Age and basal follicle stimulating hormone as predictors of in vitro fertilisation outcome

British Journal of Obstetrics and Gynaecology January 1998, Vol. 105, pp. 107-1 12 Age and basal follicle stimulating hormone as predictors of in vitro fertilisation outcome Khaldoun Sharif Lecturer, Manal Elgendy Research Fellow, Hany Lashen Research Fellow, Masoud Afnan Senior Lecturer Assisted Conception Unit, Birmingham Women j. Hospital Objective To examine the relative effect of basal follicle stimulating hormone (FSH) concentration and the woman s age on predicting the ovarian response to gonadotrophin stimulation, normal fertilisation rate and pregnancy rate in in vitro fertilisation (IVF) treatment following pituitary desensitisation. Design Descriptive cohort study. Participants Three hundred and forty-four women undergoing their first IVF cycle. Methods Basal (menstrual-day 3) FSH concentration was measured and the woman s age calculated before she underwent pituitary desensitisation followed by gonadotrophin ovarian stimulation and IVF treatment. Main outcome measures Cancellation rate due to poor ovarian response, total dose of gonadotrophin required to achieve follicular maturity, number of oocytes collected, normal fertilisation rate and pregnancy rate were compared between banded values of the variables studied. Results Increasing basal FSH concentration was associated significantly with increased cancellation rate, but increasing age was not. Both increasing basal FSH and age were associated significantly with increased total gonadotrophin dose, and reduced number of oocytes collected and pregnancy rate. Analysis of variance showed that the association for basal FSH with the number of oocytes was significant, independent of, and stronger than the effects of age. Logistic regression analysis showed that age, but not basal FSH, was independently associated with pregnancy rate. Neither basal FSH, nor age had significant association with normal fertilisation rate. Conclusion Basal FSH concentration is a better predictor of cancellation rate and of the number of oocytes collected in IVF treatment than age, but age is a stronger predictor of pregnancy rate. INTRODUCTION One of the factors that infertile couples take into consideration before deciding to embark on in vitro fertilisation treatment is the expected chances of achieving a pregnancy. Advancing age in the woman is well recognised to reduce natural fertility as well as the chances of pregnancy after all types of infertility treatment, including IVF1q2. As the menopause approaches, the chances of pregnancy decline. However, different women reach the menopause at different ages, and it is biologically plausible that the age-related decline in fertility may occur at different rates in different women. Age alone has limited predictive value, and it would be helpful, for both patient counselling and clinical management, to have another independent prognostic factor that can predict the true reproductive potential of each woman individually. One study3 has reported that women may Correspondence: Dr K. Sharif, Assisted Conception Unit, Birmingham Women s Hospital, Birmingham B15 2TG, UK. begin to have a subtle increase in their serum follicle stimulating hormone concentrations in their mid-thirties, coinciding with the time at which fertility begins to decline. Subsequently, another studf has shown that the earliest increase in FSH occurs in the early follicular phase. The use of early follicular phase (basal) serum FSH concentration was described by Scott et ~ 1 of. the ~ Jones institute, Norfolk, Virginia, USA, as a predictor of IVF outcome and was found by Toner et ~ 1 to. be ~ more predictive than age. However, Toner et al6 used pituitary desensitisation (which is now used in most IVF programmes) in only 38% of their study cycles and did not report on the analysis of those cycles separately. Furthermore, they transferred up to four embryos per woman; the maximum number allowed by law in the UK is three embryos, and many other countries have similar limits. Here again the analysis of their pregnancy rates (which are well recognised to correlate with the number of embryos transferred) may not be applicable in the UK. More recently, Cahill et al. reported on 0 RCOG 1998 British Journal of Obstetrics and Gynaecology 107

