Previous cycle cancellation due to poor follicular development as a predictor of ovarian response in cycles stimulated with gonadotrophin-releasing hormone agonist-gonadotrophin treatment

Joana Peñarrubia1, Francisco Fábregues1, Dolors Manau1, Montserrat Creus1, Francisco Carmona1, Roser Casamitjana2, Juan A. Vanrell1 and Juan Balasch1,3

1 Institut Clínic of Gynecology, Obstetrics and Neonatology and 2 Hormonal Laboratory, Faculty of Medicine—University of Barcelona, Hospital Clínic-Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona, Spain

3 To whom correspondence should be addressed at: Institut Clínic of Gynecology and Obstetrics, Hospital Clínic, C/Casanova 143, 08036 Barcelona, Spain. Email: jbalasch{at}ub.edu


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: There is scanty information analysing the predictive value of a poor response, in terms of cancellation of the IVF cycle because of poor follicular development, as a predictor of ovarian response in a subsequent treatment cycle. This study, where logistic regression analysis was used, was undertaken to investigate the relative power of the woman's age, basal FSH, and previous cycle cancellation both as single and combined predictors of ovarian response in an IVF program where pituitary desensitization is routinely used. METHODS: One hundred and twenty-nine consecutive patients having their first cycle of IVF/ICSI treatment cancelled because of poor follicular response and undergoing a second attempt within 6 months after the failed treatment cycle were initially selected (group 1). Group 2 comprised 129 patients undergoing the first cycle of IVF/ICSI treatment and who were randomly selected from our assisted reproductive treatment program matching by BMI and indication for IVF/ICSI to those in group 1. RESULTS: Cancellation rate was significantly higher but ovarian response significantly lower in group 1 as compared with group 2. As indicated by the AUCROC determined with ROC analysis, such a poor outcome in patients having a previous IVF/ICSI cycle cancelled due to poor response was observed whatever the level of basal FSH. In a logistic regression analysis and according to the odds ratio values, the predictive capacity of a previous poor response was 9 and 7.6 times higher than the predictive capacity of age and basal FSH, respectively. Any two or all three variables studied did not improve the predictive value of previous cycle cancellation alone. CONCLUSIONS: The history of an IVF/ICSI cancelled cycle due to poor follicular response in a standard stimulation protocol is a better predictor of cancellation in subsequent treatment cycles than age or FSH. The poor ovarian response associated with previous cycle cancellation occurs whatever the level of basal FSH.

Key words: age/basal FSH/IVF/ovarian reserve/poor response


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The application of gonadotrophin administration to stimulate multiple follicular development and the subsequent transfer of multiple embryos has dramatically increased the success of assisted reproductive technologies (ART). The ability of the ovaries to respond to gonadotrophins with adequate follicular development has been referred to as ovarian reserve. Remarkably, a number of women are found to respond poorly or not at all to this treatment. Such patients, whose prevalence is estimated to be 5–20% of the ART population, have been referred to as low responders and having diminished ovarian reserve (Fasouliotis et al., 2000Go; Mahutte and Arici, 2002Go; Tarlatzis et al., 2003Go). Gonadotrophin treatment in low responders may result in cycle cancellation, insufficient response with poor follicle recruitment, low levels of serum estradiol, and low pregnancy rates. This phenomenon was first described by Garcia et al. (1983)Go in patients with peak estradiol levels <300 pg/ml and decreased follicular response, expressed as fewer retrieved and fertilized oocytes and also fewer transferred embryos. Thus, such patients are unable to take full advantage of ART.

Importantly, ovarian reserve declines with age but it is a biological and not just a chronological function. The most important aspect of diminished ovarian reserve is that the timing of its onset is highly variable (Scott and Hofmann, 1995Go). Therefore, as ART are expensive, time-consuming, and stressful for patients, there is a need for predicting ovarian response. Apart from female age, a plethora of predictive tests for low ovarian response have been described, including both basal and dynamic hormone testing, ultrasonographic studies, and even ovarian biopsy (Scott and Hofmann 1995Go; Fasouliotis et al., 2000Go; Bukman and Heineman, 2001Go; Kligman and Rosenwaks, 2001Go). However, no test is absolutely predictive and it has been recently stressed that the poor responder is revealed definitively only during ovarian stimulation (Karande and Gleicher, 1999Go; Tarlatzis et al., 2003Go).

