Background: There are considerable demands to accurately measure estradiol (E2) at low concentrations (<20 pg/mL) in postmenopausal women, men, pediatric patients, and patients receiving breast cancer treatment. Most current high-sensitivity LC-MS/MS E2 methods require large sample volumes and involve complex sample preparations with dansyl chloride derivatization. Our study aims to develop a high-sensitivity, underivatized method using micro LC-MS/MS to reliably measure E2 concentrations below 5 pg/mL by the use of low sample volume.
Methods: A total of 290 μL of sample was mixed with internal standard (IS), E2-d4, and extracted with a mixture of hexane/ethyl acetate (90/10) (v/v). After extraction, sample was separated by Eksigent Ekspert™ micro LC 200 system with a flow rate of 35 μL/min in a total run time of 3.5 min and detected by SCIEX QTRAP 6500 mass spectrometer in a negative mode using transitions: 271/145 (quantifier) and 271/143 (qualifier). In this method, it was crucial to use HPLC columns with stability at a pH >10.
Results: The validation study demonstrated broad linear ranges (3.0–820.0 pg/mL) with r2 > 0.999. Total precision was below 15% at all QC levels, and limit of quantification (LOQ) was 3.0 pg/mL. Our method showed good correlation with E2 RIA (r2 = 0.96, bias = −1.0 pg/mL) and modest correlation with E2 Roche Cobas automated immunoassay (r2 = 0.86, bias = 6.0 pg/mL).
Conclusions: In conclusion, we developed and validated a routinely applicable micro LC-MS/MS method without derivatization for E2 in blood samples with an LOQ of 3.0 pg/mL.
Patients under assessment of puberty delays, pubertal growth, gynecomastia, and efficiency of aromatase inhibitor (AI) treatment will benefit from the information presented here. Evidence presented on high-sensitivity LC-MS/MS assay for E2 will allow better characterization of E2 concentration in blood in the low measurement range. Knowledge in the field of LC-MS/MS method development for E2 will be advanced by the information presented.
Estrogens are steroid hormones derived from cholesterol and are primarily secreted by the ovarian follicles or placenta in female, and testes in male. The order of potency of the 3 major naturally occurring estrogens within the male and female bodies are: estradiol (E2),3 estrone, and estriol (1, 2). Estrogens are crucial for the development of female secondary sex characteristics and reproductive functions and are also important for many other non–sex-specific processes, including growth, skeleton maintenance, cardioprotective, and neuroprotective functions (3–10).
E2, as the most potent and the most predominant estrogen in premenopausal female, is commonly measured in clinical laboratories to assess female reproductive functions (9, 11, 12) and to monitor ovulation induction and preparation in in vitro fertilization (13, 14). Compared to the E2 concentrations in premenopausal females, concentrations in children, males, and postmenopausal females are very low, typically <20 pg/mL. Reliable measurements of E2 in those populations can be useful for workups of many pathophysiological conditions, such as precocious puberty, delayed puberty, osteoporosis, gynecomastia, and ambiguous genitalia in newborns with congenital adrenal hyperplasia (4, 15–17). In an evaluation of endogenous E2 concentrations in postmenopausal women and vertebral fractures, 33% of the measurements were <5 pg/mL (16). A plasma E2 ≥10 pg/mL strongly suggests that puberty has begun and a plasma E2 ≥20 pg/mL is clearly indicative of the onset of puberty (18). In patients with estrogen receptor–positive breast cancers, aromatase inhibitors (AIs) are used for treatment to achieve a reduction of estrogen concentrations. Accurate measurements of E2 at low concentrations are important to assess the efficacy of antiestrogen therapies (15, 16). By using LC-MS/MS, a previous study measured plasma E2 in AI-treated breast cancer patients at a median concentration of 4.5 pg/mL, with a range of 0.7–44 pg/mL (19). One challenge most clinical laboratories are facing is to have adequate sensitivity in E2 measurements at low ranges for clinical indications other than assessing female reproductive functions and ovulation induction.