108 K. SHARIF ET AL. 171 IVF cycles following pituitary desensitisation and demonstrated the superior predictive value of basal FSH over age on ovarian response. However, they did not study the effect on cancellation rate (a major problem in poor responders), fertilisation rate or pregnancy rate (due to relatively small sample size). We report here on the analysis of the predictive value of basal FSH and age on cancellation rate, total dose of gonadotrophins, number of oocytes collected, normal fertilisation rate and pregnancy rate in 344 women undergoing their first IVF cycle after pituitary desensitisation. METHODS We studied 344 consecutive women undergoing their first IVF cycle at our unit between August 1994 and July 1996. Only first cycles were studied to avoid possible bias from experience of ovarian response in previous cycles. The indications for treatment were: tuba1 damage (40%), sperm dysfunction (35%) endometriosis (7%), unexplained infertility (13%) and ovulatory dysfunction (5%). Age was calculated as completed years on the day of starting ovarian stimulation. The serum FSH concentration was measured on day 3 of the period preceding the treatment cycle. Hormone assays Serum FSH concentration was measured using the Ciba Coring ACS FSH assay; a two-site chemiluminometric (sandwich) immunoassay using ACS 180 immunoanalyser (Ciba Coring Diagnostic Corporation, Massachusetts, USA). The assay was standardised against the World Health Organisation 2nd IRP 94/632 reference material. The in-house withm-batch coefficients of variation for this assay (determined by replicate analysis) were 456% at 5.13 IUL (n = 12), 4.06% at 16.56 IUL (n = 12) and 4.92% at 47 IUL (n = 12). The betweenbatch corresponding values were 7.32% at 8.64 IUL (n = 127), 6.61% at 14-71 IU/L (n = 52) and 9.53% at 51.04 IUL (n = 107). In our laboratory the upper limit of normal (95th centile) for early follicular phase FSH was 10.8 IUL. Treatment protocol We followed protocols for desensitisation, ovarian stimulation, egg collection, gamete handling, embryo transfer and luteal support as described previou~ly~>~. Basically, gonadotrophin releasing hormone analogue (Nafarelin, Syntex, Berkshire, UK), 200 pg three times daily via nasal spray, was used for pituitary desensitisation (long protocol; starting from the mid-luteal phase of the previous cycle), followed by ovarian stimulation with human menopausal gonadotrophin (Metrodin HP, Serono, Welwyn Garden City, UK). We used a flat-dose protocol where a constant daily dose of gonadotrophin was administered according to the woman s age; three ampoules (75 IU each) for women aged 35 years or younger, four ampoules for women aged between 36 and 39 years, and six ampoules for women 40 years and older. Ovarian response was monitored using vaginal ultrasound scan. Human chorionic gonadotrophin (hcg) (Profasi, Serono, UK) 10,000 units were administered when there was a leading follicle of 20-22 mm mean diameter and at least another three follicles over 14 mm. Twenty-eight women with three or fewer follicles > 14 mm mean diameter after 14 days of stimulation were considered to have poor response and counselled regarding cancelling their treatment cycle; all agreed to cancel treatment. Ultrasound directed oocyte retrieval was performed 36 hours after the hcg injection. Embryo transfer was performed 48-54 hours following oocyte retrieval and progesterone (Gestone, Paines and Byrne, Surrey, UK) 100 mg/day (intramuscular) was given for luteal support. A urinary pregnancy test was performed two weeks following the embryo transfer and, if positive, a scan was performed at four weeks post-transfer. A clinical pregnancy was defined as that with fetal heart motion on ultrasound scan and a biochemical pregnancy as that where miscarriage has occurred before demonstrating fetal heart motion on scan. Analysis of results The end points studied were: cancellation rate for poor response, total dose of gonadotrophins used (calculated as the total number of ampoules used), number of oocytes collected, normal fertilisation rate (number of embryos with two pronuclei on day 1 following egg collection divided by the number of oocytes collected) and pregnancy rate per cycle. Dichotomous categorical variables were used to describe and analyse the data on cancellation rate and pregnancy rate. Continuous variables were used with the other end points. The independent variables studied were the woman s age and her basal FSH concentration. Continuous variables were examined for normalcy; logarithmic transformation was used where necessary to produce acceptable Gaussian distribution. The variables of age and basal FSH were banded for analysis into arbitrary interval groups. For age, there were four age groups: group 1 5 30 years (n = 11 1, 32.3%); group 2 > 30-35 years (n = 119, 34.6%); group 3 > 35 to < 40 years (n = 82,23.8%); and group 4 2 40 years (n = 32, 9.3%). For FSH, there were three groups: FSH group 1 5 5.4 IU/L (n = 39, 11.3%); FSH group 2 > 5.4 to < 10.8 IU/L (n = 273,79.4%); and FSH group 3 1 10.8 IU/L (n = 32,9*3%).

