FERTILITY AND STEFULI~ Vol. 68, No. 3, September 1997 Copyright 1997 American Society for Reproductive Medicine Printed on acid-free paper in U. S. A. Published by Elsevier Science Inc. Follicle-stimulating hormone levels on cycle day 3 to predict ovarian response in women undergoing controlled ovarian hyperstimulation for in vitro fertilization using a flare-up protocol Timur Giirgan, M.D.* Bulent Urman, M.D. Hakan Yarali, M.D. Hakan E. Duran, M.D. Hacettepe University, Faculty of Medicine, Department of Obstetrics and Gynecology, Ankara, Turkey Objective: To determine whether cycle day 3 serum FSH levels predict poor ovarian response and cycle cancellation in women undergoing controlled ovarian hyperstimulation (COH) for IVF. Design: Prospective case series. Setting: University hospital IVF program. Patient(s): The study includes outcomes of 637 initial cycles of M?. Intervention(s): Controlled ovarian hyperstimulation was undertaken with a short, flareup GnRH agonist and hmg protocol. The relation of cycle day 3 FSH levels with various cycle characteristics and outcomes was determined. Main Outcome Measure(s): The accuracy of cycle day 3 FSH levels in predicting poor ovarian response and cycle cancellation was evaluated. Receiver operating characteristic curves were constructed to determine the optimum cutoff value for the cycle day 3 serum FSH level that best predicted poor ovarian response and cycle cancellation. Result(s): A cycle day 3 serum FSH level 213 miu/ml (conversion factor to SI unit, 1.00) was associated with a high likelihood of poor ovarian response and cycle cancellation. The sensitivity, specificity, and positive likelihood ratio of an FSH level of 13 miu/ml in predicting cycle cancellation were 52%, 91%, and 6.04, respectively. Conclusion(s): Cycle day 3 FSH screening appears to be a fairly accurate predictor of poor ovarian response and cycle cancellation in women undergoing COH using a short, flare-up GnRH agonist and hmg protocol. (Fertil Steril@ 1997;68:483-7. 0 1997 by American Society for Reproductive Medicine.) Key Words: Controlled ovarian hyperstimulation, day 3 FSH level, cycle cancellation, poor ovarian response, flare-up protocol The advent of intracytoplasmic sperm injection largely has obviated male infertility as a successlimiting factor in assisted reproduction. However, poor ovarian response remains a major challenge to overcome. Advanced female age is associated with poor hormonal and follicular response to controlled ovarian hyperstimulation (COH), with fewer available oocytes for retrieval and fewer embryos for transfer. These factors collectively increase cycle Received June 3, 1996; revised and accepted May 23, 1997. *Reprint requests: Timur Giirgan, M.D., Hacettepe Universitesi Tip Fakuhsi, Kadin HastalikIari ve Dogum Ana Bilim Dali, Sihhiye 06100, Ankara, Turkey (FAX: 90-312-310 7632; e- mail: HDURAN@NEKJRON.ATO.ORG.TRh cancellation and decrease pregnancy rates (PRs) (1). Nevertheless, chronologic age does not always show a linear correlation with the so-called ovarian age. Whereas some young women respond poorly to COH, some older women demonstrate an adequate response. In view of this fact, other variables besides age have been sought to predict ovarian response to exogenous gonadotropins (2). Of these, the most widely used is assessment of ovarian reserve using cycle day 3 FSH levels. Assessment of cycle day 3 FSH levels fulfills the criteria for a good screening test in that it is easily measurable, minimally invasive, and inexpensive, and it has good predictive value (3). The aim of the current study was to assess the 001850282/97/$17.00 PI1 SOO15-0282(97)00246-X
value of cycle day 3 serum FSH levels in determining ovarian response to a flare-up GnRH agonist (GnRHa) protocol combined with exogenous gonadotropins in women undergoing COH for IVF. MATERIALS AND METHODS This study included 637 initial cycles of IVF performed from May 1991 to January 1995. Cycle day 3 FSH levels were measured using the Amerlex-M FSH RIA kit (Johnson & Johnson Clinical Diagnostics Ltd., Amersham, Buckinghamshire, United Kingdom) on the third day of a spontaneous cycle, within 3 months of the IVF cycle. The within-assay and between-assay coefficients of variation of the kit were 3.2% and 8.4%, respectively. None of the patients had received any hormonal medication during the preceding menstrual cycle. Superovulation was undertaken using a short, flare-up GnRH-a protocol combined with