- Email: [email protected]
mass spectrometry in 99 human serum samples
Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97 – 104 www.elsevier.com/locate/jsbmb
Comparison of progesterone concentration determination by 12 non-isotopic immunoassays and gas chromatography/mass spectrometry in 99 human serum samples Philippe Boudou a,*, Joe¨lle Taieb b, Bruno Mathian c, Yves Badonnel d, Isabelle Lacroix e, Elisabeth Mathieu f, Franc¸oise Millot g, Nicole Queyrel h, Claude Somma-Delpero i, Marie-Claude Patricot c a
Department of Hormonal Biology, St. Louis Hospital, 1 a6enue Claude Vellefaux, 75010, Paris, France b Hormonology Laboratory, A.Be´cle`re Hospital, Clamart, France c Hormonology Laboratory, Lyon Sud Hospital, Lyon, France d Biochemistry Laboratory, A.Pinard Hospital, Nancy, France e CERBA Laboratory, Cergy Pontoise, France f Molecular Biology Laboratory, Angers Hospital, Angers, France g Biochemistry and Hormonology Laboratory, Tenon Hospital, Paris, France h Hormonology Laboratory, Versailles Hospital, Le Chesnay, France i Endocrinology Laboratory, Faculty of Medicine, Marseille, France Received 15 May 2000; accepted 9 March 2001
Abstract A single serum progesterone determination may be highly predictive for early pregnancy and in vitro fertilisation and embryo-transfer outcomes. We therefore compared 12 direct non-isotopic progesterone immunoassays with gas-chromatography/ mass spectrometry (GC/MS). For each assay, data from the analysis of 99 individual sera were compared with data obtained by GC/MS, using regression and bias plot analyses and the ratio method. We observed a larger difference in concentration between high and low values and a broader distribution of results for immunoassays than for GC/MS. All immunoassays displayed bias in the calibration process and a lack of specificity and/or sensitivity, to various degrees. We tried to identify the parameters of the assay procedure that might contribute to these discrepancies. None of the criteria investigated (antibodies, control and preparation of calibrators, blocking agents and choice of tracer) had a significant effect when studied alone. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Gas-chromatography; Immunoassays; Mass-spectrometry; Progesterone
1. Introduction Serum progesterone (P) levels are often determined to assess corpus luteum function and to detect luteal phase defects. On days 20– 23 of the menstrual cycle, serum P assays can be used to confirm that ovulation has occurred. A single P determination may be as valuable as repeated hCG measurements for estimating the risk of early pregnancy complications [1] and has greater pre* Corresponding author. Tel.: + 33-1-42499391; fax: + 33-142494280.
dictive value for pregnancy outcome than a single hCG determination [2]. A retrospective study demonstrated that a cut-off level of P\ 55.7 nmol/l could be used to identify patients at risk of ectopic pregnancy, without the need for ultrasound or invasive diagnostic techniques [3]. In contrast, decisions concerning treatment (surgery vs methotrexate) cannot be taken on the basis of a single P determination in most patients with ectopic pregnancy [1], although Ransom et al. [4] showed that in a very small group of patients a single P determination may be useful for predicting methotrexate success. Prospective studies have also shown that P
0960-0760/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0960-0760(01)00078-4
98
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
cut-off values of 1.27– 2.86 nmol/l, on the day of hCG administration, can be used to predict pregnancy rates for in vitro fertilisation and embryo tranfer [5–7]. Overall, therefore, it appears that previous studies have shown that a single serum P assay may be highly predictive of clinical outcome and may be used as a diagnostic aid for various clinical events. Nevertheless, different progesterone cut-off values have been reported for predicting abnormal pregnancy outcomes [2,8– 10]. These differences may be partly due to differences in methods [10] and to the small number of samples tested [2,8,9]. The aim of this study was to try to verify test performance by comparing 12 commercial direct automated non-isotopic P immunoassays with a reference method based on isotope dilution associated with gas chromatography/mass spectrometry (GC/MS).
France)a, ACS-180 (Chiron Diagnostics, Cergy-Pontoise, France)g, Access (Diagnostics Pasteur, Marnes-laCoquette, France)c, Immulite (Diagnostics Products Corp., La Garenne Colombes, France)f, Auto-Delfia (EG&G Instruments, Evry, France)i, AIA 600 (Eurogenetics, Orle´ ans-La Source, France)d, Magia (Merck Clevenot-Biotrol, Nogent-sur-Marne, France)h, Amerlite and Vitros Eci (Ortho Clinical Diagnostics, Issy-lesMoulineaux, France)b, Elecsys (Roche-BoehringerMannheim, Meylan, France)f. The number of replicates for each assay was as recommended by the manufacturer: one for all assays except the Amerlite assay and GC/MS, which were performed in duplicate. For both of these assays, a mean P concentration was calculated. All samples were assayed in each of the laboratories, using kits from the same batch with the same labelling procedure and the same control calibrators.
2. Materials and methods
2.3. GC/MS method 2.1. Samples tested Blood samples were drawn from female outpatients attending the endocrinology clinic of Saint-Louis Hospital. Samples containing sufficient serum for testing with all P assays were used to constitute a serum sample bank. This bank consisted of 99 samples comprising equal numbers of samples with low, medium and high P levels. The serum bank thus covered the wide range of concentrations found in clinical practice. Blood samples were collected in a glass VacutainerR system and were separated by centrifugation. No preservatives were added. All samples were dispensed in 300 ml (or 2 ml for GC/MS) aliquots in screw-capped EppendorfR tubes, frozen and stored at −70 °C. Samples were transported, on solid CO2, to the nine French hospital laboratories participating in the study. In the laboratories, the serum samples were stored at − 20 °C. They were thawed at room temperature before use and tested within 8 h of thawing. None of the samples was frozen and thawed repeatedly.
2.2. Immunological methods The study was performed from December 1997 to January 1998 and began simultaneously in all laboratories. Twelve manufacturers agreed to participate in the evaluation, providing kits and, when necessary, installing analysers. The study included one manual single-analyte and 11 automated multi-analyte systems. All were direct non-isotopic assays1: Axsym (Abbott, Rungis, France)e, Immuno-1 (Bayer Diagnostics, Puteaux, France)c, Vidas (Bio-Me´ rieux, Marcy l’Etoile, 1 Superscript letters (as for the authors’ addresses on page 1) were used to indicate which laboratory performed the immunoassay.