The above notwithstanding, chronological age and basal FSH are the most commonly used predictive tests in daily clinical practice. Recent evidence indicates that basal FSH level is the better predictor of egg production capacity whereas age is the better predictor of egg quality (Chuang et al., 2003Go; Toner, 2003Go; van Rooij et al., 2003Go). In a recent study (Klinkert et al., 2004Go), the ovarian response in subsequent cycles was investigated in patients with a poor response (defined as the collection of <4 oocytes) in their first IVF cycle. In that study it was concluded that patients with an expected poor response (defined as those having a basal FSH level ≥15 IU/l and/or aged ≥41 years) in the first cycle should be advised to withdraw from treatment after the first cycle because of a poor prognosis. In contrast, most patients with an unexpected poor response (defined as those aged <41 years with FSH levels <15 IU/l) were found to have a normal response in the second cycle.

However, there is no report analysing the predictive value of a poor response in terms of cancellation of the cycle because of poor follicular development as a predictor of ovarian response in a subsequent treatment cycle. Therefore the current study, where logistic regression analysis was used, was undertaken to investigate the relative power of the woman's age, basal FSH, and previous cycle cancellation in predicting ovarian response in an assisted reproduction program where pituitary desensitization is routinely used. These prognostic factors were studied as single and combined predictors. In addition, we used a statistical method using receiver-operating characteristic (ROC) curve analysis to test the usefulness of basal FSH serum concentrations to discriminate between cancelled versus non-cancelled cycles and conception versus non-conception cycles. It is noteworthy that in ROC curve analysis, many efficiencies of all decision levels can be calculated, resulting in an overall quantification of accuracy which is not affected by the prevalence of a condition. Also, ROC plots provide a pure index of accuracy by demonstrating the limits of a test's ability to discriminate between alternative states of health over the complete spectrum of operating conditions (Hanley and McNeil, 1982Go; Zweig and Campbell, 1993Go).


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Patients studied
The study was a retrospective analytic investigation including 258 women undergoing IVF/ICSI treatment at the Fertility Unit of the Hospital Clínic of Barcelona. From January 2000 to December 2001, 129 consecutive patients having their first cycle of IVF/ICSI treatment cancelled because of poor follicular response and undergoing a second attempt within 6 months after the failed treatment cycle were initially selected (group 1). All patients in group 1 underwent a second IVF/ICSI attempt between March 2000 and April 2002. Group 2 comprised 129 patients undergoing the first cycle of IVF/ICSI treatment between January 2002 and November 2002 and who were randomly selected from our assisted reproductive treatment program matching by body mass index (BMI) (±2 kg/m2) and indication for IVF/ICSI to those in group 1. Matching for these variables is a feature of the present study as both BMI and cause of infertility may influence ovarian response to gonadotrophins (Crosignani et al., 1994Go; Roseboom et al., 1995Go; Tinkanen et al., 1999Go; Loh et al., 2002Go).

All patients had both ovaries, no previous ovarian surgery, and normal ovulatory function according to midluteal serum progesterone concentrations and regular menses (every 25 to 34 days). All subjects had blood samples drawn on day 2 or 3 of their cycle within 3 months of the IVF/ICSI attempt for assay of basal FSH, LH and estradiol.

Institutional review board approval was not required because our assisted reproduction unit is licensed and regulated by the Spanish Health Authorities and there were no interventions other than those for standard IVF/ICSI. The study concerned a retrospective analysis of data that were recorded anonymously.

