madman
Super Moderator
ABSTRACT
Monitoring estrogen levels, especially estradiol (E2), is amongst others important for determining menopausal status and guidance of breast cancer treatment. We validated a serum E2 and estrone (E1) liquid chromatography-tandem mass spectrometry assay (LC-MS/MS) suitable for quantitation in human subjects. In addition, we compared our method with an E2 immunoassay (IA) and established preliminary reference values. Validation parameters were within the predetermined acceptance criteria. Assay linearity ranges were 4–1500 pmol/L for E1 and 4–2500 pmol/L for E2. Imprecision ranged from 7.4 to 9.6%. The lower limit of quantitation for E2 (8.0 pmol/L) was 11.4 times lower than the IA. The method comparison revealed differences in E2 quantitation up to 155% between both methods. The method allowed quantitation of E1 in all healthy volunteers, while E2 could not be detected in 95% versus 40% of the post-menopausal women using IA and LC-MS/MS, respectively. Male, pre-, peri- and postmenopausal female reference values were estimated. An LC-MS/MS-based method combining E1 and E2 analysis was validated with superior E2 analytical sensitivity when compared to the IA.
1. Introduction
In women, estrogens are important for the development and upkeep of the reproductive system, secondary sex characteristics, menstrual cycle, and pregnancy. The two most prevalent estrogens are estrone (E1) and estradiol (E2), with E2 being the most biologically active [1]. Laboratories quantitate circulating E2 levels to aid in the diagnosis of amongst others: female fertility disorders, ovarian hyperstimulation in the context of in-vitro fertilization, determination of menopausal status, and gynecomastia in males [2]. For breast cancer patients, assessment of ovary function and menopausal status is essential to guide systematic hormonal treatment. In this context, E2 quantitation is used to confirm proper suppression of ovary function in pre-and perimenopausal patients to assure treatment efficacy of aromatase inhibitors [3,4]. E1 is not commonly measured in laboratories, despite being the most abundant circulating estrogen in post-menopausal women [1,5,6].
Physiological levels of estrogens, especially in postmenopausal women, are low (E1, <148 pmol/L; E2, <77 pmol/L), and highly sensitive assays are required to allow quantitation [5,6]. For E2 analysis, laboratories mostly rely on cost-effective and high throughput immunoassays (IA) offering considerable sensitivity. However, in most postmenopausal women, specifically for breast cancer patients receiving aromatase inhibitors, circulating E2 levels measured with an IA are nonquantitable. Furthermore, IA is known to lack specificity in low concentrations due to cross-reactivity potentially resulting in unreliable quantitation of E2 [7–10]. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) based methods are considered best practice for steroid analysis and can quantitate estrogens in low pico-molar concentrations with increased specificity compared to immunoassays [11]. Although LC-MS/MS estradiol methods offer obvious advantages, they are labor-intensive and require additional expertise to perform correct analysis [12,13].
*In this study, we present the validation of an LC-MS/MS-based method for the simultaneous quantitation of E1 and E2. Furthermore, reference values were estimated and E2 results were compared with the in-house E2 IA
2.5. LC-MS/MS assay validation
(Pre-) analytical method validation was performed and included imprecision, the lower limit of quantitation, trueness, sample stability, linearity, matrix effect and extraction recovery, carry-over, and interference. Imprecision was determined by analysis of three serum pools in quadruplicate for ten consecutive runs on separate days containing concentrations distributed over the measuring range. The lower limit of quantitation was determined by measuring three serum pools of spiked double charcoal-stripped fetal bovine serum (Approximately 6, 8 and 10 pmol/L for both E1 and E2) in duplicate for six consecutive runs on separate days with analyte peaks showing an S/N > 10. The mean of the pool containing the lowest concentration and meeting the predefined criteria was accepted as LLOQ. Criteria for imprecision and the lower limit of quantitation were a total coefficient of variation (CV) below 10% and 20%, respectively. For E2, assay trueness was determined by measuring medium and high serum reference material in triplicate and low serum reference material in duplicate for four consecutive runs on separate days. A bias below 5% was considered acceptable. For E1, calibration was performed in triplicate with a European reference standard. Sample stability was evaluated for three serum pools at − 20 ◦C (2 months), 4 ◦C (2 weeks) and 20 ◦C (1 week). Linearity was evaluated at 7 levels across the measuring range. Polynomial regression was performed and linearity fit was tested in EP Evaluator (Version 12.2). Matrix effects and extraction recovery were determined by pre-and postextraction standard addition. Carry-over effects were investigated in three blank samples after injection of three consecutive high calibrator samples (E1, 1500 pmol/L; E2, 2500 pmol/L). Interference was tested by analyzing three serum pools spiked with 17-hydroxyprogesterone (5 nmol/L), 17α-estradiol (0.5 nmol/L), 17α-ethynylestradiol (0.5 nmol/L), anastrozole (250 nmol/L), androstenedione (5 nmol/L), cortisol (500 nmol/L), dehydroepiandrosterone (50 nmol/L), dihydrotestosterone (5 nmol/L), epitestosterone (50 nmol/L), exemestane (100 nmol/L), letrozole (100 nmol/L), prednisolone (1 nmol/L), prednisone (1000 nmol/L), progesterone (5 nmol/L), tamoxifen (1000 nmol/ L), testosterone (50 nmol/L), hemoglobin (1 mmol/L), bilirubin (50 µg/ mL) and intralipid (1%). A recovery within ± 10% was considered acceptable for evaluating sample stability and interference. Validation was in accordance with previously published guidelines [17].