The isotope dilution GC/MS assay has been described in detail elsewhere [11]. Briefly, analysis was carried out using a Carlo Erba chromatograph coupled to a QMD 1000 mass spectrometer (Thermoquest France, Les Ulis, France). GC separation was achieved on a DB5 (J&W Scientific, Folsom, USA) fused-silica column (30 m× 0.32 mm internal diameter; 0.25 mm film thickness) operated at a helium inlet pressure of 100 kPa and a temperature of 220 °C. Samples (2–4 ml) were injected at 275 °C using a mobile needle injector. The retention time of P was about 10 min. The ion source was operated in the electron impact mode with 70eV electron energy and the photomultiplier was set to 700V. A stable isotopically labelled steroid in which two carbon atoms in positions three and four were replaced by 13C ([3,4-13C2]-progesterone) (Commissariat a` l’Energie Atomique, Gif sur Yvette, France) was used as an internal standard, as previously described [12– 14]. Two to 10 ng of labelled P ([3,4-13C2]-progesterone) was added to 0.5–2.0 ml samples. The unlabelled/labelled P ratio was kept between 0.5 and 1.0. Extraction was carried out with 5 ml of diethyl-ether and the extract was purified on a column (6 mm internal diameter) containing 0.45 g of propanediol-impregnated celite. P was eluted with 2 ml of isooctane. P was then recovered by evaporation under nitrogen. The dry residue was dissolved in 50 ml of heptafluorobutyric anhydride and 200 ml of anhydrous acetonitrile and was incubated for 30 min at 37°C (derivatisation). The reaction mixture was subjected to evaporation under nitrogen, the residue was dissolved in 20 ml of toluene and 2–4 ml was injected into the GC/MS apparatus. P was quantified by monitoring 510 and 512 m/z ions, corresponding to the ions of P and [3,4-13C2]-progesterone derivatives respectively.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
The standard calibration curve was produced using 0.0, 0.5 and 1.0 P concentration ratios (unlabelled/labelled) and corresponded to peak-area ratios (ion 510/ ion 512). The overlapping of: (i) unlabelled P peak area at mass 512; and (ii) labelled P peak area at mass 510 was corrected for calibration curves and samples as previously described [15]. A preliminary purification step prevented the generation of interfering peaks on the fragmentogram. The detection limit was below 300 pmol/l. The inter-run coefficient of variation was lower than 2.5% for P concentrations between 3.0 and 55.0 nmol/l. Standards and samples were analysed in duplicate and the mean of the two runs was considered.
2.4. Statistical analysis The results were analysed by: (i) linear regression analysis (reference method: x values, assay tested: y values); (ii) the method of Bland and Altman (P level measured by GC/MS −P level measured by assay tested) vs (sum of P level measured by (GC/MS +assay tested)/2) [16]; (iii) the ratios method (result from GC/ MS/result from assay tested) expressed in the form of histograms (frequency (%) vs ratio) [17]. Linear regression analysis (slope, intercept, correlation coefficient (r) and residual SD of the distribution around the regression line) was performed using Statview SE Software.
3. Results The serum P levels of six of the samples were below the detection limit for the GC/MS method ( B 0.3 nmol/l) and were excluded from the overall analysis. Thus, comparisons were performed using the remaining 93 serum samples, although it was not possible to obtain a result for all of the samples in all of the assays. Indeed, some serum samples had P levels below the
99
detection limits of particular tests: Elecsys (0.48 nmol/l) (n= 21), Vidas (0.32 nmol/l) (n = 3), AIA-600 (0.32 nmol/l) (n= 3), Vitros Eci (0.25nmol/l) (n= 5). The results of the linear regression analysis are shown in Table 1. Most of the immunoassays displayed a slopeB 1 [from 0.667 (Access) to 1.034 (AIA-600)] and a greater variety of intercepts [from − 0.098 (Magia) to 3.506 nmol/l (Access)] than GC/MS. All the immunoassays had r values above 0.9, and standard deviation values (SD) were from 1.64 (Axsym) to 5.816 (AIA-600). P levels were widely distributed and large differences in the intrinsic specificity of the various P assays was noted with respect to GC/MS. The results for Bland and Altman analysis are presented in Fig. 1. Large differences in P concentration were observed between non-isotopic assays and GC/ MS. The mean differences between these non-isotopic assays and GC/MS were significantly different from zero and tended to be either positive (Immuno-1) or negative (Axsym, Vidas, Elecsys, Magia). The assays from Axsym, Vidas, Autodelfia, Amerlite, Access, Immulite, Vitros ECi and Magia underestimated P concentration at high levels (up to 10 nmol/l). In contrast, assays from Immulite, ACS-180, Immuno-1, Access, Vitros ECi, Amerlite, and Autodelfia overestimated P concentration at low levels (from 0.3 to 10.0 nmol/l). Assays from Elecsys and AIA-600 showed extensive scattering of the data over the entire concentration range, with overestimation at high concentrations. All except Axsym showed a high percentage of outliers if P levels below 5 nmol/l were excluded. Ratios (P level in each assay vs P level in GC/MS) were classified into six arbitrary groups (R B0.50, 0.505 RB 0.85, 0.855 RB 1.15, 1.155 RB2.00, 2.005 RB 5.00, R \ 5). The frequency (%) and distribution of the ratios (R) are shown in Fig. 2. A broad distribution of R was observed for each of the immunoassays. The extreme values of the ratios were as
Table 1 Comparison of 12 non-isotopic progesterone immunoassays with GC/MS by linear regression analysis Progesterone assay
n
Slope
Intercept (nmol/l)
Residual SD (nmol/l)
Correlation coefficient (r)
Abbott: Axsym Bayer Diagnostics: Immuno.1 Bio Me´ rieux: Vidas Chiron Diagnostics: ACS-180 Diagnostics Pasteur: Access Diagnostics Prod Corp: Immulite EG&G instruments: Auto-Delfia Eurogenetics: AIA-600 Merck Clevenot-Biotrol: Magia Ortho Clin Diagnostics: Amerlite Ortho Clin Diagnostics: Vitros Eci Roche Boehringer: Elecsys
93 93 90a 93 93 93 93 90a 93 93 88a 72a
0.816 0.935 0.841 0.928 0.667 0.790 0.768 1.034 0.733 0.781 0.707 0.853
1.015 3.202 0.276 1.528 3.506 2.690 2.180 0.264 −0.098 3.017 2.550 0.142
1.640 4.051 2.799 2.485 3.411 3.972 2.610 5.816 5.421 4.148 4.533 3.401
0.993 0.970 0.982 0.988 0.958 0.960 0.981 0.951 0.918 0.955 0.939 0.977
a Sample levels below the detection limit defined by the manufacturers were excluded from the analysis [Elecsys (n = 21), Vitros ECi (n= 5), Vidas (n =3), AIA-600 (n= 3)].