Stimulation regimens
In the first IVF/ICSI treatment cycle in patients in group 1 and in patients in group 2 ovarian stimulation was carried out with FSH under pituitary suppression with GnRH agonist according to a routinely used protocol previously reported (Balasch et al., 2001Go). Pituitary suppression was achieved by s.c. administration of leuprolide acetate (Procrin; Abbott Laboratories, Madrid, Spain). This treatment was started in the midluteal phase of the previous cycle and given 1 mg daily, then reduced to 0.5 mg after ovarian arrest was confirmed. Gonadotrophin stimulation of the ovaries was started when serum estradiol concentrations declined to <50 pg/ml and a vaginal ultrasonic scan showed an absence of follicles >10 mm diameter. On days 1 and 2 of ovarian stimulation, six and four ampoules/day of recombinant human FSH (75 IU per ampoule) (Gonal-F, Serono S.A., Madrid, Spain) respectively, were administered subcutaneously. On days 3 and 4 of ovarian stimulation, two ampoules per day of FSH were administered to each patient. From day 5 onward, FSH was administered on an individual basis according to the ovarian response as assessed by sequential transvaginal ultrasonography and serum estradiol measurements.

In the second IVF/ICSI treatment cycle in patients in group 1 high-dose human menopausal gonadotrophin (HMG) in association with a reduced dose leuprolide protocol were used for ovarian stimulation. Patients received 0.5 mg/day of leuprolide from the midluteal phase until pituitary suppression was achieved and then 0.25 mg/day thereafter. On days 1 and 2 of ovarian stimulation, eight ampoules/day of HMG (Pergonal, Serono S.A.) were administered i.m. On days 3 and 4 of ovarian stimulation, four ampoules per day of HMG were administered to each patient. From day 5 onward, HMG was administered on an individual basis according to the ovarian response.

The criteria for HCG administration (5000 IU; Profasi, Serono S.A.) were the presence of 2 or more follicles ≥18 mm in diameter with ≥4 follicles measuring ≥14 mm in association with a consistent rise in serum estradiol concentration. The cycle was cancelled when there were <3 follicles with diameter ≥14 mm after 8–9 days of gonadotrophin therapy or after 4–5 additional treatment days without attaining, or the imminent prospect of attaining, the criteria for HCG administration.

Oocyte aspiration was performed with vaginal ultrasonography 35–36 h after HCG administration. Embryo grading was recorded according to published criteria (Veeck, 1999Go); embryos of (Veeck) grades 1 or 2 were considered high quality. Two to three days after oocyte recovery, up to three embryos per patient (depending on the age of the patient, the indication for IVF/ICSI, and the number and quality of embryos available per replacement) were replaced and the luteal phase was supported with additional doses of HCG or vaginal micronized progesterone according to ovarian response. Pregnancy was diagnosed by incresing serum concentrations of {beta}-HCG after embryo transfer, and the subsequent demonstration of an intrauterine gestational sac by ultrasonography.

Hormone assays and ultrasonography
Hormones were measured using commercially available kits as reported previously (Balasch et al., 2001Go). Estradiol concentrations in serum were estimated by a competitive immunoenzymatic assay (Immuno 1, Technicon; Bayer, Tarrytown, NY). The sensitivity was 10 pg/ml and the interassay coefficient of variation (CV) was 5%. FSH and LH serum concentrations were measured by an immunoenzymatic assay with two mAb (Immuno 1, Technicon; Bayer) and data expressed in terms of IRP 78/549 and 68/40, respectively. The sensitivity of the assays was 0.1 IU/l for FSH and 0.3 IU/l for LH, and interassay CV were 2.7 and 3.1%, respectively. Total {beta}-HCG was measured by a solid-phase, two-site chemiluminiscent enzyme immunometric assay standardized against the Third International Standard 75/537 (Immulite, Diagnostic Products Co., Los Angeles, CA), with a detection limit of 2 IU/l. The inter-assay CV was 5.8%.

Ultrasonic scans were performed using a Toshiba Eccocee SAA-340A/EF unit (Toshiba Co., Tokyo, Japan) equipped with a 5–7 MHz endovaginal probe (PVF-641VT).