For E2 measurements, indirect RIAs provide relatively good sensitivity and specificity (20, 21), but they are time-consuming and use radioactive materials; therefore, these methods have been largely replaced by automated immunoassays. With higher throughput, faster turnaround time, and minimum sample preparation, automated immunoassays are the most widely used techniques for E2 measurements in clinical and research laboratories. However, automated immunoassays suffer from poor accuracy and precision at low physiological E2 concentrations and are susceptible to interferences due to nonspecific antibody cross-reactivity with estrogen metabolites and other exogenous estrogens (9, 22–24). To overcome these technical limitations of immunoassays, mass spectrometry assays, especially LC-MS/MS assays, have increasingly been used in the recent years. One drawback about most LC-MS/MS E2 assays is that derivatization is required in the sample preparation procedure to achieve sufficient sensitivity (4, 9, 25–27). In a routine clinical laboratory, derivatization can increase the sample preparation time and introduce more method variations. Some attempts have been made to design methods without the need of derivatization, but those methods require an extended extraction protocol and large sample volumes (23, 24, 28), which are also not optimal for routine clinical practice. Our goal in this study was to develop a high-sensitivity, underivatized method using micro LC-MS/MS to reliably measure E2 concentrations in adult men, postmenopausal women, pediatric patients, and patients undergoing antiestrogen therapies.
Materials and Methods
E2, HPLC-grade hexane, and HPLC-grade ethyl acetate were obtained from Sigma-Aldrich. Deuterated internal standard (IS), E2-2,4,16,16-d4 (E2-d4), was purchased from C/D/N Isotopes. E2 and E2-d4 have reported chemical and isotopic purity (>98%) and were used without further purification. HPLC-grade methanol and HPLC-grade acetonitrile were obtained from Fisher Scientific. The MS-grade, E2-depleted serum was supplied by Golden West Biologicals.
Preparation of calibrators and IS and QC materials
Stock solutions of E2 (1 g/L) and IS E2-d4 (1 g/L) were prepared by dissolving 1 mg of each individual compound in 1 mL of methanol. A working solution of E2-d4 was prepared by diluting E2-d4 stock solution with methanol to 2.5 ng/mL and E2 working solutions were prepared by diluting the E2 stock solution with methanol to 100, 200, 1000, 2000, 10 000, and 20 000 pg/mL, respectively. Calibrators at 5, 10, 50, 100, 500, and 1000 pg/mL were then prepared by diluting each working solution with E2-depleted serum by a factor of 20. QC materials were prepared at 4, 50, and 550 pg/mL, respectively, in E2-depleted serum. Calibrator values were verified by ARUP laboratories using their high-sensitivity, derivatized LC-MS/MS method.
A total of 290 μL of calibrators, QCs, or samples were transferred into 16 × 100 mm borosilicate glass tubes followed by 10 μL of IS working solution. Each sample was extracted with 3 mL freshly made extraction solution (90% hexane and 10% ethyl acetate). Samples were vigorously vortex-mixed and incubated at room temperature for 30 min before centrifugation at 865g for 10 min. The organic phase was transferred into a new borosilicate glass tube and evaporated under nitrogen at 37 °C, and the dried residues were reconstituted in 60 μL methanol and water (volume 1:1). Before LC-MS/MS injection, all samples were subjected to high-speed centrifugation (17 000g) for 10 min to remove any debris and ensure sufficient sample cleanup. Each LC-MS/MS analysis required 5 μL of the prepared sample.
Micro LC-MS/MS instrumentation and conditions
A QTRAP 6500 triple-quadrupole mass spectrometer (SCIEX) equipped with an IonDrive™ Turbo V source was operated in negative electrospray ionization (ESI) mode. An Eksigent Ekspert™ micro LC 200 system (SCIEX) with a CTC PAL autosampler provided chromatographic separation using a reversed-phase C18 analytical column (YMC Triart, 0.5 mm × 50 mm, 3 μm particle size) protected by a Triart C18 guard column (0.5 mm × 5 mm) of the same packing material (YMC America). The mobile phase consisted of water with 0.05% ammonium hydroxide (phase A) and methanol with 0.05% ammonium hydroxide (phase B). The micro LC system was programmed to deliver a flow rate of 35 μL/min starting with 10% mobile phase B for 0.5 min, linear gradient to 98% mobile phase B between 0.5 and 2.5 min, held steady at 98% mobile phase B for 0.7 min, and followed by reequilibration to initial conditions. The full liquid chromatography (LC) run was 3.5 min. The autosampler injection syringe and valve were washed 10 times after each injection with acetonitrile and water. The QTRAP 6500 instrument settings were optimized for E2 signal. The ESI source was operated with ion spray voltage at −4500 V at a temperature of 500 °C; declustering potential was set at −160 V and entrance potential was −10 V. A high-purity nitrogen generator (Peak Scientific) provided nitrogen gas with the following settings: nebulizer gas, 80 psi; heater gas, 80 psi; curtain gas, 20 psi; and collision gas, 11 psi. Two mass transitions were monitored for E2 [271/145 (quantifier) and 271/143 (qualifier)] and 1 transition was monitored for IS (275/147). Other instrument settings optimized for each transition are shown in Table 1. All mass spectrometry data were acquired and processed with the Analyst 1.6.2 software, and the calibration curve was constructed by plotting peak area ratio between E2 and E2-d4 vs the concentration of E2 in calibrators using the MultiQuant™ 3.0 software (SCIEX).
