Immunoassay of Estradiol: Unanticipated Suppression by Unconjugated Estriol

Transcription

1 Clinical Chemistry 50: (2004) Endocrinology and Metabolism Immunoassay of Estradiol: Unanticipated Suppression by Unconjugated Estriol Zhimin (Tim) Cao, 1* Thomas A. Swift, 2 Clint A. West, 1 Thomas G. Rosano, 2 and Robert Rej 1,3 Background: Accurate measurement of estradiol is important in clinical settings. The quality of laboratory estimations of estradiol may be assessed through external quality-assurance surveys. Methods: Estradiol was measured by microparticle enzyme immunoassay (MEIA) and other immunoassays. Proficiency testing of medical laboratories was conducted using samples prepared from normal male human serum supplemented with exogenous estradiol and other steroid and nonsteroid hormones, and participant laboratories measured estradiol by a variety of commonly used immunoassay techniques. Results: The imprecision (CV) for measurement of estradiol [ ng/l ( pmol/l)] was <22% for most analytical techniques. Greater imprecision, as high as 40% for the same concentration range, was observed for the (AxSYM) MEIA method in the proficiency testing event of September Results from this method were bimodal in distribution. We found that unconjugated estriol at concentrations >1.5 g/l (>5.2 nmol/l) interfered with the MEIA method, leading to decreased recovery of added estradiol by up to 50%. This suppression in estradiol measurement was prevented by dilution of the specimen before measurement. Addition of unconjugated estriol gave a positive bias in some other immunoassay methods for estradiol. Poor comparability among the immunoassay methods for measurement of estradiol at clinically relevant concentrations [ 60 ng/l (220 pmol/l)] was revealed. 1 Laboratory of Molecular Diagnostics, Wadsworth Center for Laboratories and Research, New York State Department of Health, Albany, NY. 2 Department of Pathology and Laboratory Medicine, Albany Medical Center Hospital and College, Albany, NY. 3 Department of Biomedical Sciences, School of Public Health, State University of New York, Albany, NY. *Address correspondence to this author at: Laboratory of Molecular Diagnostics, Wadsworth Center, New York State Department of Health, Albany, NY Fax ; tim@wadsworth.org. Received June 18, 2003; accepted October 3, Previously published online at DOI: /clinchem Conclusions: A negative interference of unconjugated estriol with the MEIA method is a source of error for estradiol measurement. Lack of specificity and lack of comparability among immunoassay methods for estradiol may have detrimental effects on medical practice American Association for Clinical Chemistry The steroid hormone estradiol is the most potent among the estrogens. Accurate determination of estradiol concentrations in human serum and plasma is important in many clinical settings (1 4). In most clinical laboratories, estradiol is measured by immunoassay methods in conjunction with sensitive detection technologies and automated instrumentation. Such measurements are usually robust, economical, and precise (5). Estradiol is classified as a type A analyte by the WHO because it is a chemically well-defined compound (6). Although its measurement should be method-independent, performance quality for estradiol assay has been questioned for more than a decade (7 11). Difficulties in measurement of estradiol may include interfering substances, low endogenous concentration in most patient samples, a lack of analytical specificity, errors in calibration, and lack of traceability. In this study, we demonstrate that the presence of unconjugated estriol interferes with the measurement of estradiol in the AxSYM microparticle enzyme immunoassay (MEIA) method; we also study the nature of the interference, determine whether such interference occurs in other immunoassay methods, and examine the comparability among methods used in clinical practice for estradiol measurement. Materials and Methods Blood specimens were collected, without additives, from patients and from a healthy volunteer, and serum was prepared by leaving the specimens at room temperature for 30 min, followed by centrifugation at 2000g for 20 min. Serum was either used immediately or stored for 14 days at or below 40 C. This study was approved by the New York State Department of Health Institutional Re- 160