hmg. Buserelin acetate nasal spray (Suprefact; Hoechst AG, Frankfurt-Main, Germany) was initiated (100 pg four times per day) on the first day of a spontaneous menstrual bleed and continued until the administration of hcg. Human menopausal gonadotropin was commenced on cycle day 3, provided a baseline vaginal ultrasonographic (US) examination showed the absence of cystic ovarian structures >lo mm in diameter, with a starting dosage of 225 U daily for 4 days. The dosage of hmg was individualized according to daily serum E2 levels and follicular response as assessed with vaginal US. Human chorionic gonadotropin was administered when at least two follicles reached or exceeded a mean diameter of 18 mm, with an Ez level >500 pg/ml (conversion factor to SI unit, 3.671). Follicular aspiration for oocyte collection was performed with transvaginal US guidance 36 hours after the administration of hcg. Sperm samples were prepared by the standard swim-up technique. Standard laboratory procedures were used for the handling of gametes and embryos (4, 5). Minimal essential medium with Earle s salts and L-glutamine (GIBCO BRL, Life Technologies Ltd. Paisley, Strathclyde, United Kingdom) were used as the insemination and culture media. The female partner s serum in 10% to 15% concentrations was added to the medium for culture of embryos. Embryo transfer was performed 48 hours after oocyte collection if fertilization was present. A maximum of four embryos were transferred. The luteal phase was supported with injections of hcg, 1,500 U, on days 3,6, and 9 after ET. Pregnancy was diagnosed by serially rising serum P-hCG (Amerlite; Amersham International, Amerikastraat, The Neth- erlands) levels, and a clinical pregnancy was confirmed by US demonstration of one or more gestational sacs. Several IVF cycle characteristics were evaluated in relation to different cutoff points of cycle day 3 FSH levels. The capacity of a given FSH level to discriminate between adequate and poor ovarian response was determined with 2 x 2 contingency tables. Cycle cancellation was undertaken, with poor ovarian response defined as a serum E2 level ~500 pg/ml and fewer than two follicles measuring 218 mm in diameter. Receiver operating characteristic curves were constructed to determine the optimum cycle day 3 FSH level that best predicted poor E2 response and cycle cancellation. Likelihood ratios were calculated to determine the predictive accuracy of cycle day 3 FSH levels to presage poor ovarian response and cycle cancellation. All data were processed with the Statistical Program for Social Sciences (SPSS Inc., Chicago, IL) on an Apple Macintosh (Bilkom, Istanbul, Turkey) personal computer. One-way analysis of variance with the Tukey Honest Significant Difference post hoc test was used for analysis of the results. Data were expressed as means + SD. Statistical significance was accepted at a P level <0.05. Receiver operating characteristic curve analysis was performed by Medcalc (Medcalc Software, Mariakerke, Belgium) on an IBM-compatible personal computer (Condor Computers, Istanbul, Turkey). RESULTS Cycle characteristics in relation to various cutoff levels of cycle day 3 FSH levels are shown in Table 1. Estradiol levels on the day of hcg administration were significantly lower when day 3 FSH levels were >15 miu/ml (conversion factor to SI unit, 1.00). Further, E2 levels of women with day 3 FSH levels > 13 miu/ml were significantly lower than those of women with day 3 FSH levels <lo miu/ml. Similarly, the number of follicles >14 mm in diameter, the number of oocytes retrieved, and ovarian response showed a decreasing trend with increasing day 3 FSH levels. Cycle cancellation rates increased from 7.9% to 55.5% in women with day 3 FSH levels of <lo and >20 miu/ml, respectively. Fertilization rates did not differ between groups. The number of embryos available for transfer was significantly less when day 3 FSH levels exceeded 15 miu/ml. The clinical PR per initiated cycle was 19.3% when the day 3 FSH level was ~10 miu/ml. This figure decreased to 6.7% when the FSH level was >15 miu/ml. No pregnancies were achieved when the FSH level was >20 miu/ml. 484 Giirgan et al. Cycle day 3 FSH levels and response to COH Fertility and Sterility@