~~ ~ AGE AND FSH AS PREDICTORS OF IVF OUTCOME 109 Table 1. Age, basal serum follicle stimulating hormone (FSH), and number of eggs collected, embryos formed, and embryos transferred in the study population (n = 344). Median (range) Age (years) 33 (22-46) Basal FSH (IU/L) 7.25 (3-27) Total dose 33 (11-104) Eggs collected 9 (1-34) Fertilisation rate (%) 53 (0-100) Embryos formed 4 (0-23) Embryos transferred 3 (1-3) Interquartile range 29-37 6-8.4 27-44 5-14 3 1-75 2-8 2-3 The effects of basal FSH concentration and age on the outcome variables were analysed using the two-sample t test and x2, as appropriate. In variables with which both FSH and age were found to have significant association, the relationship was further examined using a two-way analysis of variance (factorial ANOVA) employing the general linear model for continuous variables (total dose and number of oocytes) and logistic regression for binary variables (pregnancy). The software Stat- View (version 4.5, Abacus Concepts, California, USA) was used for analysis and statistical significance was assumed at P 0.05. RESULTS All the 344 women had successful pituitary down regulation and started ovarian stimulation. Twenty-eight women (8.1%) had cancelled cycles because of poor ovarian response and did not proceed to egg collection. The remaining 316 women (9109%) had egg collection. Thirty-five women (10.2%) had total failure of fertilisation and the remaining 281 (81.7%) proceeded to embryo transfer. Of these, 178 (51.7%) had a negative pregnancy test and the remaining 103 had a positive test, yielding a 29.9% pregnancy rate per cycle started. Of these pregnancies 82 were clinical (2343% clinical pregnancy rate per cycle), 18 (5.2%) were biochemical, and three (047%) were ectopic. Table 1 shows the distribution of the age, basal FSH concentration, and the number of oocytes collected, embryos formed and embryos transferred in the study population. As shown in Fig. la, the cancellation rate in FSH groups 1 and 2 (normal values) were 5.1% and 54?%, respectively. In FSH group 3 it was 31-2%, considered a highly significant difference (P = 0.0001). On the other hand, as shown in Fig. 2a, there was no significant difference between the cancellation rate between different age groups (P = 0.98). Increased basal FSH concentration, as well as increased age, were significantly associated with increasing total gonadotrophin dose required (Figs lb and 2b), and a reduction in the number of oocytes Table 2. Two-way analysis of variance: probability of main effects. Age FSH AgexFSH Total dose < 0~0001 < 0~0001 0.0 1 No. of eggs collected 0.07 0.002 0.8 collected (Figs lc and 2c). Of course, by virtue of our treatment protocol-where older women were given higher daily doses-there would be a significant increase in total dose with increased age, and no firm conclusions could be drawn from our study about the effect of age on the total gonadotrophin dose required. However, it is worth noting that this significant association (P = 0.0001) was also present between age groups one (I 30 years) and two (> 30 and I 35); women in both these groups received the same daily dose in our protocol, yet their total dose significantly increased with increasing age. Neither basal FSH concentration, nor the woman's age were significantly correlated with the normal fertilisation rate, as evident by the wide and overlapping confidence intervals in Figs Id and 2d. This applied whether all the couples were considered or only those with no male factor infertility (data not shown). In order to examine the effect of increased basal FSH concentration and age on pregnancy rate, these variables were analysed in a dichotomous categorical fashion; as normal and abnormal (2 10.8 IUL) for FSH, and as I 35 or > 35 years for age. This was necessary to increase the power of the analysis due to the relatively small numbers. Both variables showed a statistically significant effect on pregnancy rate (Figs le and 2e). Thus, significant associations were found between both basal FSH concentration and age on one hand and the total dose and the number of oocytes collected on the other hand. Two-way analysis of variance was then employed in order to examine the independent (main) as well as the combined effect (interaction) of FSH and age (Table 2). The analysis of variance was applied about the age of 40 years and FSH of 10.8 Iu/L (both corresponding to the 95th centile). Both basal FSH concentration and age showed significant and independent associations with the total dose. Basal FSH, but not age, showed significant independent association with the number of oocytes collected. Logistic regression analysis showed that age (P = 0-01), but not basal FSH concentration (P = 0-l), was significantly associated with pregnancy rate; the significant association between FSH and pregnancy rate disappeared when controlled for age. DISCUSSION The results of this study have shown that age and basal FSH are predictive factors of the outcome of IVF