100
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
Fig. 1.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
101
Fig. 1. Plots representing difference in P concentrations (levels measured by the assay tested minus levels measured by GC/MS) versus mean P concentrations (levels measured for each sample by the two methods/2).
follows: Immulite (0.47– 39.64), Immuno-1 (0.78– 21.46), Vitros ECi (0.40– 10.51), Autodelfia (0.60– 14.78), Amerlite (0.55– 12.00), AIA-600 (0.38– 3.41), Elecsys (0.32–3.17), Axsym (0.54– 5.54) respectively. Thirty percent of the values obtained with Elecsys, ACS-180, AIA-600 and Axsym were between 0.85 and 1.15. Immunoassays from Immulite, Amerlite and Access resulted in R values of which more than 50% were \2. The frequency of the highest R values (\ 5) was important for Autodelfia, Vitros ECi, ACS-180, Access, Immulite, Amerlite and Immuno-1 and ranged from 6.45% (Autodelfia) to 21.50% (Immuno-1).
4. Discussion A single serum P determination has been shown to be useful in clinical assessment of the risk of early pregnancy complications [1]. Retrospective and prospective studies have defined cut-off levels of P for the reliable identification of patients at high risk of ectopic pregnancy [3] and for the prediction of pregnancy rates in in vitro fertilisation and embryo transfer [5– 7]. This study focused on differences in the results obtained with 12 P
immunoassays tested in nine laboratories and with GC/ MS. Serum P assays were performed with unextracted samples. We found that most of the immunoassays underestimated serum P concentration at high levels and overestimated P concentration at low levels. There were great discrepancies between the results obtained with immunoassays and GC/MS; these depended on the immunoassay. Thus, a wide range of P ratios was obtained for all immunoassays. Some displayed high levels (ie\ 5) of ratios (R) with a high frequency (from 10% to 21.5%). The AIA-600 and Elecsys assays also resulted in extensive scattering of the data for the entire concentration range. There are several factors that might explain these discrepancies. First, these discrepancies may be due to the presence of steroids structurally similar to P (eg 5a/5b pregnane 3-20 dione, allopregnanolone, pregnanolone) which highly cross-reacted with P [18–21]. Some manufacturers had included highly specific polyclonal (Access, AIA-600, Amerlite, Autodelfia [22], Elecsys, Immulite, Vitros ECi) or monoclonal antibodies (ACS-180, Axsym [23], Immuno-1 [24], Magia, Vidas) in their immunoassays, but this did not improve specificity. Second, they may be due to the presence of binding proteins. Inclusion of blocking
102
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
Fig. 2.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
103
n
result from the assay tested vs. frequency (%) the distribution of the P concentration obtained by the two result from GC/MS methods. R were arbitrarely classified as follows: R B0.50, 0.50 =R B 0.85, 0.85 =R B 1.15, 1.15 = RB2.00, 2.00 = R B5.00, R\ 5. Fig. 2. Histograms
ratio: R:
agents (eg danazol (ACS-180, Autodelfia, Elecsys), bovine serum albumin (Axsym), products of unknown origin (Amerlite, Vitros ECi)) may prevent interference from binding proteins [25]; but there was no difference in results between immunoassays including blocking agents and those without blocking agents (eg Access, AIA-600, Immulite, Immuno-1, Magia, Vidas). None gave results very similar to those obtained by GC/MS. Third, they may be due to the choice of tracer. This parameter has been shown to be of critical importance in the development of a reliable immunoassay. Immunoassays using alkaline phosphatase (AP) gave higher P levels than those using horseradish peroxidase (HRP) or europium when only the signal generator was modified [26]. In our study, we found no clear difference in serum P levels over the entire concentration range, between AP-based immunoassays (e.g. Access, AIA-600, Axsym, Immulite, Immuno-1, Magia, Vidas) and HRP-based immunoassays (Ammerlite, Vitros ECi) (data not
shown). Fourth, other factors such as the nature of the assay buffer (composition, ionic strength, pH) and the purity of the preparation and control of calibrators might intervene. Finally, the approach for standardising calibrators by checking the standard calibration curve (e.g. Amerlite, ACS-180 and Elecsys) by GC/MS did not seem to be a criterion for reliability. Overall, our study showed large differences between non-isotopic P immunoassays and GC/MS for individual human serum samples. None of these assays gave results very similar to those obtained by GC/MS over the entire range of P concentrations tested. Improvments in calibration procedures seem to be necessary. The specificity/sensitivity criteria also require reassessment. None of these immunoassays seemed to have the high sensitivity and reliability required for screening for abnormal early pregnancy in asymptomatic women [2] or for monitoring in vitro fertilisation/embryo-transfer protocols [6,27].