Table 1 In Vitro Fertilization Cycle Characteristics in Relation to Cycle Day 3 FSH Levels* Group 1 Group 2 Group 3 Group 4 Group 5 No. of cycles 491 67 20 41 18 Age Wt 33.3 4.2 34.4 2 2.3 35.7 2 4.7 36.5 k 4.7 34.0 k 3.5 E2 level on the day of hcg adminstration (pg/ml)$ l/qi 1,430 2 674 1,386 2 695 1,095 2 859 440 2 454 370? 342 No. of follicles >14 rnrn$lln 4.2 k 2.3 3.6 2 1.4 3.0 1.5 2.1? 2.3 1.6? 0.9 Poor responsell** 56 (11.4) 10 (14.9) 4 (20) 23 (56) 12 (66.6) Cycle cancellation ** 39 (7.9) 7 (10.4) 3 (15) 19 (46.7) 10 (55.5) No. of occytes$tt 7.3 2 4.6 7.1 2 4.5 5.2 2 3.9 4.6 t 4.2 3.8 2 1.3 Fertilization rate (%I$ 56.2 2 32.5 58.9 k 30.3 60.9 k 40.1 61.0 t 38.7 52.6 2 39.2 No. of embryos transferred311 2.4 2 1.7 2.6 r 1.6 2.1 2 1.7 1.1 k 1.4 0.8 k 1.3 Clinical pregnancy per cycle (%)I/ 19.3 26.8 15 6.7 0 * FSH levels (in miu/ml; conversion factor to SI unit, 1.00): group 1, 0 to 9; group 2, 10 to 12.9; group 3, 13 to 14.9; group 4, 15 to 19.9; and group 5, 220. t Groups 3 and 4 are significantly different from Groups 1, 2, and 5. $ Values are means 2 SD. Q Conversion factor to SI unit, 3.671. 11 Groups 4 and 5 are significantly different from Groups 1, 2, and 3. 1 Group 3 is significantly different from Group 1. ** Values in parentheses are percentages. tt Groups 3,4, and 5 are significantly different from Groups 1 and 2. Regression analysis showed that age, day 3 FSH level, and etiology of infertility had significant association with both cycle cancellation rates and low E2 levels. Logistic regression analysis, however, showed that day 3 FSH level was the only independent variable that had a statistically significant association with low Ez levels on the day of hcg administration and subsequent cycle cancellation (log likelihood = 260.5, significance of log likelihood ratio (LR) < 0.001). Receiver operating characteristic curves were constructed to determine the optimum age of the woman and the day 3 FSH level at which cycle cancellation was predicted best (Figs. 1 and 2). The day 3 FSH level that predicted poor hormonal response and cy- cle cancellation most accurately was >13 miu/ml. Likewise, the age of the woman that predicted these outcomes most accurately was >36 years (Table 2). DISCUSSION This study demonstrates that FSH levels measured on the third day of a spontaneous menstrual cycle are predictive of poor hormonal response and subsequent cycle cancellation in women undergoing COH for IVF with a short, flare-up GnRH-a and hmg protocol. The likelihood of poor hormonal response, defined as an E2 level <500 pg/ml on the day of hcg administration, and the likelihood of cycle cancellation were 2.89 and 6.04, respectively, when False-positive fraction Figure 1 Receiver operating curve for cycle day 3 FSH level in predicting cycle cancellation; cycle day 3 FSH level of >13 miu/ ml (conversion factor to SI unit, 1.00) best predicted cycle cancellation. Figure 2 Receiver operating curve for cycle day 3 FSH level in predicting poor Ez response; cycle day 3 FSH level of >13 miu/ ml (conversion factor to SI unit, 1.00) best predicted poor E, response. Vol. 68, No. 3, September 1997 Giirgan et al. Cycle day 3 FSH levels and response to COH 485
Table 2 Prediction of Cycle Cancellation and Ez Level <500 pg/ml on the Day of HCG Administration When Age >36 Years and Cycle Day 3 FSH >13 miu/ml Sensitivity Specificity Positive Positive Negative likelihood ratio predictive value predictive value Prediction of cycle cancellation Age >36 y FSH >13 III/L Prediction of Ez level <500 pg/ml on the day of hcg administration Age >36 y FSH >13 miu/ml 49.2 71.0 1.7 19.5 90.7 50.0 89.3 2.89 52.6 88.2 49.2 71.0 1.7 19.5 90.7 52.2 91.4 6.04 63.2 87.1 * Conversion factor to SI unit: Ez, 3.671; FSH, 1.00. day 3 FSH levels reached or exceeded 13 miu/ml. Age alone appeared to be a less accurate predictor of ovarian response to COH than 3 day FSH levels. Ovarian response to hmg previously was shown to be predicted fairly accurately with day 3 FSH levels. Muasher et al. (6) investigated the predictive value of cycle day 3 gonadotropins in determining the extent of ovarian response to COH for IVF using FSH and hmg. Their study demonstrated the value of cycle day 3 serum gonadotropin measurements in distinguishing between different responders to exogenous gonadotropins. In addition to predicting poor ovarian response, increasing FSH levels were associated with decreasing PRs. Similar conclusions were reached in a study by Toner et al (21, who analyzed 1,478 consecutive IVF cycles using hmg and/or FSH for COH. These authors concluded that FSH level is a better predictor of IVF outcome than chronologic age. Serum cycle day 3 FSH screening also has been