Statistics
For statistical analysis the Student's t-test and the chi-square test were used as appropriate. Results are expressed as mean±SD. P<0.05 was considered significant. The discrimination attained between cancelled versus non-cancelled cycles, and conception versus non-conception cycles in the second treatment attempt in patients in group 1 was evaluated with ROC analysis (Hanley and McNeil, 1982Go; Zweig and Campbell, 1993Go). Sensitivity, specificity, and the area under the ROC curve (AUCROC) were obtained for each model. 95% confidence intervals (CI) were calculated for each of the estimates. The models' AUCROC values were compared using the method of Hanley and McNeil (1982)Go. Calculation of the AUCROC provides the quantitative measure of accuracy, i.e. the ability of a particular parameter (e.g. basal FSH serum concentrations) to discriminate between two conditions (e.g. cancelled versus non-cancelled cycles). An ROC curve representing a parameter with no discrimination at all is a 45° diagonal line from the left lower corner (0% true positive rate and 0% false positive rate) to the upper right corner (100% true positive rate and 100% false positive rate) with an area under the curve of 0.5. A parameter with no overlap between the two conditions will discriminate perfectly and has an ROC curve passing along the y-axis to the upper left corner (100% true positive rate and 0% false positive rate) to end again in the upper right corner with an area under the curve of 1.0.

Logistic regression analysis was performed using data from patients in group 2 and the second treatment attempt in patients in group 1. The relationship of age, basal FSH and previous cycle cancellation to the outcome of ovarian response (cancelled versus punctured cycle) was analysed in a logistic regression model to assess the joint effect of variables and to estimate the sensitivity and specificity values for all possible combinations of the study variables. The best joint model was obtained by an enter procedure considering P<0.05 as the P-value associated with a variable to enter in the model. The results are given in terms of odds ratio (OR) and 95% CIs, which is the measure of effects immediately given by the logistic regression with OR > 1.0 corresponding to positive associations and OR < 1.0 corresponding to negative associations. CIs which did not include the value of 1.0 correspond to P-values < 0.05 and, hence, were considered statistically significant. Data were analysed by the Statistical Package for Social Sciences (SPSS; Chicago, IL) statistical software.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Results are presented in Tables I and II and Figure 1. Table I shows patient characteristics, gonadotrophin treatment, and ovarian response in the two groups studied (data for patients in group 1 belong to the second IVF/ICSI treatment cycle). Mean age and BMI were not statistically different in groups 1 and 2 although, as expected, mean age was somewhat higher in group 1. Indications for assisted reproductive treatment were obviously identical for both groups. Basal hormone measurements were also similar in groups 1 and 2. As expected, cancellation rate was significantly higher in group 1 where as many as 54.2% of cycles were cancelled because of low response to gonadotrophin ovarian stimulation. Despite the fact that duration of stimulation was similar in both groups, the total amount of gonadotrophins used was significantly higher in group 1. This notwithstanding, parameters of ovarian response including peak estradiol concentration on the day of HCG injection, the total number of developing follicles, and the oocyte yield were all significantly lower in group 1. Accordingly, the embryo yield and quality and pregnancy rate were also significantly lower in group 1 where 34% (20/59) of patients undergoing embryo transfer had only 1–2 embryos available, and only 13.1% pregnancies per started cycle were obtained. The interval between IVF/ICSI and embryo transfer and the outcome of pregnancies were similar in groups 1 and 2 (Table I).


View this table:
[in this window]
[in a new window]
 
Table I. Patient characteristics, gonadotrophin treatment and ovarian response in two groups studied

 

View this table:
[in this window]
[in a new window]
 
Table II. Logistic regression analysis for age, cycle day 3 FSH and previous cancelled cycle and all possible combinations as prognostic indicators of ovarian response in ART cycles in the whole population studied

 


View larger version (11K):
[in this window]
[in a new window]
 
Figure 1. Receiver operating characteristic curves analysing the value of cycle day 2–3 serum FSH for discriminating (A) cancelled versus non-cancelled cycles, and (B) conception versus non-conception cycles in the second IVF/ICSI treatment cycle in patients in group 1. Discriminatory capacity of FSH serum values obtained in the basal hormone measurement preceding the first (solid line) and the second (dashed line) IVF/ICSI cycles is presented.

