. A small fraction of the collisions between the ions induce a chemical reaction in which the solvent molecule donates a proton to the analyte molecule ([M + H]+_{). The resulting ions}

(M+•_{and [M + H]}+_{) pass through a capillary orifice into the mass analyzer, as shown in Figure 5.}

However, direct ionization of an analyte molecule occurs with a low‐statistical possibility, partly due to the solvent that reduces the photons emitted by the discharge lamp. The addition of a photoionizable intermediate-acted compound named dopant (e.g., toluene) can increase the percentage of ionized analyte molecules41–43_.

(16)

10

Figure 5 Atmospheric pressure photoionization (APPI) ion source involving vaporizing the sample followed by emitted photons from the discharge lamp to produce radical cation and chemical ionization is additionally performed.

Rahkonen et al. presented a comprehensive comparison of these three ionization modes and polarity combinations for estradiol analysis. It demonstrated that APPI in negative mode had the lowest LOQ at 0.14 pg/mL. Using this approach, low concentrations in pooled serum were detected using ammonium hydroxide as an additive. Although advantages of APPI over ESI/APCI exist for less polar and non-polar analytes, APPI is not commonly explored or available in clinical laboratories, and the majority of routine clinical assays are still using ESI44_.

2.3.3.2. Derivatization

When using positive ionization modes, the sample preparation requires an extra step, in which the analytes react with a derivatization agent to generate permanently charged species, which enhances the sensitivity and overcomes poor ionization efficiency. As in GC-MS, the hydroxyl group of the phenolic A ring in the C3 position is favored during derivatization due to the electron-withdrawing effect of the aromatic ring45_.

The published derivatization methods for the analysis of estrone and estradiol involve the use of dansyl chloride46–48_{, N-methyl-nicotinic acid N-hydroxysuccinimide ester}49_{, 2-fluoro-1-methyl}

pyridinium-p-toluene sulfnate50_{, methyl-1-(5-fluoro-2, 4-dinitrophenyl)-4,4-dimethylpiperazine}51_,

isomers of 1,2-dimethylimidazole-sulfonyl chloride46,52_{, 3-bromomethyl-propyphenazone}53_{, and}

pyridine-3-sulfonyl chloride54_{. Dansyl chloride is the most commonly used derivatization agent for}

LC-MS. Figure 6 shows the reaction of the phenolic hydroxyl group of E2 with dansyl chloride via nucleophilic aromatic substitution. The presenting basic nitrogen group to the estrogen molecule allows an improvement of the positive mode ionization under acidic conditions through reducing of the pKa of the 3’-OH group of the A ring55_{. The specificity of the fragment ions of dansyl}

(17)

11 chloride derivatives is hampered since the product ions are generated by the derivative moiety, thus it is not specific for the analyte by mass50_{. Furthermore, dansylated products are sensitive to}

light, and the by-products of derivatization can interfere with analyses resulting in the natural [M+2] isotopomers of E1 may produce a background noise within the mass transition of the E2-derivative56_{. Due to the addition of extra complexity during sample preparation, it could be}

contributing toward some degree of error and increased sample preparation time. Consequently, derivatization techniques are less favored in the clinical setting. Methods without derivatization are more commonly used in negative ionization mode, reporting the lower limit of quantification of 3 pg/mL for estradiol in a 0.29 mL blood sample57_{. However, a method using the negative}

mode is still to be extended to involve estrogen metabolites and validated by standardization to reference materials. Reference materials for E2 in serum could be obtained from the Institute for Reference Materials and Measurements (IRMM, Geel, Belgium).

3 O H OH CH₃ H H H A B C D E2

+

N C H₃ CH₃ S Cl O O Dansyl chloride NaHCO₃, 60°C -HCl A N C H₃ CH₃ S O O O 3 O H C H₃ H H _H Dansyl-E2

(18)

12

2.3.3.3. Mass analyzer

The mass analyzer, also called ion separator, receives ionized analytes from the ion source and separates them based on mass-to-charge ratios (m/z). Quadrupole, ion trap, time of flight, orbitrap, or combination of two of these mass analyzers, named tandem MS or MS/MS, are frequently used for estrogen analysis. When ions flow through a magnetic or electrical field, their m/z ratio causes particular movement, and this is the main separation principle of ions in MS. In this project, a triple quadrupole mass spectrometer was used for the determination