2 Clinical Chemistry 50, No. 1, view Board. Processed male human serum used for the preparation of proficiency testing samples was from Bioresource Technology, Inc. All steroids, from Sigma Aldrich Co., were dissolved in methanol before addition to serum. The immunoassay instruments and associated reagents that were used for measurements in our laboratories were as follows: AxSYM and IMx (Abbott Laboratories), and Elecsys 2010 (Roche Diagnostics Co.). Other instruments and methods involved in the study included Architect (Abbott Laboratories); Immuno 1, ACS:180, and ADVIA Centaur Estradiol-6 assay (Bayer Corp.); Coat-A- Count, Immulite, and Immulite 2000 (Diagnostic Products Corp.); Access (Beckman Coulter Inc.); Vitros ECi (Ortho Clinical Diagnostics); and VIDAS (biomérieux, Inc.). The AxSYM method is a heterogeneous immunoassay in which estradiol from the specimen binds to rabbit polyclonal anti-estradiol antibodies that are linked to microparticles. On removal of unbound materials, estradiol alkaline phosphatase conjugate is added and binds to available sites. After washing, 4-methylumbelliferyl phosphate is added, and the fluorescent product is measured; the intensity of the signal is inversely proportional to the estradiol concentration in specimens (12). Results We prepared proficiency testing samples with estradiol concentrations ranging from 20 to 600 ng/l. 4 Specimens were distributed to 430 laboratories enrolled in the endocrinology proficiency testing program; 159 of these submitted results for estradiol testing. The mean CV for all methods was 8.4% at estradiol concentrations 100 ng/l. Greater imprecision (CV 33 40%; n 16) at estradiol concentrations ranging from 150 to 400 ng/l was observed for the (AxSYM) MEIA method in the proficiency testing event of September Results from this method were bimodal in distribution, with apparent centers at 120 and 240 ng/l. 5 We examined the possibility that components of the proficiency testing specimens interfered with the AxSYM method. Steroid components of the endocrinology proficiency testing specimens are listed in Table 1. We examined their potential to interfere with the AxSYM method by adding estradiol to processed human serum (proficiency testing serum matrix) to a final concentration of 1000 ng/l, together with one of the steroids listed in Table 1 at either of two concentrations (one within and one above the reference intervals). Estradiol in each sample was measured by the AxSYM method in an undiluted sample and a 1:5 automated dilution. (Abbott Laboratories has instructed users of the AxSYM instrument to perform a 1:5 automatic dilution of samples for some specimens.) Of the steroids tested, only unconjugated 4 The conversion factors to molar units are: estradiol, ng/l 3.67 pmol/l; unconjugated estriol, g/l 3.47 nmol/l. 5 Details regarding this proficiency test, as well as others, are available at Table 1. Effect of other steroids on measurement of estradiol by the AxSYM (MEIA) method. Steroid added Measured estradiol, ng/l Undiluted 1:5 dilution Relative recovery (undiluted/1:5 dilution), a % Unconjugated estriol 5.0 g/l g/l Cortisol 108 g/l g/l Estriol 3-sulfate 5.0 g/l g/l Testosterone 4.2 g/l g/l Progesterone 7.4 g/l g/l a A range of 86 96% was obtained with use of three fresh patient sera without addition of exogenous estradiol and unconjugated estriol. estriol showed an inhibitory effect on estradiol measurements, in the undiluted specimen. The magnitude of this effect was substantially reduced when the assay was performed with the 1:5 automated dilution protocol. To confirm the results described above and to eliminate the possibility of a matrix effect in the pooled human serum, we prepared two series of specimens by adding various amounts of unconjugated estriol and a constant concentration of estradiol (200 ng/l) to fresh human serum and to the proficiency serum matrix. The concentrations for unconjugated estriol corresponded to the range of concentrations found in nonpregnant and pregnant women (first 28 weeks), and the estradiol concentration of 200 ng/l was characteristic of late follicular and mid-luteal phases (13). Estradiol was measured by the AxSYM method in both an undiluted sample and in a 1:5 automatic dilution. The measured estradiol concentrations in the specimens of both series were substantially decreased, by up to 50%, in the presence of unconjugated estriol at 20 g/l (Fig. 1, solid lines). However, the inhibitory effect of unconjugated estriol could be attenuated, from the original 50% down to 10%, by use of the 1:5 automated dilution (Fig. 1, dashed lines). We also examined the effect of exogenous estriol on the measurement of estradiol with the IMx, an instrument that also uses the MEIA technique, at four different concentrations of estradiol (Fig. 2). At a serum estradiol concentration of 750 ng/l, unconjugated estriol at 15 g/l suppressed the measured result by 20%, and the effect on the IMx assay (Fig. 2) was generally similar to that found for the AxSYM (Fig. 1). To determine whether an increase in the estradiol concentration could reduce or eliminate the inhibitory effect of unconjugated estriol, we prepared two sets of

