Article Relevance of basal serum FSH to IVF outcome varies with patient age

Transcription

1 RBMOnline - Vol 17. No Reproductive BioMedicine Online; on web 16 May 2008 Article Relevance of basal serum FSH to IVF outcome varies with patient age Dr Luca Sabatini, MD, MRCOG, is a Consultant in Obstetrics and Gynaecology and Sub- Specialist in Reproductive Medicine and Surgery at the St Bartholomew s and the Royal London Hospital in London, UK. He obtained his MD (1991) and specialized in Clinical Pathology (1997) at the La Sapienza University, Rome where he worked in the Male Infertility Service of the Medical Pathophysiology Department. In 1997 he moved to UK where he specialized in Obstetrics and Gynaecology obtaining the sub-specialization in Reproductive Medicine. His fields of interest are male and female reproductive endocrinology and assisted reproduction. Dr Luca Sabatini L Sabatini 1,3, A Zosmer 1, EM Hennessy 2, A Tozer 1, T Al-Shawaf 1 1 Barts and the London Centre for Reproductive Medicine, St Bartholomew s and the London Hospital NHS Trust, Kenton and Lucas Wing, St Bartholomew s Hospital, West Smithfield, London, EC1A 7BB; 2 The Wolfson Institute of Preventive Medicine, Barts and the London, Queen Mary s School of Medicine and Dentistry, Charterhouse Square, London EC1M 6BQ, UK 3 Correspondence: Luca.Sabatini@bartsandthelondon.nhs.uk Abstract Live birth rate (LBR), age and basal serum FSH values were analysed in 1589 women undergoing their first cycle of IVF. Four age groups (<30, 30 34, 35 38, years) and three FSH groups (<5, 5 9.9, 10 IU/l) were established. Logistic regression analysis was used to determine the effect of age and FSH on live birth. A model to predict the probability of a live birth suggests that an additional 10 years of age reduces the odds for live birth (OR = 0.66, 95% CI ); an increase of FSH by 5 IU/l reduces the probability of live birth (OR = 0.75, 95% CI ); women 39 years have an additional reduction in probability of live birth (OR = 0.58, 95% CI ). Analysis by age and FSH categories showed that pregnancy rate (PR) did not change significantly with rising FSH for women <35 years old. In cycles started with serum FSH <5 IU/l, increasing age did not effect PR and LBR. Cycles started with serum FSH 10 IU/l had a PR and LBR of 23.6 and 16.9% respectively. The clinical relevance of elevated FSH varies according to age; younger women with elevated FSH and older women with low FSH still have an acceptable LBR. Keywords: age, FSH, IVF, live birth Introduction 10 Evaluating the fertility potential of women is a challenging task for clinicians dealing with infertile couples. Ageing of the ovaries is known to be the main cause of declining fecundity (Menken et al., 1986), but a woman s age is only partially accurate in predicting her reproductive potential (Schwartz and Mayaux, 1982; van Noord-Zaadstra et al., 1991; Akande et al., 2002; te Velde and Pearson, 2002). Early follicular phase serum FSH concentration (basal FSH) has been advocated as a further indicator of ovarian reserve (Navot et al., 1987; Lenton et al., 1988; Muasher et al., 1988), and it has been used to develop prognostic models for assisted reproductive technology (Scott et al., 1989; Toner et al., 1991, Sharif et al., 1998; Shrim et al., 2006). Age and basal FSH are the most widely studied and used parameters for the prediction of outcome in assisted reproduction, as they are inexpensive to determine and cause minimal inconvenience to the patients. The term ovarian reserve is frequently used in relation to female fecundity in assisted reproduction. It has been used in describing both ovarian responsiveness to stimulation (e.g. number of developing follicles, number of oocytes retrieved, cancellation rate) and the reproductive potential, such as implantation and delivery rates (Scott, 2004). The indiscriminate association of a raised basal FSH concentration with reduced ovarian reserve has resulted in a group of women finding difficulty in accessing fertility treatment, as they are regarded as having a very low chance of success (Nasseri et al., 1999; Levi et al., 2001; El- Toukhy et al., 2002; Scott, 2004) Published by Reproductive Healthcare Ltd, Duck End Farm, Dry Drayton, Cambridge CB3 8DB, UK

2 This controversial issue has been reviewed by several authors, who have described the basal FSH concentration to be of limited value in predicting fecundity in subfertile women undergoing assisted reproduction (Bancsi et al., 2003; Sharif and Afnan, 2003; Toner, 2004). Although basal FSH concentration is inversely related to ovarian responsiveness, the take home baby (live birth) rate seems to be less dependent on this parameter, particularly if the patient is young (Esposito et al., 2002; Abdalla and Thum, 2004; van Rooij et al., 2004; Roberts et al., 2005) and younger women with significantly elevated basal FSH can have a very successful ovarian stimulation and live birth following IVF (Tozer et al., 2003). On the basis of these reports, specialists have started to question the ethical and clinical value of excluding women with elevated baseline FSH concentration, irrespective of other factors (Bukman and Heineman, 2001; Wolff and Taylor, 2004; van der Steeg et al., 2006). The debate will continue until more sensitive