of estrogen in human plasma. A quadrupole mass analyzer contains four parallel metal rods kept at equal distance, with two opposite rods have an applied positive potential, and the other two rods have a negative potential, see Figure 7. A direct current (DC) and alternating current (AC) voltages are applied to the diagonally placed pair of rods, which affected the trajectory of ions traveling down the flight path. Ions generated in the ion source travel parallel to the rods in the z-direction with oscillatory movement in the x-y plane. Some of the ions show unstable trajectories. These ions oscillate with increasing amplitude and hit the rods of the quadrupole, thrown out of their original path, and fail to reach the detector. Ions that are successfully guided through the quadrupole filter possess stable trajectories, and these are recorded by the detector. The amplitude of oscillation can be set for desirable m/z ratios that are “stable” by changing the DC and RF voltages. Single quadrupoles were primarily established to enable operation in two modes: scan mode and selected ion monitoring mode (SIM). In scan mode, a range of mass-to-charge ratios was monitored, whereas discrete mass-to-mass-to-charge ratios were selected in SIM mode, as shown in Figure 8. SIM mode is significantly more sensitive, though information about fewer ions was provided. Scan mode is typically used for qualitative analyses or quantitation when all analyte masses are not known in advance, and SIM mode is used for quantitation and monitoring of target compounds. Improvements led to the presentation of tandem quadrupoles, consist of three quadrupoles set up in a linear model often called “triple-quadrupoles,” which support double mass filtering. In triple-quadrupole (Q1), the first quadrupole selects the precursor ion, following by collision-induced dissociation (CID) in the second stage (quadrupole or octopole), called the collision cell (Q2). Collision-induced dissociation is a process, which precursor ions are fragmented to their product ions by colliding them with neutral molecules (gas, such as nitrogen or helium). The product ions resulting from CID are associated with the molecular structure of the analyte ions and can be detected by a third quadrupole mass analyzer (Q3) to obtain structural information, known as multiple reaction mode (MRM), see Figure 9. A significant advantage of tandem MS is its capability to remove non-analyte ions in the Q1 stage. Quadrupoles tend to be the simplest and cheapest mass analyzers, and are commonly used in routine clinical analysis30,58_.

(19)

13

Figure 7 Schematic representation of a quadrupole mass analyzer. The quadrupole is composed of four parallel rods. Two opposite rods are connected and set to a positive potential and the other ensemble is set a negative potential. The settings of DC and AC voltages determine the trajectory of ion m/z59_.

Figure 8 Scan mode and SIM mode operated by the quadrupole mass analyzer. Scan over a range of mass-to-charge ratios or select between just a few60

(20)

14 Figure 10 shows different qualities of the mass resolution, also called mass resolving power, and the resolution indicates the ability of the mass analyzer to separate one mass from an adjacent peak. Orbitrap yields the highest resolution, followed by TOF, ion trap, and quadrupole58,61_.

Metabolism study is focused on the identification of a handful of metabolites or biochemical intermediates or quantitation of selected compounds that may play a role in the pathways affected by some metabolic intervention62_{. An LC-MS/MS equipment with high sample}

throughput, production of detailed and accurate structural information is desired in metabolism study. The commercially available high-resolution MS, such as Orbitrap and time of flight, is the cornerstone of any successful metabolism study62_{. The Ion trap-Orbitrap mass spectrometer}

offers a high mass accuracy and high resolution at a high price, which is still unaffordable for all drug companies and university medical centers.

Figure 10 Mass resolving power/resolution, which orbitrap profits the highest resolution followed by TOF, ion trap, and quadrupole63_.

(21)

15

2.3.4. Sample preparation

During initial analytical assay development, appropriate elimination of interfering compounds and decreasing background noise by sample preparation is crucial. Estrogens must be efficiently extracted from the matrix of choice, and this can achieve by liquid-liquid extraction (LLE), solid-phase extraction (SPE), or supported liquid extraction (SLE). Figure 11 presents the entire workflow schematically for the analysis of estrogen by GC-MS/MS and LC-MS/MS.

Figure 11 Schematic workflow for the analysis of estrogen by GC-MS/MS and LC-MS/MS, including sample treatment, derivatization, separation, and detection.

(22)

16 Liquid/liquid extraction (LLE) is an extraction technique, which offers an inexpensive process to extract the target estrogen from the aqueous phase (i.e., serum, plasma) into the organic phase, which should have good solubility for the target analytes. In this project, LLE was selected to treat the plasma samples. The extraction efficiency of the analytes from the aqueous into the organic phase is determined by the polarity of both the organic solvent and the analyte of interest. The solvent with less density will be the upper layer, while the more dense solvent will be the lower layer. After extraction, a large difference in density is essential when the two phases have to be partitioned in a separator. LLE has been widely used for the sample clean-up of

estrone and estradiol in serum. The extraction of biological samples was recommended to use less polar solvents, the less polar the solvent, the more selective it is. Therefore, the solvent of choice is typically the least polar one in which the analyte is still soluble. Methyl tert-butyl ether (MTBE), ethyl acetate, diethyl ether, dichloromethane, n-hexane, or mixtures of these organic solvents are commonly selected for extraction of estrogens into the organic phase. Diethyl ether is a highly volatile solvent with low boiling point and low viscosity, and it has a limited capacity for solubilizing water (60 g/L at 25 °C). These features are ideal for the use as the non-polar solvent in LLE. Diethyl ether had the highest extraction efficiency for the estrogens, and it could tend to form explosive peroxides9_{. Safer alternatives to diethyl ether are the MTBE and hexane-ethyl}

acetate mixture with the different volume ratio from 10% to 90%, affording relatively high recoveries and its low toxicity64_{. Disadvantages of LLE for routine analysis rely on its manual}

procedure, usually being time-consuming and sample can cause loss through transfer between the tubes, possibly contributing to irreproducibility65,66_.

