Assessment of theca cell function prior to controlled ovarian stimulation: the predictive value of serum basal/stimulated steroid levels

Transcription

1 Human Reproduction, Vol.25, No.1 pp , 2010 Advanced Access publication on November 6, 2009 doi: /humrep/dep378 REVIEW Reproductive endocrinology Assessment of theca cell function prior to controlled ovarian stimulation: the predictive value of serum basal/stimulated steroid levels Jean-Noël Hugues 1,4, Lucie Theron-Gerard 1, Christiane Coussieu 2, Maud Pasquier 1, Didier Dewailly 3, and Isabelle Cedrin-Durnerin 1 1 Unit of Reproductive Medicine, Department of Obstetrics-Gynaecology, Hôpital Jean Verdier, Assistance Publique - Hôpitaux de Paris, Université Paris XIII, Bondy, France 2 Laboratory of Biochemistry, Hôpital Pitié Salpétrière, Assistance Publique - Hôpitaux de Paris, Université Paris VII, Paris, France 3 Department of Endocrine, Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Université Lille, Lille, France 4 Correspondence address. Tel: þ ; Fax: þ ; jean-noel.hugues@jvr.aphp.fr background: Serum androgen levels correlate with ovarian sensitivity to follicle-stimulating hormone (FSH) but in practice, standard baseline serum testosterone (T) levels prior to in-vitro fertilization (IVF) may not be the most appropriate marker for determination. methods: Infertile women enrolled in an IVF programme were included in this study. Serum T and D4-androstenedione (A), and the androgen precursor 17-hydroxyprogesterone (17-OHP) were measured before and 24 h after a gonadotrophin-releasing hormone agonist stimulation test (GAST). An early follicular phase antral follicle count (AFC) was also performed. Patients were subsequently enrolled in a long gonadotrophin-releasing hormone agonist protocol with a standard FSH dose (150 IU) for 7 days to assess the association between androgen levels and ovarian responsiveness to FSH. results: The GAST elicited a significant increase in serum androgen levels that was well correlated with AFC. 17-OHP showed the greatest response to GAST and strongest correlation with AFC. The 17-OHP response to GAST differentiated patients with high ovarian reserve (OR) from those with low or normal OR as assessed by AFC, whereas only the estradiol response could differentiate those with low AFC. GAST-stimulated serum levels of 17-OHP were also correlated with ovarian response to FSH. Using receiver operating characteristic curve analysis, stimulated 17-OHP levels were predictive of the ovarian response to controlled ovarian stimulation, with similar power to that observed with AFC but lower power than with anti-müllerian hormone. conclusions: Serum androgen levels following GAST are correlated with AFC and ovarian response to FSH. Serum T is a less sensitive marker of theca cell function than 17-OHP. Key words: theca cell function / antral follicle count / gonadotrophin-releasing hormone agonist / in-vitro fertilization Introduction Assessment of serum androgen levels prior to controlled ovarian stimulation might be useful to predict the ovarian response and, thus, to adjust the starting dose of exogenous gonadotrophins. Indeed, there is some evidence that ovarian sensitivity to folliclestimulating hormone (FSH) is closely related to androgen production. For instance, in women with polycystic ovary syndrome (PCOS), high androgen levels are closely related to FSH hypersensitivity. This relationship likely results from the effect of androgens on follicle growth and number (Zawadzki and Dunaif, 1992; Jonard et al., 2003; Rotterdam Consensus, 2004; Azziz et al., 2009). Similarly, serum androgen levels in normo-ovulatory women are correlated with the ovarian response to FSH as assessed by peak estradiol (E2) levels and the number of mature follicles, oocytes and embryos (Barbieri et al., 2005). It has also been suggested that basal serum androgen levels can predict the ovarian response to FSH (Frattarelli and Peterson, 2004; Frattarelli and Gerber, 2006). & The Author Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 Theca cell function and ovarian