and cost-effective indicators are introduced to enable clinicians to be more accurate in defining a woman s ovarian reserve (Broekmans et al., 2006; Johnson et al., 2006; Meheshwari et al., 2006). In this study, the outcomes for 1589 women were reviewed following their first cycle of IVF treatment. The results were analysed with respect to age and basal FSH, in order to establish their impact on assisted reproduction outcome and birth rate. By using a logistic regression model, a normogram has been developed to predict the likelihood of achieving a live birth based on age and basal serum FSH concentration. Materials and methods All first IVF cycles of 1589 consecutive patients initiating treatment at Barts and the London Centre for Reproductive Medicine from January 1998 to February 2004 were analysed. Data were extracted from a computer database that collates patient information prospectively during fertility treatment in the centre. The upper age limit for undergoing a cycle of IVF treatment with the patient s own oocytes was 45 years. There was no upper limit to the concentration of basal serum FSH, providing there had been regular menstruation in the previous 6 months. Serum basal FSH and oestradiol concentrations were measured on day 3 (range 1 4) of a menstrual cycle, within 6 months from the start of ovarian stimulation. Since an elevated basal oestradiol concentration might suppress an elevated FSH concentration (Smotrich et al., 1995), a re-evaluation of serum basal FSH concentrations was carried out during the following cycle (day 1 2) if the serum basal oestradiol was higher than 200 pmol/ l. Typically, repeat testing on days 1 or 2 of the cycle results in oestradiol concentrations <200 pmol/l and hence a more accurate FSH value. Serum FSH values with serum oestradiol <200 pmol/l were considered as a basal FSH analysis. Hormone analysis Oestradiol and FSH measurements were performed using a Bayer immuno-1 automated analyser (Bayer, Newbury, Bucks, UK) with a 5% coefficient of variation in the range 75 13,200 pmol/l for oestradiol and 1% coefficient of variation for FSH. The serum FSH concentration was measured by calibrating to the Second International Reference Preparation of the World Health Organization (WHO) (2nd IRP 78/549). The upper level of normality in the study centre is 9.9 IU/l. IVF treatment protocol The stimulation protocol was determined by patient age, body mass index (BMI) and basal FSH concentration (Al-Shawaf et al., 2001). In brief, the exogenous FSH starting dosage to facilitate ovarian stimulation was age dependent: 110 IU/ day if <30 years of age, 150 IU/day if years, 225 IU/ day if years and 300 IU/day if >38 years. If the basal FSH concentration was >9.9 IU/l, the dose was increased by 50% and a BMI of >30 increased the dose by 50%. The dosage of gonadotrophin was further adjusted based on ultrasound ± serum oestradiol after 7 9 days of stimulation. A long luteal phase downregulation with a gonadotrophin releasing hormone analogue (GnRHa) was used in 96% of cycles. Embryo transfer was performed on day 2 or 3 following oocyte retrieval. A maximum of three embryos were transferred for women 40 years old and a maximum of two embryos were transferred for those <40 years old. Definitions A clinical pregnancy was defined as ultrasonographic confirmation of a gestation sac with fetal heartbeat movement seen 4 5 weeks following embryo transfer. A miscarriage was defined as a clinical pregnancy ending spontaneously before the age of viability (24 weeks). Live birth (LB) was defined as a delivery of one or more live babies. Twins and triplets were counted as one live birth in this analysis. An abandoned cycle was defined as the commencement of ovarian stimulation, but no oocyte retrieval was performed. A cycle was abandoned if there were fewer than three follicles 14 mm of diameter seen on ultrasound after days of ovarian stimulation. The fertilization rate was calculated as the number of two pronucleate (2PN) zygotes per number of oocytes retrieved per 100 for each IVF cycle. The implantation rate was defined as the number of gestational sacs seen at the first ultrasound divided by the number of embryos transferred. Statistical analysis Patients were categorized by age and FSH concentration. There were four age groups: <30 years (group 1), (group 2), (group 3) and years (group 4), and also three basal serum FSH concentration groups: , 5 9.9, 10 IU/l. A Kolmogorov Smirnov test was used to assess all variables for normality. As the data were not normally distributed, they are presented as median and range. The differences between categories for numerical data were analysed by Kruskal Wallis test with Dunnett s multiple comparison test, while categorical data (proportions) were analysed by chi-squared for trend and chi-squared analysis. This tested for differences between the groups and not for an effect of increasing age or FSH. Differences between the groups after adjustment for multiple comparisons identified some effects of age or of FSH. This 11