 
Such a poor outcome in patients having a previous IVF/ICSI cycle cancelled due to poor follicular response was observed whatever the level of basal FSH. To analyse the diagnostic accuracy of basal FSH to discriminate between cancelled versus non-cancelled cycles, and conception versus non-conception cycles in the second IVF/ICSI attempt in group 1, the AUCROC determined with ROC analysis is presented in Figure 1. Clearly serum FSH concentration measured on cycle day 2–3 preceding the first IVF/ICSI cycle was unable to discriminate between cancelled and non-cancelled cycles (AUCROC=0.53; 95% CI: 0.43–0.69) or conception versus non-conception cycles (AUCROC=0.57; 95% CI: 0.40–0.74). The same was true when basal FSH measured preceding the second assisted reproduction treatment cycle was considered (AUCROC for cancelled versus non-cancelled cycles=0.52; 95% CI: 0.42–0.70) (AUCROC for conception versus non-conception cycles=0.56; 95% CI: 0.41–0.74).

Table II shows logistic regression analysis and individual probability estimates for age, basal FSH and previous cancelled cycle and for all possible combinations of the three variables regarding outcome of ovarian stimulation in the whole population studied (considering the second treatment cycle for patients in group 1). On the basis of this study, the history of an IVF/ICSI cancelled cycle is a better predictor of cancellation in subsequent treatment cycles than age or FSH are. Thus, according to the ORs the predictive capacity of a previous poor response is 9 and 7.6 times higher than the predictive capacity of age and basal FSH, respectively. The analysis of the impact of age or basal FSH within the group defined by a previous cancelled cycle did not further delineate any group with higher or lower chances of cancellation. Finally, analysis of the groups created by the simultaneous evaluation of age, basal FSH and previous poor response did not delineate any group beyond that provided by previous cancellation alone.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The success of ART depends on the ability of the ovaries to respond to gonadotrophin stimulation and to develop several follicles leading to adequate oocyte yield. That response reflects the ovarian reserve and depends on the pool of primordial follicles in the ovaries. Ageing of an ovary is characterized by the reduction of the number of primordial follicles from about a million at birth to about 250 000 at menarche, to a very few at the end of reproductive life (Lass, 2001Go). This loss accelerates around the age of 35–37 years and precedes the menopause by 10–12 years but importantly, there is wide variation between women in the number and rate of depletion of follicles (Lass, 2001Go). Thus, although it is clear that fertility declines with age, the clinically relevant question of a given woman's fertility potential at a given age is far more difficult to answer. In this respect, it would certainly be ideal to be able to predict a poor response during routine screening so that such patients could be directed toward a more appropriate stimulation regimen.

Over recent years there has been a great interest in developing screening tools to assess a woman's fertility potential independent of her age. The first step in developing a test to predict fertility requires the identification of that component of the reproductive axis primarily responsible for the decrease in fecundity. The dramatic success of oocyte donation, even in women of advanced age, has clearly demonstrated that the decrease in fertility potential is predominantly caused by ovarian rather than uterine factors (Barnhart and Osheroff, 1998Go). This understanding has directed the search for screening tools towards tests which would identify women with advanced depletion of follicles or diminished oocyte quality.

An accurate test of ovarian reserve has always been an attractive proposition to clinicians because of the attendant advantages of assessing the likelihood of a sufficient ovarian response prior to commencing ART treatment. Woman's age seems to affect egg quality more than quantity (Toner, 2003Go). Thus, in a recent study it was shown that younger women with reduced ovarian reserve had a higher risk of cycle cancellation and produced fewer eggs than did women older than 40 years of age, but once eggs were retrieved, the younger women had near-normal implantation and pregnancy rates (van Rooij et al., 2003Go). However, age and regularity of menses alone are unreliable measures of predicting ovarian reserve and the biological age is more important than the chronological age (Lass, 2001Go).

Tests of functional reserve of the ovaries can often be used to predict low response to standard ovarian stimulation protocols. These include baseline serum concentrations of hormones such as FSH, LH, estradiol and inhibin B as well as FSH/LH ratio (Sharara and Scott, 1997Go; Karande and Gleicher, 1999Go; Lass, 2001Go). However, non-response to ovarian stimulation in normogonadotropic, normogonadal women has been reported (Farhi et al., 1997Go). Therefore, it has been stressed that consideration of the glandular response to stimulation may be necessary for evaluation of the reserve of the endocrine organ in question (Kim, 1998Go).