2.3.5. Analytical challenges

Estrogens are typically present at extremely low concentrations, from picomolar to nanomolar, in complex biological matrices69_{. Matrix effects, ion enhancement, and primarily ion suppression are}

generally noticed when using ESI or APCI.Subsequently, these effects are affected by endogenous particles in the sample matrix. Ion suppression, which is lowering the ionization efficiency of a species of interest by the presence of other species present in the solvent, reduces the signal intensity and may lead to a decreased the method sensitivity70_{. Besides, ESI in negative mode}

shows lower ionization efficiency of estrogens compared to positive modes, which decrease the response of the analytes and can result in a higher limit of quantification51_{. The third challenge is}

the structural-related estrogen metabolites of a similar mass, as 2-hydroxy-E2 and 4-hydroxy-E2 are each enantiomer, effort should be taken to obtain a well-resolved chromatographic

separation. Fragment ions of dansyl chloride derivatized estrogen has the same mass, which may decrease the specificity26_._{Therefore, alternative derivatization approaches should be}

investigated.

In summary, inadequate sensitivity and specificity are the current two technical difficulties for analyzing estrogen in complex biological samples for LC-MS/MS-based assays. The sensitivity of a bioanalytical method is the lowest concentration of the analyte, which can be determined with acceptable precision and accuracy. The term specificity describes the ability of the bioanalytical method to produce a signal only for the analyte of interest and not for other interfering

(23)

17

2.3.6. Parameters for consideration during method development

An adequately designed biological method development should contain the following steps: understanding the chemistry of the analytes and the drug product, preliminary method conditions to reach minimally acceptable separations, investigating sample preparation

procedures, method optimization, and method validation. In order to have an efficient method development process, the following parameters will be used for consideration: sample

preparation, chromatographic separation, and validation71_.

Sample preparation focuses on the selection of the sample solvent for extraction, sample clean-up procedures, and sample matrix amount, e.g., 500, 200, 100, or 50 l plasma. The effect of sample solvents, including the types of organic solvent, extraction techniques, and extraction efficiency will be investigated. Different extraction and clean-up techniques such as solid-phase extraction (SPE), liquid-liquid extraction (LLE), and protein precipitation (PP) or combination of these techniques have been used for steroid hormone analysis. Decrease of matrix suppression may be achieved by extensive sample clean-up, while it is a time-consuming process, these factors should be taken for consideration during method development. When estrogens ionize poorly by negative mode ESI, chemical derivatization in combination with positive mode ESI can be used to reach the detection of lower concentrations of estrogen72_.

Chromatographic stationary phases seek satisfactory retention and separation of the analytes from each other to support MS detection. Classically, a value of 1.5 is the minimum required for baseline resolution for two equally size Gaussian-shaped peaks. Resolution (Rs) of two is

suggested to use as a minimum to account for day to day variability, non-ideal peak shapes, and differences in peak sizes during method development. Resolution can be improved by increasing of three factors: retention factor (k), selectivity (), and separation efficiency, known as column plate number (N). Although the increase of selectivity is the most effective approach, it suffered from practical difficulties. The retention factor is typically the most straightforward parameter to adjust by changing the mobile phase composition. The increase of plate number is accomplished by longer columns or smaller particles, while it is eventually limited by pressure. An analysis time of about 5-10 minutes per injection is sufficient and desired in routine analyses72_.

Biological method validation is the process of verifying that an analytical method is acceptable for diagnostics routine purpose. The validation of bioanalytical methods is performed following the guidelines of ICH. Validation of the liquid chromatography-mass spectrometry methods should include an evaluation of the linearity range, limits of detection (LOD) and quantification (LOQ), specificity, precision, accuracy, sensitivity, selectivity, ruggedness, reproducibility, stability, matrix effects, and carry-over effect. It is particularly recommended to process a partial validation during optimization to examining linearity, LOQ, precision, and accuracy of the present method72_.

(24)

18

3. Development of a derivatization-free method by negative ion

mode LC-MS/MS for the analysis of estrogen in plasma

The first phase of this project was to develop an LC-MS/MS method for the detection of E1, E2, and E3 using a negative ion mode. Most reports have resorted to chemical derivatization to increase the sensitivity of estrogen by LC-MS/MS, which introduces an additional step that

increases complexity and may present variability. Few estrogen methods are available that do not require chemical derivatization, and the aim was to develop an LC-MS/MS method for estrogen in plasma without chemical derivatization. When an appropriate separation method of E1, E2, and E3 were developed, the estrogen metabolites, included 2-hydroxyestradiol (2-OHE2),

4-hydroxyestradiol (4-OHE2), 2-methoxyestradiol (2-ME2), and 4-methoxyestradiol (4-ME2), will be investigated in the next phase.