104
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
References [1] J.D. Issacs, N.S. Whitworth, B.D. Cowan, Relative operating characteristic analysis in reproductive medicine: comparison of progesterone and human chorionic gonadotropin doubling time as predictors of early gestational normalcy, Fertil. Steril. 62 (1994) 452 – 455. [2] C.A. Daily, S.L. Laurent, W.C. Nunley, The pronostic value of serum progesterone and quantitative b-human chorionic gonadotropin in early human pregnancy, Am. J. Obstet. Gynecol. 171 (1994) 380 –384. [3] M.L. McCord, D.A. Muran, J.E. Buster, K.L. Arheart, T.G. Stovall, S.A. Carson, Single serum progesterone as a screen for ectopic pregnancy exchanging specificity and sensibility to obtain optimal test performance, Fertil. Steril. 66 (1996) 513 – 516. [4] M.X. Ransom, A.J. Garcia, M. Bohrer, G.H. Corsan, E. Kemmann, Serum progesterone as predictor of methotrexate success in the treatment of ectopic pregnancy, Obstet. Gynecol. 83 (1994) 1033 – 1037. [5] W. Schoolcraft, E. Sinton, T. Schlenker, D. Huynh, F. Hamilton, D.R. Meldrum, Lower pregnancy rate with premature luteinization during pituitary suppression with leuprolide acetate, Fertil. Steril. 55 (1991) 563 –566. [6] K.M. Silverberg, W.N. Burns, D.L. Olive, R.M. Riehl, R.S. Schenken, Serum progesterone levels predict success of in vitro fertilization/embryo transfer in patients stimulated with leuprolide acetate and menopausal gonadotropins, J. Clin. Endocrinol. Metab. 73 (1991) 797 –803. [7] R. Fanchin, D. de Ziegler, J. Taieb, A. Hazout, R. Frydman, Premature elevation of plasma progesterone alters pregnancy rates of in vitro fertilization and embryo transfer, Fertil. Steril. 59 (1993) 1090 – 1094. [8] R.H. Buck, S.M. Joubert, R.J.Norman, Serum progesterone in the diagnosis of ectopic pregnancy: a valuable diagnostic test? Fertil. Steril. 50 (1988) 752 –755. [9] B.D. Cowan, D.T. Vandermolen, C.A. Long, N.S. Whitworth, Receiver-operator characteristic, efficiency analysis, and predictive value of serum progesterone concentration as a test for abnormal gestations, Am. J. Obstet. Gynecol. 166 (1992) 1729 – 1734. [10] C.M. Peterson, D. Kreber, P. Delgado, T.T. Hung, Laboratory and clinical comparison of a rapid versus a classic progesterone radioimmunoassay for use in determining abnormal and ectopic pregnancies, Am. J. Obstet. Gynecol. 166 (1992) 562 – 566. [11] B.G. Wolthers, G.P.B. Kraan, Clinical applications of gas chromatography and gas chromatography – mass spectrometry of steroids, J. Chromatogr. 843 (1999) 247 –274. [12] M.C. Patricot, B. Mathian, S. Serpentie, A. Revol, Determination of urinary estradiol using a enzymatic method during the menstrual cycle. Comparison with a method using isotope dilution – mass spectrometry (ID-MS), Clin. Chem. Acta 158 (1986) 139 – 146. [13] L. Dehennin, Estradiol-17 beta determined in plasma by gas chromatography – mass spectrometry with selected ion monitoring of mixed silyl ether- perfluoroacyl ester derivatives and use of various stable-isotope-labeled internal standards, Clin. Chem. 35 (1989) 532 – 536.
[14] L.M. Thienpont, V.I. De Brabandere, D. Stockl, A.P. De Leenheer, Use of cyclodextrins for prepurification of progesterone and testosterone from human serum prior to determination with isotope dilution gas chromatography/mass spectrometry, Anal. Chem. 66 (1994) 4116 – 4119. [15] L. Siekman, Isotope dilution mass spectrometriy of steroid hormones — a definitive method in clinical chemistry, in: A.P. De Leenheer, R. Roncucci, C. Van Peterghen (Eds.), Quantitative Mass Spectrometry in Life Science, vol. 2, Elsevier, Amersham, 1978, pp. 3 – 16. [16] J.M. Bland, D.G. Altmanm, Statistical method for assessing agreement between two methods of clinical measurement, The Lancet 8 (1986) 307 – 310. [17] K. Nahoul, L. Dehennin, R. Scholler, Radioimmunoassay of plasma progesterone after oral administration of micronized progesterone, J. Steroid Biochem. 26 (1987) 241 – 249. [18] T. Ba¨ ckstro¨ m, A. Andersson, D.T. Baird, G. Selstam, The human corpus luteum secretes 5a-pregnane-3,20-dione, Acta Endocrinologica 111 (1986) 116 – 121. [19] P.J. Schmidt, R.H. Purdy, P.H. Moore Jr, S.M. Paul, D.R. Rubinow, Circulating levels of anxiolytic steroids in the luteal phase in women with premenstrual syndrome and in control subjects, J. Clin. Endocrinol. Metab. 79 (1994) 1256 – 1260. [20] M. Wang, L. Seippel, R.H. Purdy, T. Ba¨ ckstro¨ m, Relationship between symptom severity and steroid variation in women with premenstrual syndrome: study on serum pregnenolone, pregnenolone sulfate, 5a-pregnane-3,20-dione and 3a-hydroxy-5apregnan-20-one, J. Clin. Endocrinol. Metab. 81 (1996) 1076 – 1082. [21] A.R. Genazzani, F. Petraglia, F. Bernardi, E. Casarosa, C. Salvestroni, A. Tonetti, A. R.E. Nappi, S. Luisi, M. Palumbo, R.H. Purdy, M. Luisi, Circulating levels of allopregnanolone in human: gender, age, and endocrine influences, J. Clin. Endocrinol. Metab. 83 (1998) 2099 – 2103. [22] T. Lo¨ vgren, L. Merio¨ , K. Mitrunen, M.L. Ma¨ kinen, M. Ma¨ kela¨ , K. Blomberg, T. Palenius, K. Pettersson, One-step all-in-one dry reagent immunoassays with fluorescent europium chelate label and time-resolved fluorometry, Clin. Chem. 42 (1996) 1196 –1201. [23] D.H. Wilson, W. Groskopf, S. Hsu, D. Caplan, T. Langner, M. Baumann, D. DeManno, G. Williams, D. Payette, C. Dacel, D. Lynch, G. Manderino, Rapid, automated assay for progesterone on the Abbott AxSYMTM analyzer, Clin. Chem. 44 (1998) 86 – 91. [24] A. Levesque, M. Letellier, P. Dillon, A. Grandt, Analytical performance of Bayer Immuno I™ estradiol and progesterone assays, Clin. Chem. 43 (1997) 1601 – 1609. [25] J.V. Micallef, M.M. Hayes, A. Latif, R. Ahsan, S.B. Sufi, Serum binding of steroid tracers and its possible effects on direct steroid immunoassay, Ann. Clin. Biochem. 32 (1995) 566 – 574. [26] B.S. Prakash, H.H.D. Meyer, E. Schallenberger, D.F.M. Van de Wiel, Development of a sensitive enzyme immunoassay (EIA) for progesterone determination in unextracted bovine plasma using the second antibody technique, J. Steroid Biochem. 28 (1987) 623 – 627. [27] S.D. Prien, M.S. Canez, R.H. Messer, Increases in progesterone after human chorionic gonadotropin administration may predict cycle outcome in patients undergoing in vitro fertilization-embryo transfer, Fertil. Steril. 62 (1994) 1066 – 1068.