Analytical performance validation
Precision, analytical measurement range, and limit of quantification studies.
Precision studies of this method were performed using in-house prepared QC materials. Assay repeatability was determined by consecutively measuring 3 levels of QC materials 20 times (5 replicates of each QC sample were prepared and 4 injections were made from each replicate). Carryover was assessed during the repeatability test. Reproducibility was determined by daily testing of the QC materials in duplicates over 10 days. The analytical measurement range was determined by diluting a patient sample with high E2 concentration using a blank serum by a factor of 2, 10, 20, 100, and 200, respectively. The measured results were plotted against expected results calculated by dilution factors. The limit of quantification (LOQ) was assessed by running a series of prepared pooled serum samples with low E2 concentrations over a period of 10 days, and the % CV at 20% was used as the cutoff to define LOQ.
Method comparison and recovery studies.
Samples used for method comparison were leftover samples retrieved from the clinical chemistry laboratory. E2 measurements by LC-MS/MS were compared with the Cobas 8000 E2 II automated immunoassay (Roche Diagnostics) using 42 plasma samples collected in lithium heparin tubes (BD Diagnostics) and with a laboratory-developed indirect RIA using 38 serum samples collected in plain serum tubes (BD Diagnostics). Results were analyzed using Passing–Bablok regression plot and Bland–Altman bias plot to determine the correlation and overall bias between methods. Data analysis was performed using Analyse-iT version 2.21 (Analyse-iT Software). Graphs were created using the GraphPad Prism software (GraphPad Software). To assess assay accuracy, recovery study was performed by spiking known amounts of E2 standards into blank matrix and calculating the recovery percentage (measured concentration/expected concentration × 100%).
Samples with serial concentrations of hemoglobin, bilirubin, or intralipids were prepared as described in a previous study (29) and were used to assess the effect of high concentrations of those interferents on this LC-MS/MS assay. All samples were processed in duplicates, and the serum indices were measured on a Cobas 8000 system.
Specimen-type compatibility and stability.
Samples from 5 volunteers were collected in both serum tubes and lithium heparin plasma tubes to compare the specimen-type compatibility. Serum aliquots containing low E2 concentrations were stored at 4 °C or −20 °C for 1, 5, and 7 days to assess the sample storage stability. To test the postextraction stability at 4 °C, aliquots of prepared samples ready for injections were stored up to 48 h before analysis. This study was performed under a quality improvement protocol, which did not require our institutional review board review or approval.
The chromatogram of E2 and IS extracted from serum are shown in Fig. 1. E2 concentrations were determined using mass transition of m/z 271–145. The total imprecision of this method showed CVs <11.6% for all levels of QCs (Table 2). The LOQ was determined to be 3.0 pg/mL corresponding to a CV of 20% (Fig. 2A) and an S/N of 27. The assay was linear from 3.0 pg/mL to 822.0 pg/mL with r2 > 0.999 (Fig. 2B). No carryover was detected up to E2 concentration of 577.0 pg/mL (see Supplemental Data 3 in the Data Supplement that accompanies the online version of this article at http://www.jalm.org/content/vol1/issue1).
Potential interferences from bilirubin, hemolysis, and lipemia were studied, and results are provided in the Supplemental Data. A significant negative bias was observed for E2 measurement when H-index was >300 or I-index was >5. No significant interference was observed with L-index up to 710 (Supplemental Data 1). Higher lipid concentrations will interfere with the liquid–liquid extraction process.