3 data analysis was carried out using GraphPad software (GraphPad Software Inc., San Diego, CA 92130, USA). Multivariate logistic regression analysis was performed to study the independent and combined effects of age and basal serum FSH on the live birth rate (LBR). A multivariate model combining age and the basal serum FSH concentrations was developed to predict the chance of a live birth for specific age and FSH values. Fractional polynomial models and models containing interaction terms for age and FSH concentrations were considered. This analysis used STATA 9 (Stata statistical software, Release 9, 2005; StataCorp, College Station, TX, USA). A P-value of <0.05 was considered statistically significant. Results Effect of age alone The age range of the women included in this study was years (median age 33 years). The basal FSH ranged from 0.7 to 35.2 IU/l, (median 7.2 IU/l). Three embryos were transferred in 12.4% of cases, while 79.3% of cases had two embryos transferred and 8.3% had single embryo transfer. The patient characteristics and their treatment outcomes are shown in Table 1. As expected, the basal serum FSH concentrations increased significantly with increasing female age. The number of oocytes and embryos obtained in each age group significantly diminished as patient age increased. The fertilization rate remained constant across the four age groups; however, the implantation rate was similar only in the two youngest groups and significantly decreased thereafter. Similarly, pregnancy and LBR remained stable in women up to the age of 34, with a statistically significant decline thereafter. The odds of achieving live birth in women younger than 35 years were at least 2.5-fold higher than in women 39 years or older [live birth OR: 3.2 (95% CI ) for age group 1, and live birth OR 2.8 (95% CI ) for age group 2 versus age group 4]. The miscarriage rate increased with age, and affected 25% pregnancies above the age of 38. No statistically significant differences were found in the abandoned cycle rate among the four age groups. The oldest woman to achieve a live birth was 45 years old, treated for male factor infertility and with a basal serum FSH of 6.5 IU/l. Effect of age and basal FSH concentration on IVF secondary outcome measures Table 2 presents data on the effect of basal serum FSH on ovarian stimulation outcomes in the four age groups. In each group, as FSH increased, the median number of oocytes collected and Table 1. Treatment characteristics and clinical outcome in the four age groups (Kruskal Wallis; chi-squared for trend analysis). Parameter Group 1: Group 2: Group 3: Group 4: P-value <30 years years years years No. of patients Median age (years) Median FSH (IU/l) 6.7 ( ) 7.1 ( ) 7.7 ( ) 8.0 ( ) (range) Median dose used (gonadotrophin IU) Median no. of oocytes 11 (0 32) 10 (0 34) 9 (0 37) 7 (0 30) collected (range) Median no. of embryos 6 (0 18) 6 (0 23) 5 (0 24) 4 (0 20) available (range) Fertilization rate (%) NS Implantation rate (%) a No. abandoned cycles 18 (6) 34 (5.3) 21 (4.7) 15 (7.2) NS (%) No. pregnancies (%) 133 (44.3) 259 (40.7) 158 (35.6) 47 (22.5) b No. miscarriages (%) 17 (12.8) 34 (13.1) 24 (15.2) 12 (25.5) NS c No. live births (%) 104 (34.7) 202 (31.8) 113 (25.5) 30 (14.4) d 12 NS = not statistically significant. a All P < , but Group (G)1 versus G2 NS; G3 versus G4 P < b G1 versus G2 NS; G1 versus G3 P < 0.016; G1 versus G4 P < ; G2 versus G3 NS; G2 versus G4 P < ; G3 versus G4 P < c All NS, but G1 versus G4 P < 0.041, G2 versus G4 P < d G1 versus G2 NS; G1 versus G3 P < 0.007; G1 versus G4 P < ; G2 versus G3 P < 0.025; G2 versus G4 P < ; G3 versus G4 P <

4 Table 2. IVF treatment outcome analysed according to age groups and basal FSH concentrations (Kruskal Wallis; chi-squared for trend analysis). Parameter FSH Total Group 1: Group 2: Group 3: Group 4: P-value sub-group * median <30 years years years 39 years (%) (n = 300) (n = 636) (n = 444) or more (n = 209) No. cycles A B C No. of abandoned A 13 (7.7) 2 (4.3) 10 (12.3) 1 (3.8) 0 (0.0) NS cycles (%) B 55 (4.7) 15 (6.6) 20 (4.1) 14 (4.2) 10 (7.2) NS C 16 (6.6) 1 (4.0) 4 (5.5) 6 (6.8) 5 (9.1) NS P-value NS NS o NS NS No. of eggs A (4 23) 11 (2 27) 11 (2 21) 10 (3 30) NS (range) B (0 32) 10 (0 23) 10 (0 37) 8 (0 27) h C 6 8 (4 20) 7 (0 27) 6 (1 27) 5 (1 12) j P-value a b d f No. of embryos A 6 7 (1 15) 5 (0 17) 5 (0 13) 6 (1 15) NS (range) B 5 6 (0 18) 6 (0 23) 5 (0 24) 4 (0 20) i C (1 12) 4 (0 18) 3 (0 17) 2.5 (0 8) k P-value NS c e g Fertilization A NS rate % B NS C NS P-value NS NS NS NS NS Implantation A NS rate % B < m C < n P-value < NS NS < l NS NS = not statistically significant. * A = <5 IU/l; B = IU/l; C = 10 IU/l. a A versus C P < 0.05; the rest are NS. b A versus B NS; A versus C and B versus C P < c A versus B NS; A versus C P < 0.05, B versus C P < d A versus B NS; A versus C P < 0.01, B versus C P < e A versus B NS; A versus C P < 0.05, B versus C P < f A versus B NS; A versus C and B versus C P < g A versus B NS; A versus C and B versus C P < h Group (G)1 versus G2, G2 versus G3, G3 versus G4 all NS; G1 versus G3 and G1 versus G4 P < 0.001; G2 versus G4 P < i G 1 versus G4 and G2 versus G4 P < 0.01; The rest are NS. j G1 versus G2, G2 versus G3, G3 versus G4 NS; G1 versus G3 P < 0.05; G1 versus G4 and G2 versus G4 P < k G1 versus G4 P < 0.05; The rest are NS. l A versus B NS; A versus C P < 0.02 and B versus C P < , Both odds ratio (OR) 2.7. m G1 versus G2 NS; The rest are between P < 0.02 and P < ; G1 versus G4 OR 2.4; G2 versus G4 OR 2.3. n G2 versus G4 and G3 versus G4 NS; G1 versus G2 P < 0.04, OR 2.2; G1 versus G3 P < , OR 5.3; G1 versus G4 P < 0.009, OR 3.0; G2 versus G3 P < 0.01, OR 2.4. o A versus B P < 0.002, OR 3.3; A versus C and B versus C NS. 13