sensitivity to FSH 229 Finally ovarian ageing is characterized by a reduced ovarian sensitivity to FSH, which is accompanied by a progressive decline in serum androgen levels (Davison et al., 2005). Indeed, even in women with early ovarian ageing and a preserved follicular pool, the ovarian capacity to secrete androgens is decreased under both basal and human chorionic gonadotrophin-stimulated conditions (Piltonen et al., 2003). These data highlight the close association between androgen production and ovarian sensitivity to FSH. They are in line with experiments in monkeys which demonstrated that androgens act synergistically with FSH to promote granulosa cell proliferation and reduce the rate of cellular apoptosis (Vendola et al., 1998; Weil et al., 1999). They also indicate that evaluation of theca cell function should be performed in addition to measurement of granulosa cell markers. However, there is no consensual agreement on which serum androgen is able to best predict the ovarian response to FSH. Indeed, the results of serum androgen assays are both imprecise and inconsistent (Barth et al., 2007). Measurement of serum total testosterone (T) is usually recommended during clinical work-up but this practice should be questioned (Barth et al., 2007). The inaccuracy of the measurement of low serum T levels is particularly marked among older women (Miller et al., 2004). For greatest accuracy, T assays require extraction and purification from serum, but these steps are usually omitted with the use of commercial kits, which showed substantial variability in outcomes (Boots et al., 1998). Furthermore, the normal range of female serum hormone levels is usually designated by commercial kit manufacturers after hormonal measurement among a limited number of unselected women. The described shortcomings of serum T assays dictate the need for consideration of other serum androgens and/or stimulation tests to better assess theca cell function, specifically for women whose serum T levels are at the lower end of the normal range. Few data have been published on the ability of serum androgens to quantify theca function. The most documented situation is PCOS, in which stimulation of theca cells by endogenous luteinizing hormone (LH) during a gonadotrophin-releasing hormone agonist stimulation test (GAST) induces a more rapid increase in serum levels of D4-androstenedione (A) and the androgen precursor 17-hydroxyprogesterone (17-OHP) than in T (Barnes et al., 1989; Rosenfield et al., 1993, 1994; Pasquali et al., 2007). Thus, 17-OHP and A may represent better markers of theca cell function than T. With these assays, the GAST has been able to reveal occult biochemical ovarian hyperandrogenism in normo-ovulatory women (Chang et al., 2000). On the other hand, to our knowledge, there is no data showing that the GAST is able to detect theca cell insufficiency in women with low ovarian reserve (OR). Therefore, the aim of this study was to explore the relationship between the theca cell function as assessed by serum steroid response to the GAST and the antral follicle count (AFC), which is a valid marker of OR (Broekmans et al., 2006). In addition, we compared the predictive value of these factors for the ovarian response to FSH. Materials and Methods Patients and treatment protocols Ninety-one unselected women participating in an in-vitro fertilization (IVF) programme were included in this prospective, single centre study. Written informed consent was obtained from all participants. On Day 3 of a spontaneous menstrual cycle, a blood sample to evaluate baseline hormone levels was taken in the morning, and a transvaginal ultrasound (TVUS) scan performed. A GAST was then performed, as part of the endocrinological work-up for IVF. This involved administration of a subcutaneous injection of 0.1 mg triptorelin (Decapeptyl w ; Ipsen Biotech, France) and collection of a second blood sample after 24 h. Blood samples were frozen and stored until testing. All samples were tested using the same assay for each hormone (as described below). Ultrasound