We compared E2 as measured by LC-MS/MS with automated immunoassay and indirect RIA. Indirect RIA showed good correlation with LC-MS/MS method in all patient samples (r2 = 0.96, n = 38) (data not shown). Comparison of indirect RIA vs LC-MS/MS in a range of <100 pg/mL (Fig. 3A) showed the equation from Passing–Bablok regression was: y = 0.92x + 1.18, n = 29, range = 3.0–81.0 pg/mL, r2 = 0.96, with an overall bias of −1.0 pg/mL (Bland–Altman analysis, Fig. 3B). The automated Roche Cobas E2 immunoassay demonstrated modest overall correlation with the LC-MS/MS assay (r2 = 0.93) in a measuring range from 3.0 to 202.0 pg/mL (data not shown). Analysis performed on the lower measuring region (<100 pg/mL) showed the automated immunoassay (Fig. 3C) had a slight positive proportional bias of 10% (Passing–Bablok regression: y = 1.1x + 3.6, n = 40, range 3.0–60.5 pg/mL, r2 = 0.86) and an overall bias of +6.0 pg/mL (Bland–Altman analysis, Fig. 3D). In the recovery study, all E2 samples added had a recovery ranging from 95% to 114% (Supplemental Data 4).
We observed no significant differences between serum and lithium heparin plasma for E2 measurement using this LC-MS/MS assay with the percent difference ranging from −4.7% to 2.1%. E2 samples were stable at 4 °C or −20 °C for at least 7 days, with E2 values differing from day 0 by 9.6% at 4 °C and 5.7% at −20 °C. Postextraction samples were stable at 4 °C for at least 48 h, with E2 values differing by 9.4% (data not shown). The liquid–liquid extraction recovery was approximately 70% in this assay, owing to a 20% volume discard in the supernatant transferring step after extraction, to avoid disturbing the lipid level and to ensure better matrix cleanup.
Significant ion suppression of E2 was observed in patient samples. Although the addition of IS at the beginning of the assay could normalize the ion suppression effect, the assay sensitivity was still compromised. Calibrators and QCs were prepared from charcoal-stripped serum, which had a cleaner matrix than patient samples and less ion suppression effect. Ammonium hydroxide (0.01%) was initially used in this method, and no ion suppression was observed for calibrators and QCs. Under the same conditions, some patient samples could not be fully ionized. To reduce the ion suppression effect, different concentrations of ammonium hydroxide used in the mobile phases (0.01%, 0.02%, 0.05%, and 0.10%) were tested, and patient samples with the most severe ion suppression effect were analyzed using the different mobile phases. In a patient sample that showed severe ion suppression with 0.01% ammonium hydroxide, no clear E2 peak was detected; with increasing ammonium hydroxide added, the signal significantly increased and peaked at 0.05%, resolving the ion suppression effect (Supplemental Data 2). Thus, mobile phase pH at 10.3 with 0.05% ammonium hydroxide was used in this method for all samples to ensure sufficient ionization.
Widely used automated immunoassays in clinical laboratories for E2 measurements may reach an LOQ of 20 pg/mL and may be adequate for use in premenopausal women and women undergoing fertility treatments. For other populations such as postmenopausal women, the pediatric population, men, and women receiving AI treatment, E2 concentrations are typically below the LOQ of automated immunoassays, making such assays suboptimal for use in these populations. Another significant challenge of automated immunoassays is the antibodies used in these tests may cross-react with other estrogens and their metabolites. The accuracy, specificity, and reproducibility of the measurement of E2 are critical for diagnosis and monitoring in patients requiring low E2 concentration values. The lack of analytical accuracy in the low measuring range not only affects current clinical care but also is holding back a better understanding of emerging science around the role of E2 beyond female reproduction and sexual maturity, as extensively discussed in a recent position statement from the Endocrine Society (15, 16). LC-MS/MS assays, with better low-end sensitivity and selectivity, can overcome most limitations of automated immunoassays and thus were recommended as the assay of choice for E2 measurement at the low ranges.
However, it is challenging to develop an LC-MS/MS assay for measuring E2 at very low concentrations. One of the most difficult challenges with using MS for steroid analysis is the difficulty in eliciting sufficient analyte ionization because of its poor polarity, especially for E2. Derivatization of E2 has been widely accepted as a way to improve ionization efficiency, making it more sensitive in LC-MS/MS–based detection. Most reported methods of derivatization use dansyl chloride and found improved signal intensity by 1–2 orders of magnitude when compared to an underivatized method under positive polarity mode (23–28). However, the derivatization step can add complexity in the sample preparation and may decrease the robustness of the overall method, which is not optimal for a routine clinical laboratory application.