Comparison of progesterone concentration determination by 12 non-isotopic immunoassays and gas chromatography/mass spectrometry in 99 human serum samples Philippe Boudou a,*, Joe¨lle Taieb b, Bruno Mathian c, Yves Badonnel d, Isabelle Lacroix e, Elisabeth Mathieu f, Franc¸oise Millot g, Nicole Queyrel h, Claude Somma-Delpero i, Marie-Claude Patricot c a
Department of Hormonal Biology, St. Louis Hospital, 1 a6enue Claude Vellefaux, 75010, Paris, France b Hormonology Laboratory, A.Be´cle`re Hospital, Clamart, France c Hormonology Laboratory, Lyon Sud Hospital, Lyon, France d Biochemistry Laboratory, A.Pinard Hospital, Nancy, France e CERBA Laboratory, Cergy Pontoise, France f Molecular Biology Laboratory, Angers Hospital, Angers, France g Biochemistry and Hormonology Laboratory, Tenon Hospital, Paris, France h Hormonology Laboratory, Versailles Hospital, Le Chesnay, France i Endocrinology Laboratory, Faculty of Medicine, Marseille, France Received 15 May 2000; accepted 9 March 2001
Abstract A single serum progesterone determination may be highly predictive for early pregnancy and in vitro fertilisation and embryo-transfer outcomes. We therefore compared 12 direct non-isotopic progesterone immunoassays with gas-chromatography/ mass spectrometry (GC/MS). For each assay, data from the analysis of 99 individual sera were compared with data obtained by GC/MS, using regression and bias plot analyses and the ratio method. We observed a larger difference in concentration between high and low values and a broader distribution of results for immunoassays than for GC/MS. All immunoassays displayed bias in the calibration process and a lack of specificity and/or sensitivity, to various degrees. We tried to identify the parameters of the assay procedure that might contribute to these discrepancies. None of the criteria investigated (antibodies, control and preparation of calibrators, blocking agents and choice of tracer) had a significant effect when studied alone. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Gas-chromatography; Immunoassays; Mass-spectrometry; Progesterone
1. Introduction Serum progesterone (P) levels are often determined to assess corpus luteum function and to detect luteal phase defects. On days 20– 23 of the menstrual cycle, serum P assays can be used to confirm that ovulation has occurred. A single P determination may be as valuable as repeated hCG measurements for estimating the risk of early pregnancy complications [1] and has greater pre* Corresponding author. Tel.: + 33-1-42499391; fax: + 33-142494280.
dictive value for pregnancy outcome than a single hCG determination [2]. A retrospective study demonstrated that a cut-off level of P\ 55.7 nmol/l could be used to identify patients at risk of ectopic pregnancy, without the need for ultrasound or invasive diagnostic techniques [3]. In contrast, decisions concerning treatment (surgery vs methotrexate) cannot be taken on the basis of a single P determination in most patients with ectopic pregnancy [1], although Ransom et al. [4] showed that in a very small group of patients a single P determination may be useful for predicting methotrexate success. Prospective studies have also shown that P
0960-0760/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0960-0760(01)00078-4
98
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
cut-off values of 1.27– 2.86 nmol/l, on the day of hCG administration, can be used to predict pregnancy rates for in vitro fertilisation and embryo tranfer [5–7]. Overall, therefore, it appears that previous studies have shown that a single serum P assay may be highly predictive of clinical outcome and may be used as a diagnostic aid for various clinical events. Nevertheless, different progesterone cut-off values have been reported for predicting abnormal pregnancy outcomes [2,8– 10]. These differences may be partly due to differences in methods [10] and to the small number of samples tested [2,8,9]. The aim of this study was to try to verify test performance by comparing 12 commercial direct automated non-isotopic P immunoassays with a reference method based on isotope dilution associated with gas chromatography/mass spectrometry (GC/MS).
France)a, ACS-180 (Chiron Diagnostics, Cergy-Pontoise, France)g, Access (Diagnostics Pasteur, Marnes-laCoquette, France)c, Immulite (Diagnostics Products Corp., La Garenne Colombes, France)f, Auto-Delfia (EG&G Instruments, Evry, France)i, AIA 600 (Eurogenetics, Orle´ ans-La Source, France)d, Magia (Merck Clevenot-Biotrol, Nogent-sur-Marne, France)h, Amerlite and Vitros Eci (Ortho Clinical Diagnostics, Issy-lesMoulineaux, France)b, Elecsys (Roche-BoehringerMannheim, Meylan, France)f. The number of replicates for each assay was as recommended by the manufacturer: one for all assays except the Amerlite assay and GC/MS, which were performed in duplicate. For both of these assays, a mean P concentration was calculated. All samples were assayed in each of the laboratories, using kits from the same batch with the same labelling procedure and the same control calibrators.
2. Materials and methods
2.3. GC/MS method 2.1. Samples tested Blood samples were drawn from female outpatients attending the endocrinology clinic of Saint-Louis Hospital. Samples containing sufficient serum for testing with all P assays were used to constitute a serum sample bank. This bank consisted of 99 samples comprising equal numbers of samples with low, medium and high P levels. The serum bank thus covered the wide range of concentrations found in clinical practice. Blood samples were collected in a glass VacutainerR system and were separated by centrifugation. No preservatives were added. All samples were dispensed in 300 ml (or 2 ml for GC/MS) aliquots in screw-capped EppendorfR tubes, frozen and stored at −70 °C. Samples were transported, on solid CO2, to the nine French hospital laboratories participating in the study. In the laboratories, the serum samples were stored at − 20 °C. They were thawed at room temperature before use and tested within 8 h of thawing. None of the samples was frozen and thawed repeatedly.
2.2. Immunological methods The study was performed from December 1997 to January 1998 and began simultaneously in all laboratories. Twelve manufacturers agreed to participate in the evaluation, providing kits and, when necessary, installing analysers. The study included one manual single-analyte and 11 automated multi-analyte systems. All were direct non-isotopic assays1: Axsym (Abbott, Rungis, France)e, Immuno-1 (Bayer Diagnostics, Puteaux, France)c, Vidas (Bio-Me´ rieux, Marcy l’Etoile, 1 Superscript letters (as for the authors’ addresses on page 1) were used to indicate which laboratory performed the immunoassay.