scans TVUS scans were performed using a vaginal probe. The ovarian volume and size of all follicles of 2 9 mm in diameter (the mean of two measurements) were recorded. All scans were performed by the same physician (CD). The total number of follicles of 2 9 mm observed in both ovaries on TVUS on Day 3 was used to classify patients into one of three groups based on recommendations issued by the Rotterdam Consensus (2004): Group 1:,10 follicles (low AFC, n ¼ 21); Group 2: follicles (normal AFC, n ¼ 55); Group 3:.23 follicles (high AFC, n ¼ 15). Patients in Group 1 had high plasma FSH levels on Day 3 (FSH. 10 mui/l), whereas those in Group 2 had normal FSH levels. In addition, Group 3 consisted of patients with PCOS criteria according to the Rotterdam Consensus (2004). Ovarian response to FSH Later in the same cycle, 70 of the 91 women received a long luteal phase gonadotrophin-releasing hormone (GnRH) agonist protocol. From Day 22 to 25, a single daily dose of triptorelin 0.1 mg was administered by subcutaneous injection. Once pituitary desensitization was achieved, a fixed daily dose (150 IU) of recombinant human (r-h) FSH (GONAL-f w ; Merck Serono S.A. Geneva, Switzerland) was administered for 7 days. Following this stimulation period with a fixed dose of FSH, the ovarian sensitivity to FSH was assessed on Day 8 by measurement of the serum E2 level and number of follicles of.10 and.14 mm in mean diameter. Finally, patients were classified into two groups according to the number of follicles.17 mm on the day of human chorionic gonadotrophin (hcg) administration: patients with 2 mature follicles were considered to be low responders to FSH and those with 3 mature follicles were considered to be normal responders. An assessment of predictive factors was performed by receiver operating characteristic (ROC) curve analysis. Hormonal assays Serum androgens were measured using highly specific assays. Serum total T levels were measured using a radio-immunometric technique (Spectria Orion Diagnostics, Finland). The sensitivity of the assay was 0.1 ng/ml, with intra- and inter-assay coefficients of variation of and 4.8 7%, respectively. Free T levels and the free androgen index were calculated after measurement of serum albumin and sex hormone binding globulin (Vermeulen et al., 1999). Measurement of serum extracted A levels were carried out by radio-immunoassay (Immunotech Beckman- Coulter, France): the sensitivity of the assay was 0.1 ng/ml with intraand inter-assay coefficients of variation were below or equal to 8.1 and 11.9%, respectively. Serum levels of 17-OHP were measured using a radio-immunometric technique (17OH-RIA-CT KIP1409 BioSource Europe, Nivelles, Belgium). The detection limit of the assay was 0.02 ng/ml, with intra- and inter-assay coefficients of variation of and %, respectively. Anti-Müllerian hormone (AMH) concentrations were assessed using a highly specific enzyme-linked immunosorbent assay (ELISA DSL , DSL, France). The sensitivity of this assay was ng/ml, with intra- and inter-assay coefficients of variation of 5 and 8%, respectively. E2 measurements were performed using a

3 230 Hugues et al. Table I Serum steroid levels before and 24 h after a GAST (n 5 91) T0 (n 5 91) T24 (n 5 91) Increase (%) Significance (P-value) OHP, ng/ml , T, ng/ml , A, ng/ml 1.6 ( ) 1.73 (1 2.89) E2, pg/ml 33 ( ) 116 (53 253) 251, Results are expressed as mean + standard deviation or median (10th 90th percentile). 17-OHP, 17-hydroxyprogesterone; A, D4-androstenedione; E2, estradiol; T, testosterone. radio-immunometric technique (Estradiol-2 Clinical Assays, DiaSorin). The detection limit of the assay was 2 pg/ml, with intra- and inter-assay coefficients of variation of and 4.6%, respectively. Statistical analysis Statistical analysis was performed using StatView (Abacus Concepts, Berkeley, USA). Normal distribution was tested using the Kolmogorov Smirnov test. Results are expressed as mean + standard deviation for normally distributed variables and as median (10th 90th percentile) for those not normally distributed. Normal