| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Reproductive Endocrine Unit, Department of Medicine, Massachusetts General Hospital, Boston MA 02114
aauthor for correspondence: fax 617-726-9330, e-mail psluss{at}partners.org
The direct measurement of serum estradiol offers advantages over the more traditional sample extraction and immunoassay methods with respect to technical ease of the assay and turnaround time. However, direct assays have not been demonstrated sufficiently sensitive for some important clinical applications, including the measurement of circulating estradiol in women during the early follicular phase of their menstrual cycle, in postmenopausal women, and in men. We report here a rapid sample extraction method developed to improve the detection limit of automated estradiol assays and thus enhance the potential usefulness of these rapid and nonisotopic systems for measuring low concentrations of serum estradiol.
Serum estradiol concentrations have long been used in conjunction with ultrasonography and other clinical indices to monitor ovarian stimulation by exogenous gonadotropin therapy during in vitro fertilization (IVF) procedures (1)(2)(3). With the recent use of gonadotropin-releasing hormone agonists, serum estradiol measurement has also been used to assess the adequacy of ovarian suppression before exogenous gonadotropin stimulation. In addition, it has become increasingly important to measure the low initial concentrations of serum estradiol (<150 ng/L) associated with the earlier stages of ovarian follicular stimulation (4)(5). Clinically important measurements of low estradiol concentrations are also required in assessing ovarian activity in perimenopausal women or women experiencing premature ovarian failure and in clinical investigations of the hypothalamic-pituitary-gonadal axis in men (6)(7)(8).
Methods for the direct measurement of serum estradiol are replacing the more traditional RIA methods that use preassay sample extraction followed by RIA with antecedent column chromatography. Direct methods are advantageous because they are rapid and less technically demanding. However, direct assays are often inaccurate, particularly for the measurement of relatively low concentrations of serum estradiol (9)(10)(11). The presence of sex-hormone-binding globulin in serum and the difficulty of obtaining adequate buffers have been associated with both interference and matrix effects, which further contribute to the inaccuracy of direct RIAs of estradiol (12)(13). Such effects probably underlie much of the high interinstitutional variability that has been reported with serum estradiol measurements (14). Recently, nonisotopic methods for the accurate direct measurement of serum estradiol have been developed. The Abbott IMx system used here is one such method that is completely automated and allows measurement of serum estradiol in <1 h.
The estradiol for these studies was purchased from Sigma Chemical Co. Tritiated estradiol (2,4,6,7-3H-estradiol) was purchased from Amersham. IMx reagents, controls, and calibrators were purchased from Abbott Laboratories. Hexane (99.9% purity; HPLC grade) was purchased from Baxter Chemical Co. (Miami, FL) and ethyl acetate (99.8% purity; HPLC grade) was purchased from Aldrich Chemical Co. Blood samples were obtained from IVF patients undergoing hormonal therapy monitoring or healthy men participating in clinical investigations at the Massachusetts General Hospital. Sampling of the research participants was performed after obtaining informed consent and was for the purpose of evaluating endocrine endpoints as described in research protocols reviewed and approved by the Human Studies Subcommittee (an Institutional Review Board) of the Massachusetts General Hospital. The serum samples studied were selected based on estradiol concentrations measured by an in-house RIA on extracted serum (15).
Direct measurement of serum estradiol in IVF patients was found to meet or exceed the precision and accuracy reported for this automated procedure when used to measure estradiol in normally cycling women (data not shown). In our laboratory we evaluate the limit of clinical utility of an assay by assessment of the assays limit of linearity on serial dilution of patient samples. On the basis of quadruplicate measurements, the limit of linearity for the measurement of estradiol in serum samples obtained from IVF patients was similar to the detection limit reported previously for the serum of normally cycling women (50 ng/L).
To improve the detection of low serum estradiol concentrations (<50 ng/L), a method is needed to improve the performance of the current estradiol assays in the low range of the assay. Using the IMx estradiol assay as a model detection system, we evaluated whether sample extraction could improve the detection of low concentrations of estradiol. Estradiol was extracted from serum with hexane–ethyl acetate (3:2 by volume). We used 10 mL of organic solvent per 1 mL of serum. After addition of hexane–ethyl acetate, samples were vortex-mixed for 30 s and allowed to separate by standing for several minutes. The organic layer, containing estradiol, was recovered by freezing the aqueous portion of the sample in a methanol–dry ice bath. The organic extract was dried completely at 40 °C in a vacuum centrifuge (Savant SpeedVac; Savant Instruments Inc.) and reconstituted in steroid-free human serum (Abbott Diagnostics or Diagnostic Products Co.).