The isotope dilution GC/MS assay has been described in detail elsewhere [11]. Briefly, analysis was carried out using a Carlo Erba chromatograph coupled to a QMD 1000 mass spectrometer (Thermoquest France, Les Ulis, France). GC separation was achieved on a DB5 (J&W Scientific, Folsom, USA) fused-silica column (30 m× 0.32 mm internal diameter; 0.25 mm film thickness) operated at a helium inlet pressure of 100 kPa and a temperature of 220 °C. Samples (2–4 ml) were injected at 275 °C using a mobile needle injector. The retention time of P was about 10 min. The ion source was operated in the electron impact mode with 70eV electron energy and the photomultiplier was set to 700V. A stable isotopically labelled steroid in which two carbon atoms in positions three and four were replaced by 13C ([3,4-13C2]-progesterone) (Commissariat a` l’Energie Atomique, Gif sur Yvette, France) was used as an internal standard, as previously described [12– 14]. Two to 10 ng of labelled P ([3,4-13C2]-progesterone) was added to 0.5–2.0 ml samples. The unlabelled/labelled P ratio was kept between 0.5 and 1.0. Extraction was carried out with 5 ml of diethyl-ether and the extract was purified on a column (6 mm internal diameter) containing 0.45 g of propanediol-impregnated celite. P was eluted with 2 ml of isooctane. P was then recovered by evaporation under nitrogen. The dry residue was dissolved in 50 ml of heptafluorobutyric anhydride and 200 ml of anhydrous acetonitrile and was incubated for 30 min at 37°C (derivatisation). The reaction mixture was subjected to evaporation under nitrogen, the residue was dissolved in 20 ml of toluene and 2–4 ml was injected into the GC/MS apparatus. P was quantified by monitoring 510 and 512 m/z ions, corresponding to the ions of P and [3,4-13C2]-progesterone derivatives respectively.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
The standard calibration curve was produced using 0.0, 0.5 and 1.0 P concentration ratios (unlabelled/labelled) and corresponded to peak-area ratios (ion 510/ ion 512). The overlapping of: (i) unlabelled P peak area at mass 512; and (ii) labelled P peak area at mass 510 was corrected for calibration curves and samples as previously described [15]. A preliminary purification step prevented the generation of interfering peaks on the fragmentogram. The detection limit was below 300 pmol/l. The inter-run coefficient of variation was lower than 2.5% for P concentrations between 3.0 and 55.0 nmol/l. Standards and samples were analysed in duplicate and the mean of the two runs was considered.
2.4. Statistical analysis The results were analysed by: (i) linear regression analysis (reference method: x values, assay tested: y values); (ii) the method of Bland and Altman (P level measured by GC/MS −P level measured by assay tested) vs (sum of P level measured by (GC/MS +assay tested)/2) [16]; (iii) the ratios method (result from GC/ MS/result from assay tested) expressed in the form of histograms (frequency (%) vs ratio) [17]. Linear regression analysis (slope, intercept, correlation coefficient (r) and residual SD of the distribution around the regression line) was performed using Statview SE Software.
3. Results The serum P levels of six of the samples were below the detection limit for the GC/MS method ( B 0.3 nmol/l) and were excluded from the overall analysis. Thus, comparisons were performed using the remaining 93 serum samples, although it was not possible to obtain a result for all of the samples in all of the assays. Indeed, some serum samples had P levels below the
99
detection limits of particular tests: Elecsys (0.48 nmol/l) (n= 21), Vidas (0.32 nmol/l) (n = 3), AIA-600 (0.32 nmol/l) (n= 3), Vitros Eci (0.25nmol/l) (n= 5). The results of the linear regression analysis are shown in Table 1. Most of the immunoassays displayed a slopeB 1 [from 0.667 (Access) to 1.034 (AIA-600)] and a greater variety of intercepts [from − 0.098 (Magia) to 3.506 nmol/l (Access)] than GC/MS. All the immunoassays had r values above 0.9, and standard deviation values (SD) were from 1.64 (Axsym) to 5.816 (AIA-600). P levels were widely distributed and large differences in the intrinsic specificity of the various P assays was noted with respect to GC/MS. The results for Bland and Altman analysis are presented in Fig. 1. Large differences in P concentration were observed between non-isotopic assays and GC/ MS. The mean differences between these non-isotopic assays and GC/MS were significantly different from zero and tended to be either positive (Immuno-1) or negative (Axsym, Vidas, Elecsys, Magia). The assays from Axsym, Vidas, Autodelfia, Amerlite, Access, Immulite, Vitros ECi and Magia underestimated P concentration at high levels (up to 10 nmol/l). In contrast, assays from Immulite, ACS-180, Immuno-1, Access, Vitros ECi, Amerlite, and Autodelfia overestimated P concentration at low levels (from 0.3 to 10.0 nmol/l). Assays from Elecsys and AIA-600 showed extensive scattering of the data over the entire concentration range, with overestimation at high concentrations. All except Axsym showed a high percentage of outliers if P levels below 5 nmol/l were excluded. Ratios (P level in each assay vs P level in GC/MS) were classified into six arbitrary groups (R B0.50, 0.505 RB 0.85, 0.855 RB 1.15, 1.155 RB2.00, 2.005 RB 5.00, R \ 5). The frequency (%) and distribution of the ratios (R) are shown in Fig. 2. A broad distribution of R was observed for each of the immunoassays. The extreme values of the ratios were as
Table 1 Comparison of 12 non-isotopic progesterone immunoassays with GC/MS by linear regression analysis Progesterone assay
n
Slope
Intercept (nmol/l)
Residual SD (nmol/l)
Correlation coefficient (r)
Abbott: Axsym Bayer Diagnostics: Immuno.1 Bio Me´ rieux: Vidas Chiron Diagnostics: ACS-180 Diagnostics Pasteur: Access Diagnostics Prod Corp: Immulite EG&G instruments: Auto-Delfia Eurogenetics: AIA-600 Merck Clevenot-Biotrol: Magia Ortho Clin Diagnostics: Amerlite Ortho Clin Diagnostics: Vitros Eci Roche Boehringer: Elecsys
93 93 90a 93 93 93 93 90a 93 93 88a 72a
0.816 0.935 0.841 0.928 0.667 0.790 0.768 1.034 0.733 0.781 0.707 0.853
1.015 3.202 0.276 1.528 3.506 2.690 2.180 0.264 −0.098 3.017 2.550 0.142
1.640 4.051 2.799 2.485 3.411 3.972 2.610 5.816 5.421 4.148 4.533 3.401
0.993 0.970 0.982 0.988 0.958 0.960 0.981 0.951 0.918 0.955 0.939 0.977
a Sample levels below the detection limit defined by the manufacturers were excluded from the analysis [Elecsys (n = 21), Vitros ECi (n= 5), Vidas (n =3), AIA-600 (n= 3)].