3.1. Chemicals and reagents

Estrone, estradiol, estriol, 2-hydroxyestradiol, hydroxyestradiol, 2-methoxyestradiol, and 4-methoxyestradiol, ammonium fluoride, triethylamine, diethyl ether, methyl-tert-butylether, and ethyl acetate were purchased from Sigma-Aldrich (St. Louis, MO, USA). LC-MS grade acetonitrile (ACN) and methanol (MeOH) were purchased from Biosolve BV (Valkenswaard, the Netherlands). Ultrapure (UP) water was prepared with a Millipore Milli-Q Integral 10 Water Purification System (Molsheim, France). n-Hexane was acquired from Merck (Darmstadt, Germany).

3.2. Preparation of stock solutions and calibration standards

A stock solution of 1 mmol/L from E2, and 10 mmol/L from E1, and E3 was made and dissolved in methanol. The solutions were stored in a fridge (4 °C) in sealed vials. Eight-point calibration solutions were prepared with the stock solution by dilution in a mixture of acetonitrile and water (50:50, v/v).

3.3. Sample preparation

The fresh human plasma specimens were collected from rest material; blood samples of anonymized male patients after measurements for clinical purposes at the Erasmus Medical Center, Department of Clinical Chemistry, The Netherlands. These samples were stored in a fridge at 8 °C until use. Liquid-liquid extraction was used to extract the analytes from plasma. For each sample, 500 μL plasma was added to a 5-mL glass tube. 2 mL extracting agent, diethyl ether, was added to the specimen. The estrogens were extracted once with the extracting agent by vigorous mechanical shaking for 10 min, and the resulting mixtures were centrifuged for 10 min. The upper layer was transferred to another 5-mL glass tube, and the extracting agent was dried under nitrogen at 40 °C. The residue was reconstituted with 300 μL acetonitrile-water (50:50, v/v). The solution was transferred to a 96-well plate, and 10 μL of the sample was

(25)

19

3.4. LC-MS/MS Instrument parameters

The LC-MS analyses were performed with an AB SCIEX QTRAP 5500 and an AB SCIEX QTRAP 6500 tandem mass spectrometer equipped with Shimadzu Nexera LC-30 AD system. LC-MS/MS

operation and data acquisition were controlled by Analyst® 1.5.2 from AB SCIEX.

The separation of estrogens was achieved with a C18 column (Acquity Waters UPLC HSS T3, 1.8 μm, 100 mm × 1.0 mm). The mobile phase used in the chromatographic separation consisted of a binary mixture of eluent A (0.1 mM ammonium fluoride in UP) and eluent B (0.1 mM

ammonium fluoride in acetonitrile). Gradient elution was performed over 10 min at a flow rate of 0.12 mL/min, and the linear gradient steps were as follows: from 0 min to 1.0 min, 25% eluent B, from 1.0 min to 3.0 min, from 25% to 95% eluent B; from 3.0 min to 5.5 min, 95% eluent B; from 5.5 min to 6.5 min, from 95% to 25% eluent B; from 6.5 min to 10 min, 25% eluent B. The column temperature was kept at 40 °C. The mass spectrometer was fitted with electrospray ionization (ESI) TurboIonSpray ion source, and the estrogens were detected in negative ESI mode. The dwell times were 20 msec for multiple reaction monitoring (MRM) low mass mode. The ion spray voltage was set at -4500 V, and the ionization source temperature was 550 °C. The curtain gas (CUR), ion source gas 1 (GS1), and ion source gas 2 (GS2) were set at 25 psi, 15 psi, and 10 psi, respectively.

(26)

20

3.5. Results and discussion

3.5.1. Initial LC-MS/MS conditions

MS tuning

The first practical stage is to determine the optimum conditions for ionization of the analytes using MS tuning, which is processed by introducing a standard solution directly into the ion source using a syringe infusion pump. The analyte travels from the entry point of the MS, meanwhile, a potential run across the orifice, which is the cone voltage, also called declustering potential (DP). The optimal DP is specific for each analyte and prevents ions from clustering with adducts (e.g., sodium) or neutral species. Subsequently, the analyte flows to Q1, where the precursor ion was identified. Upon entering the collision cell (Q2), an inert gas, typically nitrogen or argon, is applied to induce the fragmentation of the precursor ion to generate product ions using appropriate collision energy (CE). Collision energy is the amount of energy that the

precursor ions receive as they were enhanced into the collision cell, a proper CE will yield higher fragment ion intensity. The collision cell potential (CXP) was applied at the exit of Q2 and, this potential is tuned to ensure adequate ion acceleration out of Q2 and into Q3. The fragments move to Q3 where the product ions were identified. MRM method was set up by a negative ESI mode, which was used to identify the dehydrogenation precursor [M − H]-_{and product ions for}