reported to be a sensitive indicator of ovarian responsiveness to stimulation with hmg after pituitary suppression with a GnRH-a (buserelin) initiated in the luteal phase (7). Whether the predictive value of serum day 3 FSH screening is retained in a short, flareup GnRH-a protocol is unknown. Because a GnRH-a in a short, flare-up protocol usually is advocated in poor responders, the relation between certain cycle outcomes and day 3 serum FSH screening becomes more important (8). Padilla et al. (9) investigated the prognostic value of early FSH and LH levels and serum Ez response on IVF outcome in 100 cycles, using a flare-up GnRH-a protocol. They concluded that, although not statistically significant, an early serum FSH level >20 miu/ml was associated with a lower clinical PR. They also stated that early Ez response to leuprolide acetate probably was the best prognostic indicator of IVF outcome. Our results demonstrate that the day 3 FSH level is the only independent variable to predict a low Ez level on the day of hcg administration and subsequent cycle cancellation in a flare-up GnRH-a and hmg protocol. Using receiver operating characteristic curves, the day 3 FSH level that predicted poor hormonal response and cycle cancellation most accurately was 13 miu/ml. It is unknown whether routine screening of all IVF candidates with day 3 serum FSH levels is cost-effective. Although diminished ovarian reserve appears to be relatively rare in a young patient with two ovaries, screening regardless of age is recommended when the woman has a single ovary or a history of ovarian endometriosis or multiple ovarian operations (10). Routine screening may identify the occasional young woman with diminished ovarian reserve. Our results may be criticized because they are constructed on a single day 3 FSH level. However, Scott et al. (11) performed a paired analysis of high and low FSH cycles in women undergoing IVF and showed no differences in stimulation quality, the number of oocytes retrieved, or fertilization rates. All patients behaved as low responders in both cycles. In conclusion, this study demonstrates the accuracy of day 3 serum FSH screening in predicting poor ovarian response and subsequent cycle cancellation. A day 3 serum FSH cutoff level > 13 miu/ml is associated with a high likelihood of cycle cancellation when a short, flare-up GnRH-a protocol combined with hmg is used for COH. 1. 2. 3. 4. 5. REFERENCES Scott R, Toner J, Muasher S, Oehninger S, Robinson S, Rosenwaks Z. Follicle stimulating hormone levels on cycle day 3 are predictive of in vitro fertilization outcome. Fertil Steril 1989;51:651-4. Toner J, Philput C, Jones J, Muasher S. Basal follicle stimulating hormone level is a better predictor of in vitro fertilization performance than age. Fertil Steril 1991;55:784-91. Scott R, Hofmann G. Prognostic assessment of ovarian reserve. Fertil Steril 1995;63:1-11. Dandekar P, Quigley M. Laboratory setup for human in vitro fertilization. Fertil Steril 1984; 42:1-11. The American Fertility Society. Guidelines for in vitro fertilization, gamete intrafallopian transfer and related procedures. Fertil Steril 1991;56:194-7. 486 Giirgan et al. Cycle day 3 FSH levels and response to COH Fertility and Sterility@
6. Muasher SJ, Oehninger S, Simonetti S, Matta J, Ellis LM, Liu HC, et al. The value of basal and stimulated gonadotropin levels in prediction of stimulated response in in-vitro fertilization outcome. Fertil Steril 1988;50:298-307. 7. Cahill D, Wardle P, Ford W, Hull M. Relative influence of serum follicle stimulating hormone, age, and other factors on ovarian response to gonadotrophin stimulation. Br J Obstet Gynaecol 1994; 101:999-1002. 8. Muasher S. Controversies in assisted reproduction: treatment of low responders. J Assist Reprod Genet 1993;lO: 112-4. 9. Padilla SL, Bayati J, Garcia JE. Prognostic value of the early serum e&radio1 response to leuprolide acetate in in vitro fertilization. Fertil Ski1 1990;53:288-94. 10. Khalifa E, Toner J, Muasher S, Acosta A. Significance of basal follicle stimulating hormone levels in women with one ovary in a program of in vitro fertilization. Fertil Steril 1992;57: 835-9. 11. Scott R, Hofmann G, Oehninger S, Muasher S. Intercycle variability of day 3 follicle stimulating hormone levels and its effect on stimulation quality in in vitro fertilization. Fertil Steril 1990;54:297-302. Vol. 68, No. 3, September 1997 Giirgan et al. Cycle day 3 FSH levels and response to COH 487