110 K. SHARIF J* ET AL 6o r -1 0 1 2 3 1 2 3 1 2 3 1 2 3 40 v 8 30 c Q & 10 n I I I Normal Abnormal Fig. 1. Outcome measures (geometric means and 95% CI) related to FSH groups (group 1 I5.4 IU/L, group 2 > 5.4 to < 10.8 IU/L, group 3 2 10.8 IU/L; groups 1 and 2 normal, group 3 abnormal). *P= 0.0001, tp = 0.007 (x2 test); $ P = 0.001, P= 0.0001 (two-sample 1 test). treatment following pituitary desensitisation. Their predictive values, however, seem to manifest in different ways. This may have several implications for counselling and clinical management, particularly for women younger than 35 years with abnormally high basal FSH concentration who would have been expected to perform differently on age basis alone. Cancellation rates for poor response were similar (7% to 8%) between different age groups, but differed significantly with basal FSH, rising from about 5% in women with concentration < 10.8 IU/L to over 30% in those with 2 10-8 IU/L. To the best of our knowledge, this is the first reported study addressing this issue. Women with high basal FSH concentration should be counselled accordingly. The total gonadotrophin dose required to achieve follicular maturity increased significantly with increasing both age and basal FSH. As discussed above, part (but not all) of this increase with age could be explained by our treatment protocol. The increase with basal FSH, however, was highly significant and independent of the effect of age (Table 2). Drug cost forms a major component of the total cost of IVF treatment, and women with high basal FSH, particularly if older than 35 years, should be alerted to the high possibility of needing higher (more costly) total dose. Despite using higher total gonadotrophin dose with advancing age and increased basal FSH concentration, these two parameters were significantly associated with fewer oocytes retrieved. The effect of basal FSH concentration was independent of and more significant than that of age. Fewer oocytes will lead to fewer embryos available for either transfer or cryopreservation. Women with high basal FSH concentration should be warned that it is likely that they will have fewer oocytes collected and fewer (or perhaps no) embryos for cryopreservation.

AGE AND FSH AS PREDICTORS OF IVF OUTCOME 111 14 c I I I I 1 2 3 4-70 u) - a, 3 g 60-5 50 Q u) - 40 m c ' 30 6 70 a, L I I I I I 1 2 3 4 T T T 1 I I I - * o o 1 2 3 4 101 I I >35 135 Weeks Fig. 2. Outcome measures (geometric means and 95% CI) related to age groups (group 1 I 30 years, group 2 > 30-35 years, group 3 > 35 to < 40 years, group 4 2 40 years). *P = 0.0001, tp = 0.0007 (two-sample t test);$ P = 0.01 (x2 test). Interestingly, the normal fertilisation rate did not differ significantly with either age or basal FSH concentration. This finding is in apparent contrast to that by Toner et al6 whose results indicated that both increased age and basal FSH were significantly associated with reduced number of fertilised oocytes. A possible explanation of this difference may lie in the fact that we have looked at the fertilisation rate while they have looked at the number of fertilised oocytes; fewer oocytes collected (which is the case with increased age and basal FSH) will lead to fewer embryos formed even in the presence of constant fertilisation rate. We have also found that the pregnancy rate per cycle decreased significantly with increasing both basal FSH concentration and age. Logistic regression analysis showed that the effect of basal FSH was not independent; increasing age alone could explain the associated decrease in pregnancy rate. This finding was in contrast to that by Toner et al. who found that basal FSH concentration was independent of, and stronger than age in significantly predicting pregnancy rate after IVF6. Their sample size (n = 1478) was much larger than ours and this may explain the apparently conflicting findings. An alternative explanation might be that all our patients had prior pituitary desensitisation compared with about one-third of their patients. Pituitary desensitisation may in some way modulate the effect of high basal FSH concentration on pregnancy rate. A larger study than ours is needed to clarify this point. We did not examine the association with miscarriage rate, clinical pregnancy rate or live birth rate. The numbers in our study were too small to make such analysis statistically meaningful.