E1, E2, and E3. Analyte-dependent parameters (DP, CE, and CXP) were tuned to produce the most intense mass spectrum signals by infusion of a 1 M solution of each estrogen, as shown in Table 1 and Table 2, operated by AB Sciex 6500 and AB Sciex 5500. Two product ions with the highest intensity were selected as the quantifier ion and the qualifier ion for E1, E2, and E3. The most intensive MRM transitions of 269.20 > 145.0 m/z, 271.05 > 183.0 m/z and m/z 287.00 > m/z 171.0 were selected as quantifiers for E1, E2 and E3, respectively. E1, producing the 143 m/z fragment, and the other two estrogens, producing the 145 m/z fragments, were used as the qualifier ion. A quantifier ion was used for distinguishing the target compound, preferably separated the target compound from any others with similar retention times. The presence of qualifier ions in the correct amounts comparative to the target ion indicated the target

compound identification correctly.

Table 1 MRM parameters for quantitation and confirmation transitions for estrogens in AB Sciex

6500 Analyte Average mass Precursor ion [M − H] -Product ion (m/z) CE (V) CXP (V) DP (V) E1 270.37 269.20 183.0 -51 -20 -118 159.0 -45 -14 145.0 -46 -18 143.0 -69 -14 E2 271.38 271.05 183.0 -53 -25 -78 145.0 -50 -18 143.0 -63 -21 E3 288.38 287.00 211.0 -50 -11 -125 171.0 -48 -19 145.0 -52 -17 143.0 -63 -14

(27)

21

Table 2 MRM parameters for quantitation and confirmation transitions for estrogens in AB Sciex

5500 Analyte Average mass Precursor ion [M − H] -Product ion (m/z) CE (V) CXP (V) DP (V) E1 270.37 269.0 145.0 -49 -18 -120 143.0 -66 -18 E2 272.38 271.0 183.0 -51 -22 -74 145.0 -49 -17 143.0 -66 -20 E3 288.38 287.1 171.0 -49 -18 -125 145.0 -50 -16 143.0 -63 -15 2-OHE2 288.38 287.0 254.9 -55 -20 -82 199.1 -55 -24 161.9 -48 -16 4-OHE2 288.38 286.9 255.0 -51 -26 -78 198.7 -56 -21 160.8 -52 -15 2-ME2 302.41 301.0 286.0 -32 -23 -97 199.0 -53 -14 171.0 -60 -19 4-ME2 302.41 301.1 286.0 -30 -23 -73 159.1 -73 -21 Chromatography conditions

The initial LC condition was tested with E2, which is the major estrogen sex hormone in humans. The molecular weight of the estrogen is below 2000 g/mol and organic soluble hence reversed-phase chromatography as a separation model was preferred. Ammonium fluoride (NH4F) is

commonly used as an additive in the mobile phase. Preliminary mobile phases consisted of 0.1 mM ammonium fluoride both in water (A) and in 90/10 (v/v) acetonitrile/water (B). A previously used Waters HSS T3 C18 column (2.1 x 100 mm, 1.8 µm) was used, and the column temperature was maintained at 40 °C. Gradient elution was established with a preliminary condition of 25% B at a flow rate of 0.25 mL/min kept at 2.0 min. The mobile phase B was linearly increased from 25% to 65% within 4.0 min, maintained for 2.0 min at 65% B and, returned to the initial condition of 25% for equilibration of 4.0 min. The injection volume was 10 μL, and the total LC run time per sample was 12.0 min. For practical considerations, a flow of 0.208 mL/min is used for 2.1 mm ID columns to maintain linear velocity. The used flow rate in the literature varied from 0.20 to 0.30 mL/min. In this case, a flow of 0.25 mL/min provided reduced analysis time and appropriate peak shape with the retention time of 4.42 min for E2, as shown in Figure 12. In the negative ESI-MS/MS spectrum of E2 (Figure 13), the most intense fragment ions of E2 (183, 145, and 143 m/z) were formed with the ring cleavage during retrocyclization (modes 1 and 3)73_{, as shown in Figure}

14. These product ions were further stabilized by conjugated double bonds and the aromatic A-ring73_.

(28)

22

Figure 12 Chromatograms of E2 sample without plasma in a concentration of 1 µM, operated by AB Sciex 6500

Figure 13 Negative mode ESI-MS/MS product ion mass spectra of E2 and proposed structures for the fragment ions of deprotonated E273_{(1 µM).}

(29)

23

Figure 14 Proposed recyclization modes of E2 during negative ESI-MS/MS73

O– OH

A

B

C

D

RC Mode 3

RC Mode 4

RC Mode 2

RC Mode 1

(30)

24

3.5.2. Selection and optimization of the chromatographic system

Numerous columns and additives were examined to reach the optimum chromatographic

condition using a neat standard of estrogen (without sample matrix). It was reported that the use of acetonitrile instead of methanol resulted in a better selectivity with the appropriate

separation, thus a water/acetonitrile gradient with additive was selected for further method development4,6,7,9_{. Ammonium fluoride (NH}

4F) was found to promote estrogen ionization and

signal enhancement4,7,9_{, while Lozan et al. showed that the use of 0.1% TEA as addictive had the}

highest signal improvement with a factor of 7 compared to the signal of E2 without additives6_{. In}

the negative ionization mode, fluoride ions presented strong basic characters in the gas phase, enable to capture protons from the neutral analytes, forming hydrogen fluoride (HF), or both [M+F]−_{ions and [M+FHF]}−_clusters74,75_{. Hydrogen fluoride is a hazardous gas, and excessive using}

ammonium fluoride may permanently damage the HPLC column76_.