5 14 embryos available for transfer declined. In women with FSH >5 IU/l and those over 30 years old, the number of oocytes and embryos decreased significantly with increasing FSH. After adjustment for multiple comparisons, these secondary outcome measures were not statistically different when the FSH was <5 IU/l despite increasing age, nor in women <30 years with increasing FSH. The abandoned cycle rate was not affected by FSH or age. The fertilization rate remained stable across the different FSH groups. The overall implantation rate statistically declined with increasing basal FSH concentrations (P < ). However, as with number of oocytes retrieved and number of embryos available for transfer, this was not evident in women with FSH <5 IU/l, or in women younger than 30, where the implantation rate was stable with increasing FSH concentrations. The implantation rate was also negatively affected by increasing female age in each of the FSH sub-categories. The highest basal serum FSH concentration that resulted in a live birth was 26.2 IU/l. This patient was 43 years old, diagnosed with unexplained infertility. Effect of age and basal FSH categories on pregnancy rate and live birth rate In general, increasing FSH concentration caused a statistically significant decline in clinical pregnancy and LBR [P < and P < 0.02 respectively (chi-squared for trend)]. This decline was only statistically significant at concentrations >9.9 IU/l. The chances of live birth were at least 1.7-fold higher (95% CI ) in women with FSH <10 compared with women with FSH concentrations 10 IU/l (P = 0.013). Increasing FSH concentrations in the two younger age groups was not reflected in a significant decline in pregnancy rate (PR) and LBR. However, the PR and LBR declined statistically with increasing FSH concentrations in women 35 years or older. The decline was particularly evident with FSH concentrations above 9.9 IU/l. There were no statistically significant differences between the four age groups in clinical PR and live birth rate when the basal serum FSH remained <5 IU/l. In cycles started with a basal FSH concentration between 5 and 9.9 IU/l, the PR and LBR were similar in groups 1, 2 and 3. When the cycles were started at 39 years of age, the pregnancy and LBR dropped significantly (P < , P < , P < and P = , P < , P < for group 4 versus groups 1, 2 and 3 respectively). The odds of achieving a pregnancy and a live birth when the basal FSH was between 5 and 9.9 IU/l was approximately two-fold higher if cycles were started at 38 years compared with those started at >38 years of age (pregnancy OR = 2.5, 2.6, 2.2; live birth OR = 2.8, 3.0, 2.5 for groups 1, 2 and 3 respectively). In cycles started with an FSH 10 IU/l, the negative effect of ageing on clinical outcome was particularly evident beyond the age of 34, as the PR and LBR more than halved compared with women younger than 35 years (Table 3). When serum FSH was 10 IU/l, women younger than 35 years had nearly a three-fold higher chances of a live birth compared with women 35 years or older (OR 2.7, 95% CI ). This data analysis does not show a statistically significant relationship between basal FSH concentrations and miscarriage rate. As expected, the miscarriage rate increased with increasing age, the highest rates reported in the oldest group of women (25.5%). The miscarriage rate in women younger than 39 was 13.6%. Multivariable logistic regression and prediction model Multivariate logistic regression was used to model the relationship between the live birth rate and women s age and basal serum FSH concentrations. Both age and FSH were univariately and independently significantly correlated with the live birth rate. There was a suggestion of a non-linear effect of age, which was not significant using the fractional polynomials, and the interaction effect of FSH and age was not significant if a step effect for age was incorporated into the model. The independent effects of age and a step effect of age and FSH are shown in Table 4. An increase of 10 years in female age corresponds to a decrease of 0.34% in the odds of having a live birth. Moreover, in those over 38 years of age there was a further decrease of 42%. The choice of 39 years as a cut-off point reflected the cut-off points used for the categorisations, so there will be some uncertainty in the results for women around this age, as it is likely that if an individual woman s fertility declines quickly, it will do so at different ages. An increase of basal serum FSH of 5 IU/l in the model is associated with a decrease of 0.25 in the odds of having a live birth across all age ranges. Table 5 shows how the observed and expected LBR from the model are related for each age and FSH category. Expected values are calculated from the predicted proportions for each individual. There are apparent (but not statistically significant) differences in that the actual values are higher for women under 30 with high FSH, and lower for women aged with either low or high FSH, but not for women with FSH in the range IU/l. It is not possible to say whether the model or the data reflects what might be the true age adjusted by FSH risk. Figure 1 shows a normogram, based on the model in Table 4, of the expected live birth probabilities across FSH concentrations up to 30 IU/l for six specified age categories. Discussion This study analysed the effect of age and FSH concentration on ovarian responsiveness, pregnancy and live birth in 1589 consecutive, unselected women undergoing their first IVF cycle. The results showed that ovarian responsiveness and clinical outcome deteriorate with increasing age and serum FSH, confirming the findings of previous studies (Menken et al., 1986; Muasher et al., 1988; Scott et al., 1989). The results