variables were compared using the Student t-test or ANOVA as appropriate. ANOVA was followed by post hoc analysis using the Bonferroni correction for comparison of three groups. Paired analyses were performed for repeated measures. Nonparametric tests (Mann Whitney or Wilcoxon tests as appropriate) were used to analyse variables that were not normally distributed. Simple regression analyses were performed to correlate baseline/gaststimulated serum steroid levels to the AFC or ovarian response to FSH on Day 8 of ovarian stimulation. Every significant correlation was controlled by the others through stepwise multiple regression analyses when the independent variable was continuous and through ordinal logistic regression when it was ordinal (follicle number). P-values of,0.05 were considered significant. ROC curves were constructed to examine the diagnostic test performance, i.e. its capacity to discriminate between normal and poor responders. Sensitivity against (1-specificity) was plotted at each threshold level and the area under the curve (AUC) was computed by the non-parametric Wilcoxon test. The AUC (mean, 95% Confidence Interval) represents the probability of correctly identifying normal and poor responders. A value of 0.5 means that the test result is no better than chance. Results Ultrasound and hormonal assessment on Day 3 4 As shown in Table I, the GAST resulted in a significant increase in all mean serum steroid levels within the first 24 h. The largest increase was observed in E2 levels and the greatest increase in serum theca cell steroids was recorded for 17-OHP (Table I). In addition, significant positive correlations were found between the AFC and baseline or stimulated steroid values, except for baseline E2 (Table II). After stepwise multiple regression analysis, only stimulated E2, A and 17-OHP were retained in a significant predictive model for AFC (P, ). Analysis of the hormone tests according to AFC Analysis of baseline and stimulated steroid serum concentrations according to the AFC classification (see Materials and Methods section) is presented in Table III. By ANOVA, no baseline serum steroid mean level was significantly different between the three groups (Group 1: low AFC; Group 2: normal AFC and Group 3: excessive AFC). In contrast, all GAST-stimulated serum mean levels were significantly different. Post hoc analysis indicated, however, that these global differences were due to differences between Group 3 versus Group 1 or 2. Significance level was not reached in the comparisons between these Groups 1 and 2, except for GAST-stimulated E2 (Table III). Analysis of the ovarian response to FSH according to the hormonal test Ovarian sensitivity to FSH Data on ovarian responses to FSH are presented in Table IV. By univariate analysis, no significant correlation was found between baseline and stimulated serum T levels and ovarian response on Day 8. In contrast, baseline and stimulated serum 17-OHP and A levels were significantly correlated with the ovarian response as assessed by plasma E2 and the number of follicles of.10 or.14 mm in diameter on Day 8. Similarly, AFC and serum AMH values correlated strongly Table II Correlation between the AFC and serum steroid levels at baseline and after a GAST (n 5 91) Time r (correlation Significance coefficient) (P-value) OHP, ng/ml T T24* 0.587, T, ng/ml T T , A, ng/ml T T24* 0.482, E2, pg/ml T NS (P ¼ 0.9) T24* 0.604, OHP, 17-hydroxyprogesterone; A, D4-androstenedione; E2, estradiol; NS, not significant; T, testosterone. *Variable remaining significant (P, 0.05) after stepwise multiple regression analysis.