Moreover, HF may be formed in the waste container due to mixing with other compounds, thus proper discarding of the eluent waste is essential. AB Sciex recommended using small

concentrations and injection of numerous blank solvents after the run to help flush the lines. The concentration of the addictive affects the sensitivity of the method. Ammonium fluoride with a concentration of 0.1 mM or 0.2 mM was used in most approaches. In this project, 0.1 mM ammonium fluoride (NH4F) was selected to protect the analytical column, limiting HF forming in

the waste container, and 0.1% triethylamine (TEA) was also tested.

Stationary phase

The tested columns included a Waters HSS T3 C18 column (2.1 x 100 mm, 1.8 µm), a Waters HSS T3 C18 column (1.0 x 100 mm, 1.8 µm), and a Waters BEH C8 (2.1 x 100 mm, 1.7 µm). A used Waters C18 column with ID of 2.1 mm was initially chosen, followed by an aged column with smaller internal diameter Waters HSS T3 C18 (1.0 x 100 mm, 1.8 µm) to improve the sensitivity and increase ‘in-peak’ analyte concentration (average concentration across the peak). The intensity of a 1 M E2 standard was 5.6e6 cps using a C18 column (ID 2.1 mm) when a smaller ID of 1.0 mm of a C18 column was chosen, an intensity of 4.6e6 cps for E2 was observed with a 0.5 M standard, illustrated in Figure 15. The backpressure increased from 350 to 500 bar for the same column length. Practically, a flow rate of 0.05 mL/min was used 1.0 mm ID columns to maintain linear velocity, several flow rates (0.05, 0.075, 0.10, and 0,12 mL/min) was tested, and the flow rate of 0.12 mL/min provided a shorter analysis time with adequate resolution. The total LC run time decreased to 10 min by a gradient elution kept at 25% of B at 1.0 min, linearly

increasing from 25% to 100% of the mobile phase B within 2.5 min, maintained for 2.5 min at 100% B. The initial condition of 25% was returned for 1.0 min followed by equilibration of 3.0 min. The other conditions, such as the mobile phase, are kept the same as the initial conditions. Well-separated chromatographic peaks of the three estrogens are shown in Figure 15.

Due to the busy instrument schedule, the method optimization was transferred to the AB Sciex 5500. The optimized conditions of ionization for the analytes operated by AB Sciex 5500 were established in Table 2. Figure 16 demonstrated the chromatograms of E1, E2, and E3 with a concentration of 0.5 µM neat standard using the same C18 column (ID 1.0 mm). The other settings, such as the mobile phase, are kept the same as AB Sciex 6500. Comparing to AB Sciex 6500, a decrease in intensity by a factor of 255, 122, and 160 were observed for E1, E2, and E3, respectively, operated by AB Sciex 5500, as shown in Figure 15 and Figure 16. This extremely reduced sensitivity and intensity were not expected since AB Sciex 6500 was an improved system

(31)

25 of 5500. Moreover, several research studies were frequently performed by AB Sciex 5500 with diverse mobile phases. Periodic maintenance of the HPLC, specifically pump seals and the solvent lines cleaning, is recommended. Although the three estrogens were well-separated in 4.5 min, E3 suffered from peak splitting problems. This problem can be solved by revisiting the method parameters such as the mobile phase composition, flow rate, or column.

Figure 15 Extracted ion chromatograms of target estrogen from a neat standard of 0.5 µM (sample without plasma) with a C18 column (1.0 x 100 mm), operated by AB Sciex 6500.

Figure 16 Extracted ion chromatograms of target estrogen from a neat standard of 0.5 µM (sample without plasma) with a C18 column (1.0 x 100 mm), operated by AB Sciex 5500.

(32)

26 After numerous injections on a C18 column (ID 1.0 mm) without changing the method, a sharp rise in the retention times of the estrogen was observed (0.4 - 1.1 min), as shown in Figure 17A. These fluctuations in retention times were intolerable when integration of analyte peak would occur, causing timewasting manual integration and inaccurate quantitation. Although Pesek et al. and Li et al. found a small retention time shift using ammonium fluoride as a mobile phase

additive4,76_{. Several possible parameters of retention time shift were investigated, including the}

HPLC pumps, preparing fresh eluent, backpressure, equilibration time, and pH-values. Ammonium fluoride mobile phases reached a pH barely under 7, which slightly hydrolysis of the bonded phase of the column could occur. Waters recommended the working pH range for HSS (C18) columns is 2 to 8, and for BEH (C18, C8) is 1 to 12, temperatures between 20-45 °C, and tolerable pressures up to 1241 bar77_{. In this case, the pH of the eluents was outside the intermediate}

pH-values, around pH 3 to 5, which could be hydrolytic instable. A higher concentration of acetonitrile was used to wash out this hydrolyzed bonded phase of the column, and in the meantime, a Waters C8 column (ID 2.1 mm) was examined to shorter the LC run time with appropriate separation efficiency.