Dynamic tests of ovarian reserve include the clomiphene citrate challenge test, the GnRH agonist stimulation test, and the exogenous follicle stimulating hormone ovarian reserve test. Ultrasonographic tests have also been suggested to predict ovarian response (Karande and Gleicher, 1999Go; Bukman and Heineman, 2001Go; Lass, 2001Go) and thus it has been reported that the number of antral follicles as counted early in the follicular phase provides better prognostic information on the occurrence of poor response during hormone stimulation for IVF than does the patient's chronological age and the currently used endocrine markers (Scheffer et al., 1999Go; Bancsi et al., 2002Go, Bancsi et al., 2004Go). However, despite the validity of all these tests, there still remain patients who respond poorly to stimulation despite having normal tests of ovarian reserve. This supports the idea that ovarian reserve is not a simple static anatomic number of follicles but rather a dynamic process, the mechanism of which is not yet fully understood (Lass, 2001Go). In fact, it has been recently stressed that the ideal ovarian reserve test is the response of the ovaries to a ‘normal’ or ‘standard’ ovarian stimulation protocol (Karande and Gleicher, 1999Go; Tarlatzis et al., 2003Go).

In the first ART cycle patients in groups 1 and 2 were treated with a step-down regimen of recombinant FSH administration under pituitary suppression. In this tapering regimen the highest dose of FSH is given on stimulation days 1 and 2 (six and four 75 IU FSH ampoules) and is then reduced to two ampoules daily once follicular recruitment has been achieved. This regimen has proved to be clinically efficacious (Davis and Rosenwaks, 1996Go; Balasch et al., 2001Go; Peñarrubia et al., 2003Go) and is further supported by the following. First, it has been shown that for successful induction of multiple folliculogenesis in normally ovulating women, there is a critical period during the early follicular phase of the cycle when FSH values should remain above the physiological level to maximimally stimulate follicle recruitment in the primary cohort (Messinis and Templeton, 1990Go; Lolis et al., 1995Go). Second, follicles recruited by exogenous FSH require an FSH threshold concentration that is higher than that in the natural cycle (Lolis et al., 1995Go). Third, marked patient variability exists in FSH thresholds as well as in FSH metabolic clearance and ovarian sensitivity to FSH (Porchet et al., 1994Go; Ben-Rafael et al., 1995Go; van Santbrink et al., 1995Go). Remarkably, in clinical studies, such a threshold level was reached with a single injection of six ampoules of FSH on cycle day 2 and further growth of the follicles was obtained with extra FSH from cycle day 4 onward at the daily dose of two ampoules (Lolis et al., 1995Go). The above notwithstanding, all patients in group 1 failed to respond in their first treatment cycle.

In their second attempt of ovarian stimulation for ART, patients in group 1 were treated with high dose HMG and a 50% reduction in the standard dose of leuprolide acetate which are both measures proposed for treatment of low responder patients (Davis and Rosenwaks, 1996Go; Fasouliotis et al., 2000Go; Tarlatzis et al., 2003Go). In spite of this, ovarian response and ART outcome were very poor and as many as 54.2% of cycles were cancelled due to poor follicular recruitment whereas pregnancy rate per started cycle was only 13.1%. Notably, the predictive capacity of ovarian response of a previous cancelled cycle was significantly higher than that of basal FSH and age as indicated by the OR, a way of representing probability. The ORs provide an estimate (with CI) for the relationship between two binary (‘yes or no’) variables. In addition, they enable us to examine the effects of other variables on that relationship, using logistic regression (Bland and Altman, 2000Go).