100
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
Fig. 1.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
101
Fig. 1. Plots representing difference in P concentrations (levels measured by the assay tested minus levels measured by GC/MS) versus mean P concentrations (levels measured for each sample by the two methods/2).
follows: Immulite (0.47– 39.64), Immuno-1 (0.78– 21.46), Vitros ECi (0.40– 10.51), Autodelfia (0.60– 14.78), Amerlite (0.55– 12.00), AIA-600 (0.38– 3.41), Elecsys (0.32–3.17), Axsym (0.54– 5.54) respectively. Thirty percent of the values obtained with Elecsys, ACS-180, AIA-600 and Axsym were between 0.85 and 1.15. Immunoassays from Immulite, Amerlite and Access resulted in R values of which more than 50% were \2. The frequency of the highest R values (\ 5) was important for Autodelfia, Vitros ECi, ACS-180, Access, Immulite, Amerlite and Immuno-1 and ranged from 6.45% (Autodelfia) to 21.50% (Immuno-1).
4. Discussion A single serum P determination has been shown to be useful in clinical assessment of the risk of early pregnancy complications [1]. Retrospective and prospective studies have defined cut-off levels of P for the reliable identification of patients at high risk of ectopic pregnancy [3] and for the prediction of pregnancy rates in in vitro fertilisation and embryo transfer [5– 7]. This study focused on differences in the results obtained with 12 P
immunoassays tested in nine laboratories and with GC/ MS. Serum P assays were performed with unextracted samples. We found that most of the immunoassays underestimated serum P concentration at high levels and overestimated P concentration at low levels. There were great discrepancies between the results obtained with immunoassays and GC/MS; these depended on the immunoassay. Thus, a wide range of P ratios was obtained for all immunoassays. Some displayed high levels (ie\ 5) of ratios (R) with a high frequency (from 10% to 21.5%). The AIA-600 and Elecsys assays also resulted in extensive scattering of the data for the entire concentration range. There are several factors that might explain these discrepancies. First, these discrepancies may be due to the presence of steroids structurally similar to P (eg 5a/5b pregnane 3-20 dione, allopregnanolone, pregnanolone) which highly cross-reacted with P [18–21]. Some manufacturers had included highly specific polyclonal (Access, AIA-600, Amerlite, Autodelfia [22], Elecsys, Immulite, Vitros ECi) or monoclonal antibodies (ACS-180, Axsym [23], Immuno-1 [24], Magia, Vidas) in their immunoassays, but this did not improve specificity. Second, they may be due to the presence of binding proteins. Inclusion of blocking
102
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
Fig. 2.
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
103
n
result from the assay tested vs. frequency (%) the distribution of the P concentration obtained by the two result from GC/MS methods. R were arbitrarely classified as follows: R B0.50, 0.50 =R B 0.85, 0.85 =R B 1.15, 1.15 = RB2.00, 2.00 = R B5.00, R\ 5. Fig. 2. Histograms
ratio: R:
agents (eg danazol (ACS-180, Autodelfia, Elecsys), bovine serum albumin (Axsym), products of unknown origin (Amerlite, Vitros ECi)) may prevent interference from binding proteins [25]; but there was no difference in results between immunoassays including blocking agents and those without blocking agents (eg Access, AIA-600, Immulite, Immuno-1, Magia, Vidas). None gave results very similar to those obtained by GC/MS. Third, they may be due to the choice of tracer. This parameter has been shown to be of critical importance in the development of a reliable immunoassay. Immunoassays using alkaline phosphatase (AP) gave higher P levels than those using horseradish peroxidase (HRP) or europium when only the signal generator was modified [26]. In our study, we found no clear difference in serum P levels over the entire concentration range, between AP-based immunoassays (e.g. Access, AIA-600, Axsym, Immulite, Immuno-1, Magia, Vidas) and HRP-based immunoassays (Ammerlite, Vitros ECi) (data not
shown). Fourth, other factors such as the nature of the assay buffer (composition, ionic strength, pH) and the purity of the preparation and control of calibrators might intervene. Finally, the approach for standardising calibrators by checking the standard calibration curve (e.g. Amerlite, ACS-180 and Elecsys) by GC/MS did not seem to be a criterion for reliability. Overall, our study showed large differences between non-isotopic P immunoassays and GC/MS for individual human serum samples. None of these assays gave results very similar to those obtained by GC/MS over the entire range of P concentrations tested. Improvments in calibration procedures seem to be necessary. The specificity/sensitivity criteria also require reassessment. None of these immunoassays seemed to have the high sensitivity and reliability required for screening for abnormal early pregnancy in asymptomatic women [2] or for monitoring in vitro fertilisation/embryo-transfer protocols [6,27].