Mobile phase

The initially mobile phases consisted of 0.1 mM ammonium fluoride both in water (A) and in 90/10 (v/v) acetonitrile/water (B), were adjusted to eluent A (water + 0.1 mM NH4F) and eluent B

(ACN + 0.1 mM NH4F) with linearly increasing from 25% to 95% of B. A decrease in intensity and

retention time of the analytes was predictable when C8 column (ID 2.1 mm) was run instant of C18 column (ID 1.0 mm), as shown in Figure 17. The baseline noise had increased significantly for E1 and E3 when 0.1% TEA was used as mobile phase additives, though a gain in intensity for the three estrogens was further detected, shown in Figure 18.

Among the columns, Waters UPLC HSS T3 C18 column (1.0 x 100 mm, particle size 1.8 µm) was found to offer the best separation of the estrogens and generating the most intensive signals. Therefore, this column was selected as the choice of stationary phase for the method. Final gradient elution of a C18 column (ID 1.0 mm) at a flow rate of 0.12 mL/min with eluent A (water + 0.1 mM NH4F) and eluent B (ACN + 0.1 mM NH4F) was as follows: 0-1 min, 25% B; 1-4 min, 95% B;

4-5.5 min, 95% B; and 5.5-6.5 min, 25% B. Column equilibration time was set to 3.5 min. The injection volume was 10 μL, and the total LC run time per sample was 10.0 min.

(33)

27

Figure 17 Extracted ion chromatograms of target estrogen from a neat standard of 0.5 µM (sample without plasma) with (A): C18 column (1.0 x 100 mm) and (B): C8 column (2.1 x 100 mm), operated by AB Sciex 5500.

Figure 18 Extracted ion chromatograms of target estrogen from a neat standard of 0.5 µM (sample without plasma) with C8 column (2.1 x 100 mm) + 0.1% TEA as mobile phase additives, operated by AB Sciex 5500.

(34)

28

Metabolites

After the separation of E1, E2, and E3 was approved, the method was further established, including several estrogen metabolites: 2-hydroxyestradiol (2-OHE2), 4-hydroxyestradiol (4-OHE2), 2-methoxyestradiol (2-ME2), and 4-methoxyestradiol (4-ME2), chemical structures are shown in Figure 19. The optimized MS parameters of the metabolites, i.e., DP, CE, and CXP were shown in Table 2. In negative ESI mode, the significant fragment was the ion with 199 m/z for hydroxy-E2 variants, and 286 m/z for methoxy-E2 metabolites. Typical retention times for the isomers of hydroxyestradiol (2-OHE2, 4-OHE2) were 4.01 min, and 2-ME2/4-ME2 isomers eluted at 4.40, 4.33 min, respectively, as shown in Figure 20. Mixed analytes sample gave extensively overlapped chromatographic peaks among the isomers of the metabolite. Therefore independent quantitation of the isomers 2-OHE2/4-OHE2 and 2-ME2/4-ME2 based on detected masses were not appropriate due to the mass similarity of the fragmentation ions, as shown in Figure 21 and Figure 22. The isomers 2-OHE2/4-OHE2 had the most intense MRM transition of 287.0 > 162.0 m/z, illustrated in Figure 21. Nevertheless, the transition of 287.0 > 199.0 m/z revealed an increase in sensitivity resulted from conjugation, was selected as quantifier ion. The structures for fragment ions of the E2 metabolites have not yet been explored in the literature, thus they were predicted based on the E2 fragmentation pathways73_{. The 2-ME2/4-ME2 isomers gave an}

intense product ion of 301.0 > 286.0 m/z by possibly loss of methyl group, presented in Figure 22. Fragments of 285 m/z and 255 m/z were also formed for 4-ME2, which may be resulted through the loss of hydroxyl group and C2H5OH, respectively. Although fragment of 255 m/z

from 4-ME2 yielded the most intense signal in the mass spectrum, baseline noise was observed in the obtained ion chromatogram. It was noticed that the stability of the working standard of metabolites reduced significantly when it, specially hydroxyestradiol, dissolved in 100%

ultrapure water or high percentage water contained solution, peak consistently disappeared for hydroxyestradiol the next day. The target hydroxyestradiol might degrade or converted into other metabolic products when it dissolved in water. Another possibility is that the adsorption of the hydroxyestradiol to the glass tube, the target compound was stuck to the glass surface. Therefore a mixture of acetonitrile with water (50:50, v/v) was preferred to dissolve and reconstitute estrogens.