3 162 Cao et al.: Estriol Interference in Estradiol Immunoassays Fig. 1. Effect of unconjugated estriol on the measurement of estradiol by the AxSYM instrument. Unconjugated estriol was added at the concentrations shown on the abscissa to a specimen containing estradiol at 200 ng/l. Specimens were proficiency test serum (F) and patient serum (f). Estradiol concentrations were determined by the AxSYM method, in either undiluted samples (solid lines) or in a 1:5 automatic dilution (dashed lines). The measured estradiol concentration at 1 g/l of unconjugated estriol was considered 100%. A separate experiment (not shown) showed no significant difference in recovery between the samples containing unconjugated estriol at 1 g/l and those without exogenous estriol. Data represent results of four measurements (error bars, SD). specimens using both fresh human serum and proficiency serum matrix. To each set of samples, unconjugated estriol was added at a concentration of 10 g/l and estradiol was added at concentrations ranging from 100 to 1000 ng/l. Estradiol was then measured by the AxSYM method without dilution. Increasing estradiol concentrations did not increase the percentage recovery of estradiol, nor did they reduce the inhibitory effect of unconjugated estriol (Fig. 3). To determine the relationship between the concentrations of estradiol and unconjugated estriol in association with the observed suppression, we prepared a series of 24 samples, using six fresh patient sera, with various concentrations of endogenous and/or added estradiol ( ng/l). These sera were aliquoted into four subspecimens, and unconjugated estriol (up to 30 g/l) was added. Estradiol was measured on the AxSYM instrument without dilution. We observed three different types of effect depending on the amount of estradiol in the specimens: a positive interference for estradiol concentrations of 100 ng/l; no apparent interference for estradiol at 100 ng/l; and a negative interference for estradiol 150 ng/l (Fig. 4). We examined the effect of unconjugated estriol on other immunoassay methods for estradiol by preparing two proficiency samples: one containing unconjugated estriol (at 4 g/l), and the other with no such addition. These samples were analyzed by participant laboratories using various immunoassay methods in the May 2001 and January 2002 proficiency testing events conducted by our laboratory. The results, summarized in Table 2, show a positive interference of unconjugated estriol in all of these methods, with one exception; however, the effect seen in this latter method was inconclusive because of the low number of participants and relatively high imprecision. The increase in the measured estradiol concentration attributable to addition of unconjugated estriol ranged from 21% to 389%. We repeated this study using patient serum, rather than proficiency serum matrix, and similar results were obtained (data not shown). Fig. 2. Effect of unconjugated estriol on the measurement of estradiol by the IMx instrument. Unconjugated estriol was added at the concentrations shown on the abscissa to proficiency test serum containing estradiol at 360 ( ), 750 (f), 1200 (F), or 1400 (Œ) ng/l. Estradiol concentrations were determined by the IMx method with no dilution. Data represent results of four measurements (error bars, SD). Fig. 3. Effect of estradiol concentration on the estradiol recovery decreased by the presence of unconjugated estriol. Specimens were proficiency test serum matrix (circles) and fresh patient serum (squares) containing estradiol ( ng/l) and unconjugated estriol at 10 g/l. Estradiol concentrations were measured by the AxSYM method without dilution. Closed symbols represent the mean of four measurements (error bars, SD). Open symbols indicate single measurements.