4 Theca cell function and ovarian sensitivity to FSH 231 Table III Serum steroid levels at baseline and after a GAST for three groups classified according to the AFC P after post hoc analysis Group 2 versus Group 3 P after post hoc analysis Group 1 versus Group 3 P after post hoc analysis Group 1 versus Group 2 p by ANOVA Group 3 high AFC (n 5 15) Group 2 normal AFC (n 5 55) Group 1 low AFC (n 5 21) OHP, ng/ml T NA NA NA T T, ng/ml T NA NA NA T A, ng/ml T0 1.4 ( ) 1.7 (1 2.4) 1.7 ( ) NA NA NA T ( ) 1.74 ( ) 2.11 ( ) E2, pg/ml T0 31 (19 49) 33 (20 52) 34 (21 87) NA NA NA T24 70 (39 93) 119 (82 217) 234 (73 352) Results are expressed as mean + standard deviation of the mean or as median (10th 90th percentile) for non-parametric values. 17-OHP, 17-hydroxyprogesterone; A, D4-androstenedione; E2, estradiol; NS, not significant; T, testosterone. NA: not applicable. with these markers of ovarian response while baseline FSH levels correlated poorly (Table IV). After ordinal logistic regression analysis entering the thecal steroids as dependent variables, only poststimulated 17-OHP levels correlated significantly to the number of follicles.10 mm and no parameter correlated to the number of follicles.14 mm although there was a trend for post-stimulated 17-OHP levels (P ¼ 0.077). After stepwise multiple regression analysis entering the theca steroids as dependent variables, only poststimulated 17-OHP was included in a significant predictive model for E2 on Day 8 (P, ). Prediction of the ovarian response to FSH Of the 70 women who proceeded to triggering of ovulation, 50 had a normal response to FSH while 20 were considered as low responders according to the number of mature follicles at the time of hcg administration (see Materials and Methods section). A comparative analysis of clinical characteristics, hormonal and TVUS data between normal and low responders showed that they actually differed by age [median and (5th 95th percentiles): 30.0 (20 40) versus 35.0 (20 38) years, respectively, P ¼ 0.042], AFC [17.0 (7 40) versus 12.0 (5 40), respectively, P ¼ 0.014], serum AMH levels [1.98 ( ) versus 0.83 (0 6.1) ng/ml, respectively, P, 0.001] and stimulated levels of 17-OHP [1.80 ( ) versus 1.40 ( ) ng/ml, respectively, P ¼ 0.04). ROC curve analysis was then performed to analyse the ability of each parameter for predicting low response. Patients with high AFC (n ¼ 11) were therefore excluded from this analysis. As shown in Table V, significant AUCs were obtained only for AMH, AFC and stimulated 17-OHP and E2. Interestingly, stimulated 17-OHP yielded an AUC very close to the one of AFC and slightly lower than the one of AMH. The best compromise between sensitivity (60%) and specificity (82%) was obtained with a threshold of stimulated 17-OHP at 1.55 ng/ml. Discussion Our data show that serum theca steroid levels following stimulation by a GAST are well correlated with the AFC and with the ovarian sensitivity to FSH. This supports the concept of a close relationship between theca cell secretion and folliculogenesis. They indicate also that assessment of theca cell function under stimulated conditions is a valuable tool to identify patients at risk of a low response to FSH. For that purpose, measurement of serum 17-OHP and A are more informative than T. As far as we know, the usefulness of androgen measurement in clinical work-up for assisted reproductive technology (ART) has never been documented while evaluation of conventional granulosa cell markers has been extensively studied (Broekmans et al., 2006). Besides, several methodological shortcomings related to the measurement of female serum androgens have been reported. Firstly, the use of commercial kits without serum extraction of androgens carries risks of a misdiagnosis of hyperandrogenism (Azziz, 2003) or androgen deficiency (Davison et al., 2005). Substantial variability in total serum T levels detected by commercially available kits has been reported, and this accounts largely for the limited utility of these assays in clinical practice (Boots et al., 1998). Furthermore, many variables may also alter serum T levels (Barbieri et al., 2005; Barth et al., 2007).