2.6. Estimation of reference values
To estimate the reference values, serum samples from healthy males (n = 124), healthy females aged 18–40 years (n = 121), healthy females aged 41–60 years (n = 128), and healthy females aged ≥61 years (n = 122) were separately studied. The sample size was based on recommendations made by the Clinical and Laboratory Standards Institute.
4. Discussion
Here, we successfully validated an LC-MS/MS assay for measurement of E1 and E2 allowing over 11 times more sensitive E2 quantitation than the in-house routinely applied IA. Furthermore, E1 concentrations were quantifiable in all male and female samples. To investigate whether our method can quantitate estrogens in healthy volunteers, we determined preliminary reference values for males aged at least 18 years, females aged 18–40 years, females aged between 41 and 60 years, and females aged at least 61 years. To this end, in-house biobank samples were used in the absence of information on the female subjects’ menopausal status, menstrual cycle period, or use of birth control pills. Therefore, no definite reference values for both estrogens in females in regard to menstrual cycle period and menopausal status could be determined. We separated female samples by age to assess the effect of menopause on the circulating concentrations of E1 and E2. Although the onset of menopause is known to be influenced by race, ethnicity, and lifestyle factors, the overall median age at menopause is between 50 and 52 years with the vast majority of women being premenopausal before the age of 45 and most being postmenopausal after the age of 55 years [19–21]. To increase the chance that the large majority of the pre-and post-menopausal females were indeed in this menopausal stage, broad age cut-offs at 40 and 60 years for peri-menopausal female subjects were selected. Significant differences in estrogen levels between premenopausal and postmenopausal as defined by our age classification were observed for both E1 and E2.
In literature, well-established estrogen reference values using LC-MS/MS methodology is relatively scarce. Four studies have previously described reference values for E1 and E2 [5,6,10,22], while another study recently published reference intervals only for E2 [23]. For both estrogens, considerable variations in reference values are observed. This could be explained by 1) differences in population selection and characterization, 2) poor standardization between methods, 3) selection of direct or derivatization procedures and 4) various statistical approaches in determination of the reference range (i.e. 95%CI, IQR or whole range) [24,25].
Additionally, we investigated the differences in E2 quantitation by our in-house routine IA and the newly developed LC-MS/MS method in healthy volunteers. The first observation was that the LC-MS/MS was able to quantitate E2 levels in a significantly larger number of samples in all groups (p-values below 0.001). In the second analysis, relative differences up to 155% were detected, especially in the lower concentration ranges found in males and females aged above 60 years old. As our E2 LC-MS/MS method has superior specificity over the IA and was standardized against certified reference material, these findings suggest unreliable quantitation of E2 by the IA in lower concentration ranges. This could potentially be an issue for breast cancer patients in which ovarian function assessment is necessary to determine whether aromatase inhibitor treatment is appropriate [3,4].
In summary, we have successfully validated a serum estrogen LC-MS/MS method that was considered suitable for application in human subjects. Significant discrepancies were demonstrated in low circulating E2 levels with the in-house IA. Furthermore, using biobank samples, we estimated the reference values for pre-, peri- and postmenopausal women and in males. While these results clearly show the technical benefit of using LC-MS/MS-based estrogen analysis instead of IA technology, future studies are necessary to determine its potential in breast cancer patients.