4 Clinical Chemistry 50, No. 1, Fig. 4. Pattern of unconjugated estriol interference with measurements of estradiol. Unconjugated estriol (0 30 g/l) and estradiol ( ng/l) were added to fresh patient sera, and estradiol concentrations were measured on an AxSYM instrument without dilution. The value of each point represents the mean of four measurements (error bars, SD). The measured estradiol concentrations on the ordinate are expressed in logarithmic scale. Further examination of Table 2 shows a high variability among the methods used to measure estradiol in the absence of unconjugated estriol. The overall method mean was 40.4 ng/l with a range of ng/l; the relative difference for each method is presented in Fig. 5. To rule out the possibility of a matrix effect, we also performed this experiment with serum from a nonpregnant female and obtained results similar to those shown in Fig. 5 (data not shown). Discussion Although estriol is the most abundant of the estrogens, it is less potent than estradiol. In pregnancy, the concentration of estriol can be increased several hundredfold. Interference attributable to unconjugated estriol in immunoassays has been described and is usually the result of positive cross-reactivity with the antibodies used for estradiol measurement (14). Unconjugated estriol is usually present at concentrations far exceeding the concentration of estradiol (15). However, a negative interference of unconjugated estriol in estradiol immunoassays has not been reported previously. We found that unconjugated estriol lowers the recovery of estradiol in the AxSYM (MEIA) method at several concentrations of the two steroids and in both patient specimens containing exogenous estriol and estradiol and in proficiency testing specimens. Our study revealed three types of effects of unconjugated estriol on estradiol measurements; the type of effect varied according to the concentration of estradiol in the specimen (Fig. 4). At estradiol concentrations 100 ng/l, the effect of estriol resembled that of a cross-reacting interferent. At estradiol concentrations 200 ng/l, the effect of estriol was to suppress the measured concentration of estradiol. In the range of ng/l estradiol, there was little apparent effect, presumably because of a Table 2. Effect of unconjugated estriol on immunoassays for estradiol measurement. Method Estradiol measured, ng/l (CV) a No unconjugated estriol added Unconjugated estriol added (4 g/l) Change attributable to addition of unconjugated estriol, % Architect 48.8 (18%) 39.3 (31%) 20 (n 3) b AxSYM 29.7 (17%) 48.0 (13%) 62 (n 16) IMx 40.9 (46%) 74.0 (23%) 81 (n 3) VIDAS 31.8 (25%) 99.3 (12%) 212 (n 4) ACS: (16%) (14%) 75 (n 14) ADVIA Centaur 64.6 (20%) (9%) 70 (n 30) Access 32.5 (24%) (9%) 389 (n 9) Immuno (10%) 45.1 (7%) 21 (n 5) Immulite 40.6 (24%) 71.1 (13%) 75 (n 32) Immulite (23%) 65.0 (12%) 106 (n 15) Coat-A-Count 46.4 (11%) 61.5 (20%) 33 (n 13) Vitros ECi 26.6 (13%) 69.0 (25%) 159 (n 5) Elecsys (25%) 69.0 (33%) 126 (n 6) Overall mean (CV) 40.4 (31%) 78.5 (42%) 94 a Results from laboratories participating in proficiency testing over the period May 2001 to January 2002; mean and interlaboratory CV (%). b Number of participants. cancellation effect between the two phenomena. Although we did not investigate the mechanism of these interferences, a suggestion put forth by Valdes and Jortani (12) sheds light on the possible mechanism of the negative type of effect. They proposed that interfering substances bind to antibodies during the initial reaction and then dissociate during the wash step. As a consequence, relatively more binding sites become available for phosphatase-labeled conjugate to occupy, giving an increased amount of fluorescent product and a decreased analytical result. Negative interference with the MEIA technique, similar to the one we present here, has been reported for the measurement of digoxin (16 20). Further investigations would be required to confirm whether this may be a general problem of the assay design in MEIA technology. The accuracy of measurement of specimens containing estradiol 200 ng/l and an increased concentration of estriol was improved by dilution. Abbott Laboratories, through the AxSYM estradiol package insert, currently has explicit instructions to users that patient specimens with an estradiol assay value greater than or equal to 250 pg/ml (ng/l) must be diluted, retested, and the diluted