6 Table 3. Clinical outcome analysed according to different age groups and basal FSH concentrations (chi-squared for trend analysis). Parameter FSH Total no. Group 1: Group 2: Group 3: Group 4: P-value sub-group * (%) <30 years years years 39 years (n = 300) (n = 636) (n = 444) or more (n = 209) No. cycles A B C No. of clinical A 62 (36.9) 22 (47.8) 28 (34.6) 7 (26.9) 6 (40) NS pregnancies (%) B 426 (36.1) 89 (38.9) 189 (39.2) 119 (36.1) 28 (20.1) c C 56 (23.2) 13 (52) 21 (28.8) 12 (13.6) 10 (18.2) d P-value a NS NS b NS No. of A 9 (14.5) 2 (9.1) 4 (14.3) 1 (16.7) 2 (33.3) NS miscarriages (%) B 70 (16.4) 15 (16.9) 27 (14.2) 21 (17.6) 7 (25) NS C 8 (14.3) 0 (0.0) 3 (14.3) 2 (16.7) 3 (30) NS P-value NS NS NS NS NS No. of live births A 50 (29.8) 17 (36.9) 24 (29.6) 6 (23.1) 3 (20) NS (%) B 351 (29.7) 74 (32.3) 160 (33.2) 97 (29.4) 20 (14.4) g C 48 (19.9) 13 (52) 18 (24.7) 10 (11.4) 7 (12.7) h P-value e NS NS f NS NS = not statistically significant. * A = <5 IU/l; B = IU/l; C = 10 IU/l. a A versus B NS; A versus C P < 0.02; B versus C P < b A versus B NS; A versus C NS; B versus C P < c Group (G) 1 versus G2 and G1 versus G3 NS; G1 versus G4 P < , odds ratio (OR) 2.5; G2 versus G3 NS; G2 versus G4 P < , OR 2.6; G3 versus G4 P < , OR 2.2. d G1 versus G2 P < 0.035, OR 2.7; G1 versus G3 P < , OR 6.9; G1 versus G4 P < 0.002, OR 4.9; G2 versus G3 P < 0.018, OR 2.6; G2 versus G4 NS; G3 versus G4 NS. e A versus B NS; A versus C P < 0.022, OR 1.7; B versus C P < 0.002, OR 1.7. f A versus B and A versus C NS; B versus C P < , OR 3.2. g G1 versus G4 P < , OR 2.8; G2 versus G4 P < OR 3.0; G3 versus G4 P < , OR 2.5. G1 versus G2 versus G3 (chi-squared for trend) NS. h G1 versus G2 P < 0.011, OR 3.3; G1 versus G3 P < , OR 8.5; G1 versus G4 P < , OR 7.4; G2 versus G3 P < 0.027, OR 2.6; G2 versus G3 versus G4 (chi-squared for trend) NS; G1 + G2 versus G3 + G4 P < , OR 2.7, Table 4. Multivariable logistic regression model of age and basal serum FSH concentration dependent effect on live birth rate in the first cycle of IVF. Variable OR 95% CI P-value Age: increase of 10 years Over 38 years of age FSH: increase of 5 IU/l CI = confidence interval; OR = odds ratio. 15

7 Table 5. Live birth rate and predicted live birth rate by age and FSH category. Age group FSH category (IU/l) (years) < < (40) 32.3 (36) 52.0 (29) a (34) 33.2 (31) 26.9 (25) 0.0 (14) (31) 29.4 (27) 11.6 (22) 0.0 (10) (18) 14.4 (15) 12.2 (12) 16.7 (6) Total 30.0 (34) 30.0 (29) 21.0 (21) 7.0 (10) Values not in brackets are percentage of women in each group having a live birth. Values in brackets are predicted percentage of live births from the logistic regression model for the age and FSH values of the women in each group. a No women younger than 30 years of age had an FSH 20 IU/l. The model predicted, however, a live birth rate of >14%. Pearson goodness of fit test P = Live birth OR <20 years FSH IU/l Figure 1. Model-derived probability (%) of live birth for basal serum FSH (2.5 IU/l increments) in relation to a woman s age. 16 were stratified by age and FSH categories, validating them by a logistic regression model. This model could help the clinical counselling process by providing the patient with visual information on their chances of success. Several other new and interesting points have emerged. First, an age of <35 years seems to protect a woman against the negative effect of a raised FSH concentration. Second, an acceptable LBR has been achieved even in older women in the presence of a raised FSH concentration. These results confirm and expand on reports from others (Hanoch et al.,1998; Esposito et al., 2002; Abdalla and Thum, 2004; van Rooij et al., 2004; Roberts et al., 2005; Caroppo et al., 2006) and are in contrast to studies suggesting that women with elevated basal serum FSH concentrations have a very poor response to ovarian stimulation and a disappointing PR following assisted reproduction, regardless of their age (Muasher et al., 1988; Scott et al., 1989; Cahill et al., 1994; Martin et al., 1996; Levi et al., 2001; El-Toukhy et al., 2002). Many clinics discourage women with raised basal FSH from assisted reproduction based on this principle. This practice has a devastating effect on couples and conflicts with available evidence, i.e. the protective effect that young age confers on raised FSH concentrations in pregnancy outcome. This has also been demonstrated in both spontaneous pregnancies (van Montfrans et al., 2000) and in ovulation induction (Check et al., 1998). Van der Steeg et al. (2006), in a recent study, demonstrated how gynaecologists overestimate the importance of basal FSH concentration during the fertility work-up with inevitable implications for their clinical decision process. In agreement with recent studies (Esposito et al., 2002, van Rooij et al., 2004), it was found that young women with a raised FSH concentration have an IVF outcome that is similar to both those of a similar age and to older women with normal FSH. In this large study group of first cycle IVF, it was shown that pregnancy rates and live birth rates remained acceptable