5 232 Hugues et al. Table IV The ovarian response to FSH at Day 8 of stimulation and serum steroid levels at baseline and after a GAST (n 5 70) Follicles >10 mm Follicles >14 mm E OHP, ng/ml T0 r ¼ P ¼ r ¼ NS (P ¼ 0.06) r ¼ P ¼ T24 r ¼ P, r ¼ P, r ¼ P, * T, ng/ml T0 r ¼ NS r ¼ NS r ¼ NS T24 r ¼ NS r ¼ 0.04 NS r ¼ NS A, ng/ml T0 r ¼ P ¼ r ¼ P ¼ 0.03 r ¼ 0,440 P ¼ T24 r ¼ P, r ¼ P ¼ r ¼ P ¼ E2, pg/ml T0 r ¼ NS r ¼ NS r ¼ NS T24 r ¼ P ¼ r ¼ P ¼ r ¼ P ¼ FSH T0 r ¼ P ¼ r ¼ NS r ¼ NS (P ¼ 0.056) AMH T0 r ¼ P, * r ¼ P ¼ r ¼ P, * AFC T0 r ¼ P, r ¼ P ¼ r ¼ P, The ovarian response was assessed by determination of the number of follicles of.10 mm and/or.14 mm and/or E2 levels on Day 8 of FSH stimulation. Significance was assumed if P, OHP, 17-hydroxyprogesterone; A, D4-androstenedione; AFC, antral follicle count; AMH, anti-müllerian hormone; E2, estradiol; NS, not significant; T, testosterone. *Variable remaining significant (P, 0.05) after stepwise multiple regression analysis. The use of stimulation tests has been suggested to avoid the problems related to the measurement of basal serum T and to improve the diagnostic value of androgen assessment. Stimulation with a GnRH-agonist induces the release of endogenous LH and increases production and secretion of theca cell steroids usually within the first 24 h of exposure (Barnes et al., 1989). In women with PCOS, serum 17-OHP levels appeared to be the most sensitive to the GAST (Barnes et al., 1989; Rosenfield et al., 1993, 1994). Similar findings were obtained in the present study in unselected infertile women, for which we observed that administration of a GnRH agonist elicited a more marked rise in serum 17-OHP levels than in T or A levels. This observation is consistent with the physiological concept that LH exerts positive control on the 17 hydroxylase enzyme activity of the ovarian P450c17P (Sahin and Kelestimur, 1997). The data are also in accordance with another report that an intravenous infusion of r-hlh has more rapid effects on the early steps of steroidogenesis (reflected by changes in serum 17-OHP levels) than on later steps (reflected by changes in serum A and T levels) (McCartney et al., 2004). Conversely, ovarian T production, which is primarily dependent on 17b hydroxysteroid dehydrogenase activity type 5, is less dependent on LH control (Nelson et al., 2001). Additionally, our data provide evidence that, in an unselected infertile population, the theca steroid response to GAST is highly correlated with the AFC, with the strongest correlation being observed for 17-OHP. This observation concurs with a previous report (Jonard et al., 2003), and supports the notion that androgens act positively on granulosa cell proliferation and follicular recruitment, as previously shown in monkeys (Vendola et al., 1998; Weil et al., 1998). Indeed, androgens exert paracrine effects via specific granulosa cell receptors (Horie et al., 1992; Hillier et al., 1997). Through this mechanism, they control the expression of FSH receptors as shown in humans (Suikkari et al., 1995) and in monkeys (Weil et al., 1999). In addition, among theca steroids, stimulated serum 17-OHP levels correlated the best with the markers of ovarian responsiveness to exogenous FSH at Day 8 and, according to ROC analysis, they were able to predict a low response to FSH with a power very close to that observed with AFC and slightly lower than with AMH. Therefore, the GAST-stimulated 17-OHP may be more useful than baseline serum T in distinguishing between patients with a normal or a low OR. However, whether the GAST adds information to AFC and AMH for predicting the ovarian response to FSH needs further verification by prospective studies. Finally, our results might explain why the addition of exogenous LH to stimulation protocols could be beneficial for women with reduced theca cell secretion. This population may include women aged over 35 years (Marrs et al., 2003; Humaidan et al., 2004) or those with a poor ovarian response to FSH related to strong pituitary desensitization Table V Areas under the ROC curves (AUC) evaluating the prediction of low response to COH in women with low or normal AFC (n 5 59) AUC Significance 95% Confidence Interval... AMH AFC Basal 17-OHP Stimulated 17-OHP Basal A Stimulated A Basal T Stimulated T Basal E Stimulated E OHP, 17-hydroxyprogesterone; A, D4-androstenedione; AFC, antral follicle count; AMH, anti-müllerian hormone; E2, estradiol; T, testosterone.