5 164 Cao et al.: Estriol Interference in Estradiol Immunoassays Fig. 5. Comparison of immunoassay methods measuring estradiol. Aliquots of pooled male sera without added estradiol and unconjugated estriol were measured for estradiol concentrations using the following methods: A, Architect; B, AxSYM; C, IMx; D, VIDAS; E, ACS:180; F, ADVIA Centaur; G, Immuno 1; H, Access; I, Immulite; J, Immulite 2000; K, Coat-A-Count; L, Vitros ECi; M, Elecsys The mean of all estradiol measurements (40.4 ng/l) was considered as 0%. result reported. However, at various time points this concentration was either 150 or 250 ng/l. This procedure was to minimize issues associated with depressed results for fertility management patients. The cause of interference was not known, and we here identify unconjugated estriol as the likely source of error. Results from our proficiency surveys indicate that laboratories using the AxSYM method do not uniformly follow the manufacturer s guidance. We therefore observed a bimodal distribution of results from this method, with apparent centers of 120 and 240 ng/l. Participant laboratories in the former group reported results from 99 to 140 ng/l and likely failed to dilute the specimen. Laboratories in the latter group reported results from 201 to 275 ng/l and likely did perform the dilution that minimized the interference of estriol. No laboratories reported results in the range ng/l. The use of differing protocols among participant laboratories was likely the origin of the large CV obtained for the AxSYM method in the proficiency surveys. The important role of proficiency testing in the unmasking of laboratory errors is apparent from this study. Those proficiency samples that were found to have associated with them an increased imprecision for measurement of estradiol contained unconjugated estriol at concentrations 3 g/l. Concentrations far exceeding 3 g/l may be found during pregnancy, and even some nonpregnant females have been shown to have serum concentrations up to 2.4 g/l (15). Any concentration of estriol exceeding 1.5 g/l was found to depress the measured concentration of estradiol when estradiol exceeded 200 ng/l. The concentration range of estradiol that we examined also mimicked concentrations characteristic of pregnancy ( 200 ng/l) and of the early follicular phase ( 100 ng/l). Using patient specimens with added estriol and estradiol, we were able to reproduce all of the effects found with pooled serum or proficiency test samples. We therefore believe that the samples used in this study were commutable (21), both for analyte concentration and for the matrix used. Definitive confirmation of our findings would require obtaining clinical specimens with endogenous concentrations of estradiol and estriol similar to those that we prepared by adding exogenous steroids. Although the error attributable to the interference of estriol in the MEIA procedure may approach 50%, the clinical impact of this inhibitory effect may not be widespread because of the typically low concentration of unconjugated estriol in healthy nonpregnant women and because of the use of the recommended 1:5 dilution of patient samples by a proportion of laboratories. However, for certain medical conditions in which the unconjugated estriol concentration increases, e.g., pregnancy and obesity (22), results of estradiol measurements using the MEIA method on an AxSYM or IMx instrument may need to be confirmed by another technology. Poor agreement among methods was also revealed by proficiency testing (Table 2 and Fig. 5). We also found similar poor agreement when patient specimens were tested. This lack of agreement among methods may be related to issues of calibration and/or antibody specificity. A strategy for improving intermethod agreement has emerged from an experiment involving recalibration of