8 in cycles started with basal FSH concentrations >9.9 IU/l, with one out of five women (19.9%) achieving a successful live birth in their first IVF attempt. Although pregnancy and live birth rates dropped sharply in women over 34 with a basal FSH >9.9 IU/l, their live birth rate per started IVF cycle was nonetheless acceptable (12%), including women over 38 years old. Other groups have previously warned that raised FSH is an indication of impaired oocyte and embryo quality, and have suggested that these are the causes for low fecundity in those women (Nasseri et al., 1999; Levi et al., 2001; El Toukhy et al., 2002). In the present study, although increasing age and FSH concentrations negatively affected the ovarian response, the ability of those oocytes to fertilize remained unchanged. This might suggest that rising age and basal serum FSH affected oocyte number more than oocyte quality (Poe- Zeigler et al., 1994, Grimbizis et al., 1998) or developmental potential (Oehninger et al., 1995; Devroey et al., 1996; Hull et al., 1996). However, interestingly it was found that the implantation rate declined with rising FSH concentrations even in women as young as 30, suggesting that there could be a negative association between FSH concentrations and oocyte quality. Age is a factor in reducing implantation rate and multiple pregnancy in assisted reproduction. The effect of increased basal serum FSH on multiple pregnancies was not studied. A larger number of cycles need to be analysed to reach a valid conclusion. The miscarriage rate was not affected by rising FSH concentrations (Table 3). This result challenges previous reports suggesting a high miscarriage rate (Levi et al., 2001; El Toukhy et al., 2002) and aneuploidy rate (Nasseri et al., 1999) in women with raised FSH, regardless of their age. Another finding from this study is that women older than 38 with basal serum FSH concentrations <5 IU/l conceive as well as younger women, and have a good clinical outcome despite a sharp increase in miscarriages. Their good ovarian response and live birth rate implies that that they have an ovarian reserve of oocytes that is higher than average, given their age. Previous reports have indirectly supported this finding. Al Shawaf et al. (1992) studied the influence of ovarian response in gamete intra-fallopian transfer (GIFT) and demonstrated that although pregnancy and live birth rate declined in women older than 40 years, the LBR increased as a function of the number of oocytes retrieved. They concluded that among older women, there is a sub-group who respond well to ovarian stimulation by producing an adequate number of oocytes and demonstrating fecundity similar to that of much younger women. This subgroup may consist of women who have an ovarian biological age younger than their chronological age, and thus a low basal serum FSH concentration. The term ovarian reserve has been used to describe both ovarian responsiveness to stimulation (e.g. number of eggs, cancellation rate) and the reproductive potential, such as implantation and delivery rates (Scott, 2004). The indiscriminate association of a raised basal FSH concentration with reduced ovarian reserve has resulted in a group of women finding difficulty in accessing fertility treatment, as they are regarded as having a potentially poor outcome (Nasseri et al., 1999; Levi et al., 2001; El Toukhy et al., 2002; Scott, 2004). It appears that a critical predictor in achieving successful assisted reproduction treatment is the number and quality of embryos available for transfer. In this regard, the importance of knowing basal FSH concentration would lie in developing a tailored stimulation protocol that can maximize the ovarian response by achieving the highest number of high quality embryos. One of the main differences between the present study and that of previous published articles is that the starting dose of gonadotrophin used to achieve ovarian stimulation was individualized according to the patient s age, BMI, and basal FSH. This could be responsible for the relatively constant abandoned cycle rate across the different age and FSH groups. Most of the previous published articles used either a fixed standard dose of gonadotrophins (El Toukhy et al., 2002; Abdalla et al., 2004; van Rooij et al., 2004) to achieve ovarian stimulation, or they did not make this information available (Levi et al., 2001; Frazier et al., 2004) in their articles. Despite adjusting the ovarian stimulation gonadotrophin dosage, increasing age combined with increasing FSH remain negatively confounding variables to the number of oocytes retrieved (Table 2). What also distinguishes this study from others is that it reports on the effect of age and FSH concentration on live birth rather than