6 Theca cell function and ovarian sensitivity to FSH 233 following administration of a depot formulation of GnRH agonist (Ferraretti et al., 2004; De Placido et al., 2005; Ruvolo et al., 2007). Indeed, it is feasible that intra-ovarian secretion of steroid by theca cells in response to LH contributes to follicular development. Therefore, one could speculate that, among women with a low AFC, only those with a positive GAST response would benefit from exogenous LH supplementation. In conclusion, this study shows that measurement of 17-OHP following a GAST may be considered to be a valuable, albeit indirect, method for assessing ovarian function. A better understanding of the association between theca cell production and folliculogenesis could assist in identification of women who would benefit from LH supplementation of ART cycles. Acknowledgement The authors thank Hannah Wills of Caudex Medical (supported by Merck Serono S.A. Geneva) for editorial assistance. References Azziz R. Androgen excess is the key element in polycystic ovary syndrome. Fertil Steril 2003;80: Azziz R, Carmina E, Dewailly D, Diamanti-Kandarakis E, Escobar-Morreale HF, Futterweit W, Janssen OE, Legro R, Norman RJ, Taylor AE et al. The androgen excess and PCOS society criteria for the polycystic ovary syndrome: the complete task force report. Fertil Steril 2009;91: Barbieri RL, Sluss PM, Powers RD, McShane PM, Vitonis A, Ginsburg E, Cramer DC. Association of body mass index, age, and cigarette smoking with serum testosterone levels in cycling women undergoing in vitro fertilization. Fertil Steril 2005;83: Barnes BR, Rosenfield RL, Burstein S, Ehrmann DA. Pituitary-ovarian responses to nafarelin testing in the polycystic ovary syndrome. N Engl J Med 1989;320: Barth JH, Yasmin E, Balen AH. The diagnosis of polycystic ovary syndrome: the criteria are insufficiently robust for clinical research. Clin Endocrinol (Oxf) 2007;67: Boots LR, Porter S, Potter HD, Azziz R. Measurement of total serum testosterone levels using commercially available kits. High degree of between kit variability. Fertil Steril 1998;69: Broekmans FJ, Kwee J, Hendricks DJ, Mol BW, Lambalk CB. A systematic review of tests predicting ovarian reserve and IVF outcome. Hum Reprod Update 2006;12: Chang LP, Lindheim SR, Lowre C, Ferin M, Gonzales F, Berglund L, Carmina E, Sauer M, Lobo RA. Normal ovulatory women with polycystic ovaries have hyperandrogenic pituitary-ovarian responses to gonadotropin-releasing hormone-agonist testing. J Clin Endocrinol Metab 2000;85: Davison SL, Bell R, Montalto JG, Sikaris K, Donath S, Stanczyck FZ, Davis SR. Measurement of total testosterone in women. Comparison of a direct radioimmunoassay versus radioimmunoassay after organic solvent extraction and celite column partition chromatography. Fertil Steril 2005;84: De Placido G, Alviggi C, Perino A, Strina I, Lisi F, De Palo R, Ranieri A, Colacurci N, Mollo A, on behalf of the Italian Collaborative Group on Recombinant Human Luteinizing Hormone. Recombinant human LH supplementation versus recombinant human FSH (rfsh) step-up protocol during controlled ovarian stimulation in normogonadotrophic women with initial inadequate ovarian response to rfsh. A multicentre, prospective, randomized controlled trial. Hum Reprod 2005;20: Ferraretti AP, Gianaroli L, Magli MC, D Angelo A, Farfalli V, Montanaro N. Exogenous luteinizing hormone in controlled ovarian stimulation for assisted reproduction techniques. Fertil Steril 2004;82: Frattarelli JL, Gerber MD. Basal and cycle androgen levels correlate with in vitro fertilization stimulation parameters but do not predict pregnancy outcome. Fertil Steril 2006;86: Frattarelli JL, Peterson EH. Effect of androgen levels on in vitro fertilization cycles. Fertil Steril 2004;81: Hillier SG, Tetsuka M, Fraser H. Location and developmental regulation of androgen receptor in primate ovary. Hum Reprod 1997;12: Horie K, Takakura K, Fujiwara H. Immunohistochemical localization of androgen receptor in the human ovary throughout the menstrual cycle in relation to oestrogen and progesterone receptor expression. Hum Reprod 1992;7: Humaidan P, Bungum