To monitor recovery, we added 3H-estradiol (15 000 cpm) to serum samples and allowed them to equilibrate for 1 h at room temperature before extraction. An aliquot of the reconstituted sample was combined with liquid scintillation fluid (Beckman), and the amount of radioactivity present relative to the amount added before extraction was used to calculate the recovery of estradiol. With two extractions, 97.8% of the radiolabeled estradiol added to the serum was recovered: extract 1 contained a mean (SD) of 92.4 (6.3)% of the added counts, and extract 2 contained 5.4 (1.3)% (n = 6). The remainder probably represents procedural losses such as incomplete liquid transfers and glass losses. Less than 3% of the radiolabeled estradiol was unextractable and remained in the aqueous fraction. This residual is likely to represent water-soluble estradiol metabolites or tritium contamination of the radiolabeled material. These observations indicated that hexane–ethyl acetate could be expected to remove essentially all of the endogenous estradiol from patient serum samples.
On the basis of these results, we used two extractions with 10 mL of organic solvent per 1 mL of serum. This procedure was used for all data presented in this study. We compared estradiol measurements after extraction with results of direct assays of the same samples (Fig. 1A
). The results indicated that the extraction of endogenous estradiol was essentially complete and that estradiol was accurately quantified in the reconstituted serum extracts.
|
We evaluated extraction of estradiol from serum followed by reconstitution of the dry extract in a minimal volume before assay as a means to improve the practical lower limit of the IMx estradiol assay. Serum was pooled from IVF patients with relatively high concentrations of endogenous hormone. Estradiol was then extracted from the pooled sample and reconstituted at an eightfold concentration with steroid-free IMx specimen diluent. The linearity of serial dilutions of the unextracted sample and of the concentrated sample was then examined in the IMx assay system (Fig. 1B
). For estradiol >100 ng/L, direct measurements agreed closely with measurements of estradiol after extraction. For concentrations <100 ng/L, the unextracted sample deviated from linearity on dilution. In contrast, the extracted sample was linear to the equivalent of 8 ng/L in the original serum sample.
The limit of detection of estradiol in unextracted patient samples was poor considering the need to quantify the clinically important low concentrations of estradiol in serum during the follicular phase of the menstrual cycle. In our laboratory, the dose at which serial dilutions of patient samples deviates from linearity based on the actual concentration of estradiol is considered a limit of clinical utility of an assay. In individual serum samples this concentration ranged from 
25 to 100 ng/L. Furthermore, at estradiol concentrations below 100 ng/L the measurements were associated with a CV >20%. Indeed, testing of individual samples in quadruplicate gave a lower apparent limit of detection because of the improved precision of determining the point at which the serial dilutions deviated from linearity. Using quadruplicate determinations, this limit for serum from IVF patients was not different from the 25 ng/L reported for normally cycling women (13).
As reported here, it was possible to achieve a detection limit of <10 ng/L with single-point determinations by extracting and concentrating the sample, thereby taking advantage of the very good accuracy and precision of the IMx assay at estradiol concentrations >50–100 ng/L. The described extraction procedure provides a means for the precise and accurate measurement of serum estradiol concentrations to 8 ng/L. The linearity of the concentrated sample indicates the absence of nonspecific interferences in the concentrated sample, and excellent agreement was observed between direct measurements and assay of extracted samples. Recovery of estradiol added to IVF patient sera by use of the described hexane–ethyl acetate procedure was essentially complete, and monitoring of recovery in each sample does not appear to be routinely necessary. Although we did not evaluate serum from postmenopausal women, it seems likely that extraction and concentration of samples before testing with the IMx assay offers considerable potential for measuring the low (<20 to 40 ng/L) concentrations of estradiol often encountered in such specimens.
Acknowledgments
This research was supported by the National Cooperative Program for Infertility Research (Grant HD 29164) and the Reproductive Endocrine Sciences Center (Grant HD 28138).
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) and concentrated (•) sample was measured over a series of dilutions to examine assay sensitivity. Error bars, 1 SD from the mean of quadruplicate measurements. The dashed line indicates a line with a slope of 1.