all methods, using a standardized reference material (23). However, the continued existence of poor method agreement (Table 2 and Fig. 5) indicates that the latter model has yet to be fully adopted or that unstandardized reference methods are used by manufacturers within the in vitro diagnostics industry. The existing reference methods vary in their internal standards, derivatives, extraction procedures, and gas chromatography mass spectrometry conditions (24 29). Any of these variables can potentially influence measurements (29). For example, two different derivatization techniques to determine estradiol produced significantly different results (26); perhaps, therefore, standardization of the so-called reference methods for estradiol measurements may be required. Accurate measurement of estradiol is particularly important in an in vitro fertilization setting. Estradiol concentration is often used for dose adjustment during gonadotropin treatment and as an estimate of ovarian reserve, ovarian stimulation response, and pregnancy outcome (1 4). The discrepancies in the literature (30 32) regarding the recommendations for decisions based on serum estradiol concentration may well be attributable to the differential effect of estriol on different analytical methods or to the differences among the methods that we describe here (Table 2 and Figs. 1 5). For the specimen without exogenous estriol, the reported concentrations of estradiol ranged from 26.6 to 64.6 ng/l depending on the method used (Table 2); individual laboratory results ranged from 20 to 191 ng/l. For the specimen with exogenous estriol at 4 g/l, the reported concentrations of estradiol ranged from 39 to 159 ng/l, depending on the

6 Clinical Chemistry 50, No. 1, method used (Table 2); individual laboratory results ranged from 23 to 196 ng/l. These data show that the current state of practice for measurement of estradiol does not meet clinical needs, at least for certain applications (31). The presence of unconjugated estriol in specimens further compromises reliability (Table 2). In summary, an external quality-assurance/proficiency testing program revealed a source of error in the estimation of estradiol in serum. Deficiencies identified in this study occurred in three areas: (a) unconjugated estriol interferes with the measurement of estradiol in the AxSYM (MEIA) method; (b) inconsistency of participant laboratories in compliance with the manufacturer-recommended dilution protocol for the AxSYM instrument leads to poor interlaboratory agreement; and (c) there is a lack of agreement in results among immunoassay techniques for measurement of estradiol. Abbott Laboratories (Abbott Park, IL) provided reagents supporting portions of this project. References 1. Evers JL, Slaats P, Land JA, Dumoulin JC, Dunselman GA. Elevated levels of basal estradiol-17 predict poor response in patients with normal basal levels of follicle-stimulating hormone undergoing in vitro fertilization. Fertil Steril 1998;69: Mikkelsen AL, Andersson AM, Skakkebaek NE, Lindenberg S. Basal concentrations of oestradiol may predict the outcome of in-vitro maturation in regularly menstruating women. Hum Reprod 2001;16: Licciardi FL, Liu HC, Rosenwaks Z. Day 3 estradiol serum concentrations as prognosticators of ovarian stimulation response and pregnancy outcome in patients undergoing in vitro fertilization. Fertil Steril 1995;64: Phelps JY, Levine AS, Hickman TN, Zacur HA, Wallach EE, Hinton EL. Day 4 estradiol levels predict pregnancy success in women undergoing controlled ovarian hyperstimulation for IVF. Fertil Steril 1998;69: Ross JW, Fraser MD, Moore TD. Analytic clinical laboratory precision state of the art for thirty-one analysis. Am J Clin Pathol 1980;74(4 Suppl): World Health Organization. WHO Consultation on International Biological Standards for In Vitro Diagnostic Procedures, September 14 15, 2000, Geneva, Switzerland. biologicals/meeting-reports/doc/bivd00sep14.pdf (Accessed May 2003). 