104
P. Boudou et al. / Journal of Steroid Biochemistry & Molecular Biology 78 (2001) 97–104
References [1] J.D. Issacs, N.S. Whitworth, B.D. Cowan, Relative operating characteristic analysis in reproductive medicine: comparison of progesterone and human chorionic gonadotropin doubling time as predictors of early gestational normalcy, Fertil. Steril. 62 (1994) 452 – 455. [2] C.A. Daily, S.L. Laurent, W.C. Nunley, The pronostic value of serum progesterone and quantitative b-human chorionic gonadotropin in early human pregnancy, Am. J. Obstet. Gynecol. 171 (1994) 380 –384. [3] M.L. McCord, D.A. Muran, J.E. Buster, K.L. Arheart, T.G. Stovall, S.A. Carson, Single serum progesterone as a screen for ectopic pregnancy exchanging specificity and sensibility to obtain optimal test performance, Fertil. Steril. 66 (1996) 513 – 516. [4] M.X. Ransom, A.J. Garcia, M. Bohrer, G.H. Corsan, E. Kemmann, Serum progesterone as predictor of methotrexate success in the treatment of ectopic pregnancy, Obstet. Gynecol. 83 (1994) 1033 – 1037. [5] W. Schoolcraft, E. Sinton, T. Schlenker, D. Huynh, F. Hamilton, D.R. Meldrum, Lower pregnancy rate with premature luteinization during pituitary suppression with leuprolide acetate, Fertil. Steril. 55 (1991) 563 –566. [6] K.M. Silverberg, W.N. Burns, D.L. Olive, R.M. Riehl, R.S. Schenken, Serum progesterone levels predict success of in vitro fertilization/embryo transfer in patients stimulated with leuprolide acetate and menopausal gonadotropins, J. Clin. Endocrinol. Metab. 73 (1991) 797 –803. [7] R. Fanchin, D. de Ziegler, J. Taieb, A. Hazout, R. Frydman, Premature elevation of plasma progesterone alters pregnancy rates of in vitro fertilization and embryo transfer, Fertil. Steril. 59 (1993) 1090 – 1094. [8] R.H. Buck, S.M. Joubert, R.J.Norman, Serum progesterone in the diagnosis of ectopic pregnancy: a valuable diagnostic test? Fertil. Steril. 50 (1988) 752 –755. [9] B.D. Cowan, D.T. Vandermolen, C.A. Long, N.S. Whitworth, Receiver-operator characteristic, efficiency analysis, and predictive value of serum progesterone concentration as a test for abnormal gestations, Am. J. Obstet. Gynecol. 166 (1992) 1729 – 1734. [10] C.M. Peterson, D. Kreber, P. Delgado, T.T. Hung, Laboratory and clinical comparison of a rapid versus a classic progesterone radioimmunoassay for use in determining abnormal and ectopic pregnancies, Am. J. Obstet. Gynecol. 166 (1992) 562 – 566. [11] B.G. Wolthers, G.P.B. Kraan, Clinical applications of gas chromatography and gas chromatography – mass spectrometry of steroids, J. Chromatogr. 843 (1999) 247 –274. [12] M.C. Patricot, B. Mathian, S. Serpentie, A. Revol, Determination of urinary estradiol using a enzymatic method during the menstrual cycle. Comparison with a method using isotope dilution – mass spectrometry (ID-MS), Clin. Chem. Acta 158 (1986) 139 – 146. [13] L. Dehennin, Estradiol-17 beta determined in plasma by gas chromatography – mass spectrometry with selected ion monitoring of mixed silyl ether- perfluoroacyl ester derivatives and use of various stable-isotope-labeled internal standards, Clin. Chem. 35 (1989) 532 – 536.
[14] L.M. Thienpont, V.I. De Brabandere, D. Stockl, A.P. De Leenheer, Use of cyclodextrins for prepurification of progesterone and testosterone from human serum prior to determination with isotope dilution gas chromatography/mass spectrometry, Anal. Chem. 66 (1994) 4116 – 4119. [15] L. Siekman, Isotope dilution mass spectrometriy of steroid hormones — a definitive method in clinical chemistry, in: A.P. De Leenheer, R. Roncucci, C. Van Peterghen (Eds.), Quantitative Mass Spectrometry in Life Science, vol. 2, Elsevier, Amersham, 1978, pp. 3 – 16. [16] J.M. Bland, D.G. Altmanm, Statistical method for assessing agreement between two methods of clinical measurement, The Lancet 8 (1986) 307 – 310. [17] K. Nahoul, L. Dehennin, R. Scholler, Radioimmunoassay of plasma progesterone after oral administration of micronized progesterone, J. Steroid Biochem. 26 (1987) 241 – 249. [18] T. Ba¨ ckstro¨ m, A. Andersson, D.T. Baird, G. Selstam, The human corpus luteum secretes 5a-pregnane-3,20-dione, Acta Endocrinologica 111 (1986) 116 – 121. [19] P.J. Schmidt, R.H. Purdy, P.H. Moore Jr, S.M. Paul, D.R. Rubinow, Circulating levels of anxiolytic steroids in the luteal phase in women with premenstrual syndrome and in control subjects, J. Clin. Endocrinol. Metab. 79 (1994) 1256 – 1260. [20] M. Wang, L. Seippel, R.H. Purdy, T. Ba¨ ckstro¨ m, Relationship between symptom severity and steroid variation in women with premenstrual syndrome: study on serum pregnenolone, pregnenolone sulfate, 5a-pregnane-3,20-dione and 3a-hydroxy-5apregnan-20-one, J. Clin. Endocrinol. Metab. 81 (1996) 1076 – 1082. [21] A.R. Genazzani, F. Petraglia, F. Bernardi, E. Casarosa, C. Salvestroni, A. Tonetti, A. R.E. Nappi, S. Luisi, M. Palumbo, R.H. Purdy, M. Luisi, Circulating levels of allopregnanolone in human: gender, age, and endocrine influences, J. Clin. Endocrinol. Metab. 83 (1998) 2099 – 2103. [22] T. Lo¨ vgren, L. Merio¨ , K. Mitrunen, M.L. Ma¨ kinen, M. Ma¨ kela¨ , K. Blomberg, T. Palenius, K. Pettersson, One-step all-in-one dry reagent immunoassays with fluorescent europium chelate label and time-resolved fluorometry, Clin. Chem. 42 (1996) 1196 –1201. [23] D.H. Wilson, W. Groskopf, S. Hsu, D. Caplan, T. Langner, M. Baumann, D. DeManno, G. Williams, D. Payette, C. Dacel, D. Lynch, G. Manderino, Rapid, automated assay for progesterone on the Abbott AxSYMTM analyzer, Clin. Chem. 44 (1998) 86 – 91. [24] A. Levesque, M. Letellier, P. Dillon, A. Grandt, Analytical performance of Bayer Immuno I™ estradiol and progesterone assays, Clin. Chem. 43 (1997) 1601 – 1609. [25] J.V. Micallef, M.M. Hayes, A. Latif, R. Ahsan, S.B. Sufi, Serum binding of steroid tracers and its possible effects on direct steroid immunoassay, Ann. Clin. Biochem. 32 (1995) 566 – 574. [26] B.S. Prakash, H.H.D. Meyer, E. Schallenberger, D.F.M. Van de Wiel, Development of a sensitive enzyme immunoassay (EIA) for progesterone determination in unextracted bovine plasma using the second antibody technique, J. Steroid Biochem. 28 (1987) 623 – 627. [27] S.D. Prien, M.S. Canez, R.H. Messer, Increases in progesterone after human chorionic gonadotropin administration may predict cycle outcome in patients undergoing in vitro fertilization-embryo transfer, Fertil. Steril. 62 (1994) 1066 – 1068.