We have developed and validated a micro LC-MS/MS assay for E2 that does not require derivatization to achieve comparable sensitivity (3.0 pg/mL) to some MS assays that require derivatization. To achieve this, a miniature-scale LC instrument was used. There are two broad classes of miniature-scale LC systems: nano LC (flow rate <1 μL/min) and micro LC (flow rate of 1–50 μL/min). A general benefit of using miniature-scale LC systems in LC-MS/MS assays is the relatively higher sensitivity and lower background noise compared to conventional-scale LC. Nano LC is often required in proteomic studies because of its high sensitivity and drastically reduced sample volume consumption. However, the lower throughput of nano LC with cycle times of 40 min to 1 h per sample limits its application in routine clinical laboratories. In contrast, micro LC can reach an optimal balance of sensitivity, throughput, and sample consumption, making it more suitable for clinical laboratories. In this method, we used a dedicated micro LC system, Eksigent Ekspert micro LC 200, which can operate at pressures up to 10 000 psi with a flow rate of 1–50 μL/min. To achieve the required sensitivity, we coupled the Eskigent micro LC 200 to a QTRAP 6500 triple quadrupole system and operated under negative ionization mode. Previous studies have shown that assays without derivatization exhibit better E2 ionization and lower background noise when ESI operated in negative mode vs positive mode. Positive atmospheric pressure chemical ionization and atmospheric pressure photoionization produced a good response for E2, but with higher background noise. The LOQs of E2 assays using positive atmospheric pressure chemical ionization and atmospheric pressure photoionization were comparable with LOQs of assays using negative ESI (30–32). In this assay, negative ESI was used and in an effort to optimize the ESI response, mobile phase modification was needed. Previous literature showed that adding ammonium hydroxide in the mobile phase could increase the [M-H]− ion intensity (30, 31).
Ionization suppression, a major concern in the LC-MS/MS method, is a compound-dependent effect and is more severe when operating in ESI, particularly for steroid estrogens (31). We observed significant ion suppression in some extracted patient samples, whereas the effect was not as severe in calibrators and QCs, which were prepared using a cleaner matrix. The main reason of ionization suppression is the competition between matrix components and the analytes of interest to be ionized in the ESI source (32, 33). To reduce the ion suppression effect in this assay, the key was to use high pH conditions. We found that by adding 0.05% ammonium hydroxide in the mobile phase to adjust the eluent pH to be above 10 was necessary to retain sufficient signal intensity to achieve the desired analytical sensitivity in patient serum samples. A high pH can increase ionization efficiency and reduce ion suppression by creating a better condition for forming anions in the ESI source and by facilitating the deprotonation of E2, which has a pKa of 10.7. However, care must be taken in choosing an LC column that can operate robustly at the high pH range. The commercial availability of a micro LC column is not as extensive as the conventional columns. However, we found the YMC Triart C18 column had excellent stability operating in high pH conditions and an acceptable usage life in our method. In addition, sufficient removal of extraneous matrix components could also help to reduce ion suppression. In this assay, the procedure was optimized to ensure efficient cleanup in the liquid–liquid extraction step, and high-speed centrifugation was performed before sample injection.
A reference interval study was not performed for this assay, since the comparison study showed an excellent correlation: [RIA E2] (pg/mL) = 0.92 [LC-MS E2] (pg/mL) + 1.18, r2= 0.96, mean bias = −1.0 pg/mL (n = 29). In consultation with our endocrinologists, we determined that both assays are analytically and clinically equivalent, and the existing reference intervals established by the classic RIA method developed in our institution and published by our endocrinologists in the literature (34–38) are transferrable to the LC-MS/MS assay. The bias of this assay at the LOQ level was also not assessed, since there are currently no low E2 reference materials available, and bias study by spiking standards and serially diluting up to 200-fold could result in large variation at the low end. Therefore, we used the functional sensitivity concept, 20% imprecision, as the goal to define the assay LOQ (39).
In summary, we developed and validated an underivatized micro LC-MS/MS method, which is analytically robust and sensitive enough to measure E2 in men, postmenopausal women, pediatric patients, and women receiving AI treatment.
We would like to thank SCIEX Diagnostics for providing the technical support in the development of this high-sensitivity E2 assay.
↵3 Nonstandard abbreviations:
- aromatase inhibitor
- internal standard
- electrospray ionization
- liquid chromatography
- limit of quantification.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form.
Employment or Leadership: None declared.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: K.-T.J. Yeo, SCIEX Diagnostics.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, and final approval of manuscript.
- Received April 26, 2016.
- Accepted May 2, 2016.
- © 2016 by American Association for Clinical Chemistry