112 K. SHARIF ET AL. Some statistical issues in our study deserve comment. When similar data have been handled in the literature previously, two statistical methods have been used. The first is linear regression analysis, where full advantage is taken of continuous data with out introducing arbitrary cut off points (e.g. different FSH and age group) as these may bias the results6. The second method used is to divide the continuous variables into banded values and analyse them as groups. We have elected to use the latter method in presenting these data for several reasons. First, clinicians usually think of, and work with, numerical results in a categorical fashion: normal or abnormal FSH (> 95th centile). Second, we believe that this method lends itself to more clarity when presented in graphs. Finally, we have also analysed our data by regression analysis (results not shown) and found similar results. The difference is in the form, not the substance. Currently it is not known whether the ovarian response and IVF outcome of women with high basal FSH concentration could be improved by altering the stimulation regimen (e.g. using the flare-up protocol or increasing the dose of gonadotrophin). The available data relate to high dose stimulation in those who have shown poor response in previous IVF cycles, regardless of their basal FSH concentrationlo. These data suggest that although the number of oocytes is significantly increased, the pregnancy rate remained very low lo. Currently some IVF clinicians would choose to extrapolate from those data and keep women with high basal FSH undergoing their first IVF cycle on the standard regimen, while others would increase their stimulation dose. The correct action is not known and perhaps the best action is to enrol these in a prospective randomised trial. Finally, a few practical points are worth mentioning in the context of measuring basal FSH concentration. First, we have measured FSH on menstrual day 3 to achieve consistency and comparability with previous studies. Nonetheless, it has been shown that the variability in FSH concentrations during menstrual days 2-5 is not significant, and it could be measured in practice on any of those days*. Second, basal FSH concentration does not have to be measured during the cycle immediately preceding the IVF treatment; the small intercycle variability is such that abnormally high levels in any previous cycle carry the same prognostic significancei2. The final point is to do with the upper limit of normal: it is very different in different assays (ranging from 9 to 22 IU/L)6,7. Clinicians should be aware of the normal limits in their local laboratories. CONCLUSION Our study has demonstrated that, in IVF treatment following pituitary desensitisation, basal FSH concentration is a better predictor of cancellation rate and of the number of oocytes collected than age, but age is a stronger predictor of pregnancy rate. The effect of basal FSH on pregnancy rate lost its significance when controlled for age. Many prognostic factors have been shown to correlate with the chances of achieving pregnancy after IVF treatment. These include the duration of infertility, previous pregnancy, the number and outcome of previous IVF attempts, and the woman s age. We recommend the addition of basal FSH concentration to those factors to individualise pre-treatment counselling for patients. Acknowledgement We would like to thank the clinical, nursing and laboratory staff at the Assisted Conception Unit for their assistance in collecting these data. References Federation CECOS, Schwartz D, Mayaux BA. Female fecundity as a function of age: results of artificial insemination in 2193 nulliparous women with azoospermic husbands. N Engl J Med 1982; 306: 404406. Templeton A, Moms JK, Parslow W. Factors that affect outcome of in-vitro fertilisation treatment. Lancet 1996; 348: 1402-1406. Sherman BM, Korenman SG. Hormonal characteristics of the human menstrual cycle throughout reproductive life. J Clin Invest 1975; 55: 699-706. Lenton EA, Sexton L, Lee S, Cook ID. Progressive changes in LH and FSH and LH: FSH ratio in women throughout reproductive life. Maturitas 1988; 103543. Scott RT, Toner JP, Muasher SJ, Oehninger S, Robinson S, Rosenwaks 2. Follicle-stimulating hormone levels on cycle day 3 are predictive of in vitro fertilization outcome. Ferfil Steril 1989; 51: 65 1-654. Toner JP, Philput CB, Jones GS, Muasher SJ. Basal follicle stimulating hormone level is a better predictor of in vitro fertilisation performance than age. Feriil Steril1991; 55: 784-791. Cahill DJ, Prosser CJ, Wardle PG, Ford WCL, Hull MGR. Relative influence of serum follicle stimulating hormone, age, and other factors on ovarian response to gonadotrophin stimulation. Br J Obstet Gynaecoll994; 101: 999-1002. Sharif K, Afnan M, Lenton W. Mock embryo transfer with a full-bladder immediately before the real transfer for in-vitro fertilisation treatment: the Birmingham experience of 113 cases. Hum Reprod 1995; 10: 1715-1718. Sharif K, Afnan M, Lenton W, Bilalis D, Hunjan M, Khalaf Y. Transmyometrial embryo transfer following difficult immediate mock transcervical transfer. Fertil Steril1996: 56: 107 1-1 074. 10 Land JA, Yarmolinskaya MI, Dumoulin JCM, Evers JLH. High-dose human menopausal gonadotropin stimulation in poor responders does not improve in vitro fertilization outcome. Fertil Steril 1996; 65: 96 1-965. 11 Hansen LM, Batzer FR, Gutmann JN, Corson SL, Kelly MP, Gocial B. Evaluating ovarian reserve: follicle stimulating hormone and oestradiol variability during cycle days 2-5. Hum Reprod 1996; 11: 486489. 12 Scott RT, Hofmann GE, Oehninger S, Muasher S3. Intercycle vanability of day 3 follicle-stimulating hormone levels and its effect on stimulation quality in in vitro fertilization. Fertil Steril 1990; 54: 297-302. Received 26 February 1997 Returnedfor revision 4 July 1997 Accepted 23 July 1997