7. Taieb J, Benattar C, Birr AS, Lindenbaum A. Limitations of steroid determination by direct immunoassay. Clin Chem 2002;48: Middle J. Standardization of steroid hormone assays. Ann Clin Biochem 1998;35: Diver MJ, Nisbet JA. Warning on plasma oestradiol measurement. Lancet 1987;2: Sinicco A, Raiteri R, Rossati A, Savarino A, Perri GD. Efavirenz interference in estradiol ELISA assay. Clin Chem 2000;46: Mikkelsen AL, Borggaard B, Lebech PE. Results of serial measurement of estradiol in serum with six different methods during ovarian stimulation. Gynecol Obstet Invest 1996;41: Valdes R Jr, Jortani S. Unexpected suppression of immunoassay results by cross-reactivity: now a demonstrated cause for concern. Clin Chem 2002;48: Carr B. Disorders of the ovaries and female reproductive tract. In: Wilson JD, Foster DW, Kronenberg HM, Larse PR, et al. eds. Williams textbook of endocrinology, 9th ed. Philadelphia: WB Saunders, 1998: Kim JB, Barnard GJ, Collins WP, Kohen F, Lindner HR, Eshhar Z. Measurement of plasma estradiol-17 by solid-phase chemiluminescence immunoassay. Clin Chem 1982;28: Wright JV, Schliesman B, Robinson L. Comparative measurements of serum estriol, estradiol, and estrone in non-pregnant, premenopausal women; a preliminary investigation. Altern Med Rev 1999;4: Steimer W, Muller C, Eber B, Emmanuilidis K. Intoxication due to negative canrenone interference in digoxin drug monitoring. Lancet 1999;354: Steimer W, Müller C, Eber B. Digoxin assays: frequent, substantial, and potentially dangerous interference by spironolactone, canrenone, and other steroids. Clin Chem 2002;48: Dasgupta A, Trejo O. Suppression of total digoxin concentrations by digoxin-like immunoreactive substances in the MEIA digoxin assay. Elimination of negative interference by monitoring free digoxin concentrations. Am J Clin Pathol 1999;111: Dasgupta A, Biddle DA, Wells A, Datta P. Positive and negative interference of the Chinese medicine Chan Su in serum digoxin measurement. Elimination of interference by using a monoclonal chemiluminescent digoxin assay or monitoring free digoxin concentration. Am J Clin Pathol 2000;114: Datta P, Dasgupta A. Bidirectional (positive/negative) interference in a digoxin immunoassay: importance of antibody specificity. Ther Drug Monit 1998;20: Rej R, Proficiency testing, matrix effects, and method evaluation [Editorial]. Clin Chem 1994;40: Fishman J, Boyar RM, Hellman L. Influence of body weight on estradiol metabolism in young women. J Clin Endocrinol Metab 1975;41: Thienpont LM, De Leenheer AP. Efforts by industry toward standardization of serum estradiol-17 measurements. Clin Chem 1998;44: Siekmann L. Determination of oestradiol-17 in human serum by isotope dilution-mass spectrometry. Definitive methods in clinical chemistry, II. J Clin Chem Clin Biochem 1984;22: Thienpont LM, Verhaeghe PG, Van Brussel KA, and De Leenheer AP. Estradiol-17 quantified in serum by isotope dilution-gas chromatography-mass spectrometry: reversed-phase C18 highperformance liquid chromatography compared with immuno-affinity chromatography for sample pretreatment. Clin Chem 1988;34: Dehennin L. Estradiol-17 determined in plasma by gas chromatography-mass spectrometry with selected ion monitoring of mixed silyl ether-perfluoroacyl ester derivatives and use of various stableisotope-labeled internal standards. Clin Chem 1989;35: Gaskell SJ, Brownsey BG. Immunoadsorption to improve gas chromatography/high-resolution mass spectrometry of estradiol- 17 in plasma. Clin Chem 1983;29: Wu H, Ramsay C, Ozaeta P, Liu L, Aboleneen H. Serum estradiol quantified by isotope dilution gas chromatography/mass spectrometry. Clin Chem 2002;48: Wolthers BG, Kraan GP. Clinical applications of gas chromatography and gas chromatography-mass spectrometry of steroids. J Chromatogr A 1999;843: Ravhon A, Lavery S, Trew G, Margara R, Winston R. Predictive value of serum estradiol levels for IVF pregnancy outcome after gonadotropin stimulation? Fertil Steril 1999;71: Phophong P, Ranieri DM, Khadum I, Meo F, Serhal P. Basal 17 -estradiol did not correlate with ovarian response and in vitro fertilization treatment outcome. Fertil Steril 2000;74: Check JH. Low and high responders at what levels of serum estradiol do things start to get fuzzy? Fertil Steril 1999;71:582 3.