solely on clinical PR or number of oocytes obtained (Sharif et al., 1998; Esposito et al., 2002; Bancsi et al., 2003; van Rooij et al., 2004; Yih et al., 2005). This study was not designed to identify whether age or basal serum FSH was the more predictive of IVF success, as this is probably not relevant in the clinical setting. Instead, the study has aimed to provide a more precise prediction of a woman s chance of taking home a baby based purely on her age and FSH prior to starting an assisted reproduction cycle. It is acknowledged that other indicators such as antral follicle count, ovarian volume or serum anti-müllerian hormone have been shown, either singularly or in combination, to outperform age and FSH in predicting IVF success in a research setting (van Rooij et al., 2002; Kwee et al., 2003; Bukulmez and Arici, 2004); age and FSH remain the most inexpensive, easily reproducible and widely available predictors in the majority of IVF clinics worldwide. There are sources of bias in the present study. The retrospective nature is common to studies of this type (Muasher et al., 1988; Abdalla and Thum, 2004; van Rooij et al., 2004; Caroppo et al., 2006; Shrim et al., 2006). The normogram (Figure 1) must be used with caution for the unlikely combinations, e.g. very high FSH >20 IU/l in very young women or very low FSH in the oldest), as the data for these combinations are based on small or no numbers, but in any case there will be little information available to guide decisions in these women. In conclusion, a raised basal serum FSH concentration >9.9IU/l still leads to an acceptable PR and LBR, and its use to exclude women from assisted reproduction programmes does not appear to be justified. Women of 39 years of age and above and whose serum basal FSH concentrations remain less than 5 IU/l have an LBR per first cycle of 20%. An age-corrected interpretation of FSH values is necessary when counselling patients in IVF. References Abdalla H, Thum MY 2004 An elevated basal FSH reflects a quantitative rather than qualitative decline of the ovarian reserve. Human Reproduction 19,

9 18 Akande VA, Fleming CF, Hunt LP et al Biological versus chronological ageing of oocytes, distinguishable by raised FSH levels in relation to the success of IVF treatment. Human Reproduction 17, Al-Shawaf T, Zosmer A, Hussain S et al Prevention of severe ovarian hyperstimulation syndrome in IVF with or without ICSI and embryo transfer: a modified coasting strategy based on ultrasound for identification of high-risk patients. Human Reproduction 16, Al-Shawaf T, Nolan A, Guirgis R et al The influence of ovarian response on gamete intra-fallopian transfer outcome in older women. Human Reproduction 7, Bancsi LF, Broekmans FJ, Mol BW et al Performance of basal follicle-stimulating hormone in the prediction of poor ovarian response and failure to become pregnant after in vitro fertilization: a meta-analysis. Fertility and Sterility 79, Broekmans FJ, Kwee J, Hendricks DJ et al A systematic review of tests predicting ovarian reserve and IVF outcome. Human Reproduction Update, 12, Bukman A, Heineman MJ 2001 Ovarian reserve testing and the use of prognostic models in patients with subfertility. Human Reproduction Update 7, Bukulmez O, Arici A 2004 Assessment of ovarian reserve. Current Opinion in Obstetrics and Gynecology 16, Cahill DJ, Prosser CJ, Wardle PG et al Relative influence of serum follicle stimulating hormone, age and other factors on ovarian response to gonadotrophin stimulation. British Journal of Obstetrics and Gynaecology 101, Caroppo E, Matteo M, Schonauer LM, et al Basal FSH concentration as a predictor of IVF outcome in older women undergoing stimulation with GnRH antagonist. Reproductive BioMedicine Online 13, Check JH, Peymer M, Lurie D 1998 Effect of age on pregnancy outcome without assisted reproductive technology in women with elevated early follicular phase serum follicle-stimulating hormone levels. Gynecologic and Obstetric Investigation 45, Devroey P, Godoy H, Smitz J et al Female age predicts embryonic implantation after ICSI: a case-controlled study. Human Reproduction 11, El-Toukhy T, Khalaf Y, Hart R et al Young age does not protect against the adverse effects of reduced ovarian reserve an eight year study. Human Reproduction 17, Esposito MA, Coutifaris C, Barnhart KT 2002 A moderately elevated day 3 FSH concentration has limited predictive value, especially in younger women. Human Reproduction 17, Frazier LM, Grainger DA, Schieve LA, Toner JP 2004 Folliclestimulating hormone and estradiol levels independently predict the success of assisted reproductive technology treatment. Fertility and Sterility 82, Grimbizis G, Vandervorst M, Camus M et al Intracytoplasmic sperm injection, results in women older than 39, according to age and the number of embryos replaced in selective or non-selective transfers. Human Reproduction 13, Hanoch J, Lavy Y, Holzer H et al Young low responders protected from untoward effects of reduced ovarian response. Fertility and Sterility 69, Hull MG, Fleming CF, Hughes AO, McDermott A 1996 The agerelated decline in female fecundity: a quantitative controlled study of implanting capacity and survival of individual embryos after in vitro fertilization. Fertility and Sterility 65, Johnson NP, Bagrie