Monitoring estrogen levels, especially estradiol (E2), is amongst others important for determining menopausal status and guidance of breast cancer treatment. We validated a serum E2 and estrone (E1) liquid chromatography-tandem mass spectrometry assay (LC-MS/MS) suitable for quantitation in human subjects. In addition, we compared our method with an E2 immunoassay (IA) and established preliminary reference values. Validation parameters were within the predetermined acceptance criteria. Assay linearity ranges were 4–1500 pmol/L for E1 and 4–2500 pmol/L for E2. Imprecision ranged from 7.4 to 9.6%. The lower limit of quantitation for E2 (8.0 pmol/L) was 11.4 times lower than the IA. The method comparison revealed differences in E2 quantitation up to 155% between both methods. The method allowed quantitation of E1 in all healthy volunteers, while E2 could not be detected in 95% versus 40% of the post-menopausal women using IA and LC-MS/MS, respectively. Male, pre-, peri- and postmenopausal female reference values were estimated. An LC-MS/MS-based method combining E1 and E2 analysis was validated with superior E2 analytical sensitivity when compared to the IA.
1. Introduction
In women, estrogens are important for the development and upkeep of the reproductive system, secondary sex characteristics, menstrual cycle, and pregnancy. The two most prevalent estrogens are estrone (E1) and estradiol (E2), with E2 being the most biologically active [1]. Laboratories quantitate circulating E2 levels to aid in the diagnosis of amongst others: female fertility disorders, ovarian hyperstimulation in the context of in-vitro fertilization, determination of menopausal status, and gynecomastia in males [2]. For breast cancer patients, assessment of ovary function and menopausal status is essential to guide systematic hormonal treatment. In this context, E2 quantitation is used to confirm proper suppression of ovary function in pre-and perimenopausal patients to assure treatment efficacy of aromatase inhibitors [3,4]. E1 is not commonly measured in laboratories, despite being the most abundant circulating estrogen in post-menopausal women [1,5,6].
Physiological levels of estrogens, especially in postmenopausal women, are low (E1, <148 pmol/L; E2, <77 pmol/L), and highly sensitive assays are required to allow quantitation [5,6]. For E2 analysis, laboratories mostly rely on cost-effective and high throughput immunoassays (IA) offering considerable sensitivity. However, in most postmenopausal women, specifically for breast cancer patients receiving aromatase inhibitors, circulating E2 levels measured with an IA are nonquantitable. Furthermore, IA is known to lack specificity in low concentrations due to cross-reactivity potentially resulting in unreliable quantitation of E2 [7–10]. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) based methods are considered best practice for steroid analysis and can quantitate estrogens in low pico-molar concentrations with increased specificity compared to immunoassays [11]. Although LC-MS/MS estradiol methods offer obvious advantages, they are labor-intensive and require additional expertise to perform correct analysis [12,13].
*In this study, we present the validation of an LC-MS/MS-based method for the simultaneous quantitation of E1 and E2. Furthermore, reference values were estimated and E2 results were compared with the in-house E2 IA
2.5. LC-MS/MS assay validation
(Pre-) analytical method validation was performed and included imprecision, the lower limit of quantitation, trueness, sample stability, linearity, matrix effect and extraction recovery, carry-over, and interference. Imprecision was determined by analysis of three serum pools in quadruplicate for ten consecutive runs on separate days containing concentrations distributed over the measuring range. The lower limit of quantitation was determined by measuring three serum pools of spiked double charcoal-stripped fetal bovine serum (Approximately 6, 8 and 10 pmol/L for both E1 and E2) in duplicate for six consecutive runs on separate days with analyte peaks showing an S/N > 10. The mean of the pool containing the lowest concentration and meeting the predefined criteria was accepted as LLOQ. Criteria for imprecision and the lower limit of quantitation were a total coefficient of variation (CV) below 10% and 20%, respectively. For E2, assay trueness was determined by measuring medium and high serum reference material in triplicate and low serum reference material in duplicate for four consecutive runs on separate days. A bias below 5% was considered acceptable. For E1, calibration was performed in triplicate with a European reference standard. Sample stability was evaluated for three serum pools at − 20 ◦C (2 months), 4 ◦C (2 weeks) and 20 ◦C (1 week). Linearity was evaluated at 7 levels across the measuring range. Polynomial regression was performed and linearity fit was tested in EP Evaluator (Version 12.2). Matrix effects and extraction recovery were determined by pre-and postextraction standard addition. Carry-over effects were investigated in three blank samples after injection of three consecutive high calibrator samples (E1, 1500 pmol/L; E2, 2500 pmol/L). Interference was tested by analyzing three serum pools spiked with 17-hydroxyprogesterone (5 nmol/L), 17α-estradiol (0.5 nmol/L), 17α-ethynylestradiol (0.5 nmol/L), anastrozole (250 nmol/L), androstenedione (5 nmol/L), cortisol (500 nmol/L), dehydroepiandrosterone (50 nmol/L), dihydrotestosterone (5 nmol/L), epitestosterone (50 nmol/L), exemestane (100 nmol/L), letrozole (100 nmol/L), prednisolone (1 nmol/L), prednisone (1000 nmol/L), progesterone (5 nmol/L), tamoxifen (1000 nmol/ L), testosterone (50 nmol/L), hemoglobin (1 mmol/L), bilirubin (50 µg/ mL) and intralipid (1%). A recovery within ± 10% was considered acceptable for evaluating sample stability and interference. Validation was in accordance with previously published guidelines [17].