Detection of the estrogen included their metabolites in human plasma or serum using a non-derivatization LC-MS/MS method have been infrequently explored. Several studies of estrogens analysis using a positive ESI LC-MS/MS method combined with a derivatization agent have been published, and these reports presented a proper separation of estrogen including their

metabolites. The reported derivatization agents involved dansyl chloride46–48_,

3-Bromomethyl-propyphenazone53_{, pyridine-3-sulfonyl chloride}54_{, methyl piperazine}51_{. From these, dansyl}

chloride is most common, which provided increased method sensitivity. However, the generated fragment ions are unspecific for all estrogens by mass, since they hail from the derivative, thus the metabolites may not be completely separated using the same C18 (ID 1.0 mm) column. A derivative producing analyte-specific precursor and product ions is desirable to increase

specificity, FMP-TS (2-fluoro-1-methylpyridinium-p-toluenesulfonate) was reported as an optional compound-specific derivative for E1 and E250_{. However, a method for the determination of the}

estrogen metabolites using this FMP-TS agent is still lacking.

The use of acetonitrile as a mobile phase eluent to expand selectivity was described in various studies of estrogen4,6,7,9_{. Interestingly, Kruve mentioned that pure acetonitrile acted as a poor}

(35)

29 methanol, water, and water/organic mixtures32_{. The different behavior of acetonitrile and}

methanol can be explained by ion solvation from the ESI droplet. During the ion solvation,

hydrogen bonding was a crucial factor for consideration. In pure methanol, ions were respectably solvated to form hydrogen bonds, whereas acetonitrile behaved like a poor hydrogen bond donor. Consequently, ions with loosely bound solvent shells were left the ESI droplets.

Additionally, Huffman et al. showed that methanol exhibited the most considerable improvement in response compared with water or acetonitrile as an organic modifier for the analysis of

steroids78_{. To study the effect of organic modifier comprehensively in the mobile phase on the}

chromatographic separation and negative ion ESI response is suggested.

Moreover, the selected C18 column with an ID of 1.0 mm provided higher sensitivity, robust performance of column for the analysis of estrogen was restricted, predictable modification in backpressure, and relatively more equilibration time to the initial solvent composition was noticed. Stationary phases with phenyl-hexyl characteristics are not widely investigated for estrogen analysis compared to reversed-phase columns. Several studies have been achieved the chromatographic separation of estrogen using a phenyl-hexyl column52,79–81_{. Column selectivity is}

determined by varieties of interactions, such as hydrophobicity, hydrogen bonding, dipole-dipole ion exchange, and π-π interactions between the stationary phase and analyte. Estradiol is a nonpolar and hydrophobic compound, contained an aromatic A ring. Phenyl-Hexyl column, which is selective for π-π interactions in addition to hydrophobic interactions, could proceed with molecular interactions of the phenyl-hexyl moiety with aromatic groups of estrogen. Petruczynik et al. showed that the phenyl-hexyl column with methanol as an organic modifier obtained the most symmetrical peaks compared with acetonitrile82_{. However, high column efficiency on the}

reversed-phase column using acetonitrile as a mobile phase solvent was found. Besides, methanol and acetonitrile as an organic modifier on the same phenyl-hexyl column showed a difference in retention. The selective π-π interactions between the analyte and the π-ligand in the stationary phase can be suppressed by acetonitrile, which occurs to interfere with π-π

interactions due to competition from the nitrile group of acetonitrile. Therefore the retention is reduced, and it will be significantly determined by the hydrophobic interactions82_.

(36)

30 2-hydroxyestradiol (2-OHE2) 10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H OH A B C D 10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H O H A B C D 10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H O C H₃ A B C D 10 5 1 4 2 3 8 7 9 6 13 14 12 11 17 16 15 O H OH CH₃ 18 H H H O CH₃ A B C D 4-hydroxyestradiol (4-OHE2)

2-methoxyestradiol (2-ME2) 4-methoxyestradiol (4-ME2)

Figure 19 Chemical structures of the estrogen metabolites, including hydroxyestradiol (OHE2), 4-hydroxyestradiol (4-OHE2), 2-methoxyestradiol (2-ME2), and 4-2-methoxyestradiol (4-ME2).

(37)

31

Figure 20 Extracted ion chromatograms of E1, E2, E3, 2-OHE2, 4-OHE2, 2-ME2, and 4-ME2 from a neat standard of 250 nM (sample without plasma) with C18 column (1.0 x 100 mm) + 0.1 mM NH4F as mobile phase additives, operated by AB Sciex 5500.

(38)

32

Figure 21 Negative mode ESI-MS/MS product ion mass spectra of isomers 2-OHE2/4-OHE2 and predicted structures for the fragment ions.

(39)

33

Figure 22 Negative mode ESI-MS/MS product ion mass spectra of isomers 2-ME2/4-ME2 and predicted structures for the fragment ions.