M, Bungum L, Yding Andersen C. Effects of recombinant LH supplementation in women undergoing assisted reproduction with GnRH agonist down-regulation and stimulation with recombinant FSH: an opening study. Reprod BioMed Online 2004; 8: Jonard S, Robert Y, Cortet-Rudelli C, Pigny P, Decanter C, Dewailly D. Ultrasound examination of polycystic ovaries: is it worth counting the follicles? Hum Reprod 2003;18: Marrs R, Meldrum D, Muasher S, Schoolcraft W, Werlin L, Kelly E. Randomized trial to compare the effect of recombinant human FSH (follitropin alpha) with or without recombinant human LH in women undergoing assisted reproduction treatment. Reprod BioMed Online 2003;8: McCartney CR, Bellows AB, Gingrich MB, Hu Y, Evans WS, Marshall JC, Veldhuis JD. Exaggerated 17-hydroxyprogesterone response to intravenous infusions of recombinant human LH in women with polycystic ovary syndrome. Am J Physiol Endocrinol Metab 2004;286:E902 E908. Miller KK, Rosner W, Lee H, Hier J, Sesmilo G, Schoenfeld D, Neubauer G, Klibanskyi A. Measurement of free testosterone in normal women and women with androgen deficiency. Comparison of methods. J Clin Endocrinol Metab 2004;89: Nelson VL, Qin K, Rosenfield RL, Wood JR, Penning TM, Legro RS, Strauss JF 3rd, McAllister JM. The biochemical basis for increased testosterone production in theca cells propagated from patients with polycystic ovary syndrome. J Clin Endocrinol Metab 2001;86: Pasquali R, Patton L, Pocognoli P, Cognigni GE, Gambineri A. 17-hydroxyprogesterone responses to gonadotropin-releasing hormone disclose distinct phenotypes of functional ovarian hyperandrogenism and polycystic ovary syndrome. J Clin Endocrinol Metab 2007;11: Piltonen T, Koivunen R, Ruokonen A, Tapanainen JS. Ovarian age-related responsiveness to human chorionic gonadotropin. J Clin Endocrinol Metab 2003;88: Rosenfield RL, Ehrmann DA, Barnes RB, Sheikh Z. Gonadotropin-releasing hormone agonist as a probe for the pathogenesis and diagnosis of ovarian hyperandrogenism. Ann N Y Acad Sci 1993;687: Rosenfield RL, Barnes RB, Ehrmann DA. Studies of the nature of 17- hydroxyprogesterone hyperresonsiveness to gonadotropin-releasing hormone agonist challenge in functional ovarian hyperandrogenism. J Clin Endocrinol Metab 1994;79: Rotterdam ESHRE/ASRM-sponsored PCOS consensus workshop group revised on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 2004;19: Ruvolo G, Bosco L, Pane A, Morici G, Cittadini E, Roccheri MC. Lower apoptosis rate in human cumulus cells after administration

7 234 Hugues et al. of recombinant luteinizing hormone to women undergoing ovarian stimulation for in vitro procedures. Fertil Steril 2007;87: Sahin Y, Kelestimur F. 17-Hydroxyprogesterone responses to gonadotrophin-releasing hormone agonist buserelin and adrenocorticotrophin in polycystic ovary syndrome: investigation of adrenal and ovarian cytochrome P450c17alpha dysregulation. Hum Reprod 1997;12: Suikkari AM, McLachlan V, Montalto J, Calderon I, Healy D, McLachlan RI. Ultrasonographic appearance of polycystic ovaries is associated with exaggerated ovarian androgen and oestradiol responses to gonadotrophin-releasing hormone agonist in women undergoing assisted reproduction treatment. Hum Reprod 1995;10: Vendola KA, Zhou J, Adesanya OO, Weil SJ, Bondy CA. Androgens stimulate early stages of follicular growth in the primate ovary. J Clin Invest 1998;101: Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84: Weil SJ, Vendola K, Zhou J, Adesanya OO, Wang J, Okafor J, Bondy CA. Androgen receptor gene expression in the primate ovary: cellular localization, regulation, and functional correlations. J Clin Endocrinol Metab 1998;83: Weil S, Vendola K, Zhou J, Bondy CA. Androgen and follicle-stimulating hormone interactions in primate ovarian follicle development. J Clin Endocrinol Metab 1999;84: Zawadzki J, Dunaif A. Diagnosis criteria for polycystic ovary syndrome: towards a rational approach. In Dunaif A, Givens JR, Haseltine FP, Merriam GR (eds). Polycystic Ovary Syndrome. Boston, MA: Blackwell Scientific Publications, 1992, Submitted on March 20, 2009; resubmitted on September 17, 2009; accepted on October 2, 2009