EM, Coomarasamy A et al Ovarian reserve tests for predicting fertility outcomes for assisted reproductive technology. The international systematic review of ovarian reserve test accuracy. British Journal of Obstetrics and Gynaecology 113, Kwee J, Elting MW, Schats R et al Comparison of endocrine tests with respect to their predictive value on the outcome of ovarian hyperstimulation in IVF treatment: result of a prospective randomized study. Human Reproduction 18, Lenton EA, Sexton L, Lee S, Cooke ID 1988 Progressive changes in LH and FSH and LH:FSH ratio in women throughout reproductive life. Maturitas 10, Levi AJ, Raynault MF, Bergh PA et al Reproductive outcome in patients with diminished ovarian reserve. Fertility and Sterility 76, Martin JS, Nisker JA, Tummon IS et al Future in vitro fertilization pregnancy potential of women with variably elevated day 3 follicle-stimulating hormone levels. Fertility and Sterility 65, Meheshwari A, Fowler P, Bhattachrya S 2006 Assessment of ovarian reserve-should we perform tests ovarian reserve routinely? Human Reproduction 21, Menken J, Trussell J, Larsen U 1986 Age and infertility. Science 233, Erratum in: Science (5775), 413. Muasher SJ, Oehninger S, Simonetti S et al The value of basal and/or stimulated serum gonadotropin levels in prediction of stimulation response and in vitro fertilization outcome. Fertility and Sterility 50, Nasseri A, Mukherjee T, Grifo JA et al Elevated day 3 serum follicle stimulating hormone and/or estradiol may predict fetal aneuploidy. Fertility and Sterility 71, Navot D, Rosenwaks Z, Margalioth EJ 1987 Prognostic assessment of female fecundity. Lancet 19, Oehninger S, Veeck L, Lanzendorf S et al Intracytoplasmic sperm injection: achievement of high pregnancy rates in couples with severe male factor infertility is dependent primarily upon female and not male factors. Fertility and Sterility 64, Poe-Zeigler R, Toner JP, Oehninger S, Muasher SJ 1994 Basal FSH affects IVF pregnancy rates primarily by influencing oocyte numbers and not their quality (abstract). Fertility and Sterility 63 (Suppl. 1), S8. Roberts JE, Spandorfer S, Fasouliotis SJ et al Taking a basal follicle-stimulating hormone history is essential before initiating in vitro fertilization. Fertility and Sterility 83, Schwartz D, Mayaux MJ 1982 Female fecundity as a function of age: results of artificial insemination in 2193 nulliparous women with azoospermic husbands. Federation CECOS. New England Journal of Medicine 306, Scott RT Jr 2004 Diminished ovarian reserve and access to care. Fertility and Sterility 81, Scott RT, Toner JP, Muasher SJ et al.1989 Follicle-stimulating hormone levels on cycle day 3 are predictive of in vitro fertilization outcome. Fertility and Sterility 51, Sharif K, Afnan M 2003 The IVF league tables: time for a reality check. Human Reproduction 18, Sharif K, Elgendy M, Lashen H, Afnan M 1998 Age and basal follicle stimulating hormone as predictors of in vitro fertilisation outcome. British Journal of Obstetrics and Gynaecology 105, Shrim A, Elizur SE, Seidman DS et al Elevated day 3 FSH/ LH ratio due to low LH concentrations predicts reduced ovarian response. Reproductive BioMedicine Online 12, Smotrich DB, Widra EA, Gindoff PR et al Prognostic value of day 3 estradiol on in vitro fertilization outcome. Fertility and Sterility 64, te Velde ER, Pearson PL 2002 The variability of female reproductive ageing. Human Reproduction Update 8, Toner JP 2004 Modest follicle-stimulating hormone elevations in younger women: warn but don t disqualify. Fertility and Sterility 81, (discussion ). Toner JP, Philput CB, Jones GS, Muasher SJ 1991 Basal folliclestimulating hormone level is a better predictor of in vitro fertilization performance than age. Fertility and Sterility 55, Tozer AJ, Al-Shawaf T, Gillott CM et al Excessive follicular response to controlled ovarian stimulation in a woman with menopausal FSH levels: case report. Human Reproduction 19, van der Steeg JW, Steures P, Eijkemans MJC et al Which factors play a role in clinical decision-making in subfertility? Reproductive BioMedicine Online 12, van Montfrans JM, Hoek A, van Hooff MH et al Predictive value of basal follicle-stimulating hormone concentrations in a

10 general subfertility population. Fertility and Sterility 74, van Noord-Zaadstra BM, Looman CW, Alsbach H et al Delaying childbearing: effect of age on fecundity and outcome of pregnancy. British Medical Journal 302, van Rooij IA, de Jong E, Broekmans FJ et al High folliclestimulating hormone levels should not necessarily lead to the exclusion of subfertile patients from treatment. Fertility and Sterility 81, van Rooij IA, Broekmans FJ, te Velde ER et al Serum antimullerian hormone levels: a novel measure of ovarian reserve. Human Reproduction 17, Wolff EF, Taylor HS 2004 Value of the day 3 follicle-stimulating hormone measurement. Fertility and Sterility 81, Yih MC, Spandorfer SD, Rosenwaks Z 2005 Egg production predicts a doubling of in vitro fertilization pregnancy rates even within defined age and ovarian reserve categories. Fertility and Sterility 83, Declaration: The authors report no financial or commercial conflicts of interest. Received 24 August 2007; refereed 14 September 2007; revised and resubmitted 17 December 2007; accepted 22 February