2.6. Estimation of reference values
To estimate the reference values, serum samples from healthy males (n = 124), healthy females aged 18–40 years (n = 121), healthy females aged 41–60 years (n = 128), and healthy females aged ≥61 years (n = 122) were separately studied. The sample size was based on recommendations made by the Clinical and Laboratory Standards Institute.
4. Discussion
Here, we successfully validated an LC-MS/MS assay for measurement of E1 and E2 allowing over 11 times more sensitive E2 quantitation than the in-house routinely applied IA. Furthermore, E1 concentrations were quantifiable in all male and female samples. To investigate whether our method can quantitate estrogens in healthy volunteers, we determined preliminary reference values for males aged at least 18 years, females aged 18–40 years, females aged between 41 and 60 years, and females aged at least 61 years. To this end, in-house biobank samples were used in the absence of information on the female subjects’ menopausal status, menstrual cycle period, or use of birth control pills. Therefore, no definite reference values for both estrogens in females in regard to menstrual cycle period and menopausal status could be determined. We separated female samples by age to assess the effect of menopause on the circulating concentrations of E1 and E2. Although the onset of menopause is known to be influenced by race, ethnicity, and lifestyle factors, the overall median age at menopause is between 50 and 52 years with the vast majority of women being premenopausal before the age of 45 and most being postmenopausal after the age of 55 years [19–21]. To increase the chance that the large majority of the pre-and post-menopausal females were indeed in this menopausal stage, broad age cut-offs at 40 and 60 years for peri-menopausal female subjects were selected. Significant differences in estrogen levels between premenopausal and postmenopausal as defined by our age classification were observed for both E1 and E2.
In literature, well-established estrogen reference values using LC-MS/MS methodology is relatively scarce. Four studies have previously described reference values for E1 and E2 [5,6,10,22], while another study recently published reference intervals only for E2 [23]. For both estrogens, considerable variations in reference values are observed. This could be explained by 1) differences in population selection and characterization, 2) poor standardization between methods, 3) selection of direct or derivatization procedures and 4) various statistical approaches in determination of the reference range (i.e. 95%CI, IQR or whole range) [24,25].
Additionally, we investigated the differences in E2 quantitation by our in-house routine IA and the newly developed LC-MS/MS method in healthy volunteers. The first observation was that the LC-MS/MS was able to quantitate E2 levels in a significantly larger number of samples in all groups (p-values below 0.001). In the second analysis, relative differences up to 155% were detected, especially in the lower concentration ranges found in males and females aged above 60 years old. As our E2 LC-MS/MS method has superior specificity over the IA and was standardized against certified reference material, these findings suggest unreliable quantitation of E2 by the IA in lower concentration ranges. This could potentially be an issue for breast cancer patients in which ovarian function assessment is necessary to determine whether aromatase inhibitor treatment is appropriate [3,4].
In summary, we have successfully validated a serum estrogen LC-MS/MS method that was considered suitable for application in human subjects. Significant discrepancies were demonstrated in low circulating E2 levels with the in-house IA. Furthermore, using biobank samples, we estimated the reference values for pre-, peri- and postmenopausal women and in males. While these results clearly show the technical benefit of using LC-MS/MS-based estrogen analysis instead of IA technology, future studies are necessary to determine its potential in breast cancer patients.