Three features deserve comment in the current investigation. First, GnRH agonists may have a direct, deleterious effect on the ovary, which could be especially important for poor responders. In contrast, GnRH antagonists are given in the late follicular phase of controlled ovarian stimulation and are not involved in the stage of the follicle recruitment, which may be critical for poor responders (van Loenen et al., 2002Go). Thus, it may be postulated that a GnRH antagonist protocol could have changed the results in some women. Preliminary results from our group comparing high dose gonadotropins in association with reduced dose of GnRH agonist versus a multiple dose GnRH antagonist protocol in previous poor responders suggest no significant differences in ovarian response (our unpublished observations). Thus, we agree with those authors stressing that according to available limited data in the literature, there is no advantage from the use of GnRH antagonists (van Loenen et al., 2002Go; Tarlatzis et al., 2003Go). Larger, controlled, prospecitve randomized trials using GnRH antagonists are necessary to investigate this issue. To date, however, no one single adapted stimulation protocol has been proven effective in poor responders in a controlled randomized trial and thus it seems that the protocol, after a certain point, does not make a difference (Mahutte and Arici, 2002Go).

Second, a case-control but not a cohort study design was used. With such a study design a variety of patient characteristics (e.g. variables such as woman's age, basal FSH, and previous cycle cancellation both as single and combined predictors) can be studied in relation to a single outcome (e.g. cycle cancellation), whereas a typical cohort study is used to examine multiple outcomes related to a single variable (Cramer, 1994Go; Grimes and Schulz, 2002Go). In addition, multivariate analysis is a statistical technique for dealing with a number of potential confounding variables simultaneously and is frequently used in case-control studies. After such an analysis, the investigator will present the ‘adjusted OR’ that presumably reflects an unconfounded association (Cramer, 1994Go). Our study design, however, leaves open the question of whether knowing basal FSH or female age may have any modulating effect on the predictive value of previous cycle cancellation.

Finally, among 63 deliveries obtained in the whole population included in this study there were 16 (25%) multiple births. It has been reported that in all situations where more than four embryos were available for transfer, there was no greater birth rate for women receiving three embryos than for those receiving two, but there was a considerable increase in the rate of multiple births, and especially triplets (ESHRE Capri Workshop Group, 2000Go). Remarkably, in the present study twin and triplet rates were similar in previous poor responders (group 1) having reduced embryo yield and quality than in patients undergoing the first IVF/ICSI attempt (group 2). This further stresses the need to limit the number of embryos replaced to reduce the risk of multiple births and supports the idea that in most circumstances two embryos can be transferred without anticipating any significant diminution in the expected pregnancy or birth rates (ESHRE Capri Workshop Group, 2000Go).

In conclusion, this study shows that the history of an IVF/ICSI cancelled cycle due to poor follicular response in a standard stimulation protocol is a better predictor of cancellation in subsequent treatment cycles than age or FSH are. Notably, the poor ovarian response and ART outcome associated with previous cycle cancellation occur whatever the level of basal FSH.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The authors thank Mrs Paquita Antonell for the technical assistance. This work was supported in part by grants from the Instituto de Salud Carlos III (RCMN C03/08), and the Comissionat per a Universitat i Recerca–Generalitat de Catalunya (2001SGR 00372) to J.B.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Balasch J, Vidal E, Peñarrubia J, Casamitjana R, Carmona F, Creus M, Fábregues F and Vanrell JA (2001) Suppression of LH during ovarian stimulation: analysing threshold values and effects on ovarian response and the outcome of assisted reproduction in down-regulated women stimulated with recombinant FSH. Hum Reprod 16, 1636–1643.[Abstract/Free Full Text]

Bancsi LF, Broekmans FJ, Eijkemans MJ, de Jong FH, Habbema JD and te Velde ER (2002) Predictors of poor ovarian response in in vitro fertilization: a prospective study comparing basal markers of ovarian reserve. Fertil Steril 77, 328–336.[CrossRef][ISI][Medline]

Bancsi LF, Broekmans FJ, looman CW, Habbema JD and te Velde ER (2004) Impact of repeated antral follicle counts on the prediction of poor ovarian response in women undergoing in vitro fertilization. Fertil Steril 81, 35–41.[ISI]

Barnhart K and Osheroff J (1998) Follicle stimulating hormone as a predictor of fertility. Curr Opin Obstet Gynecol 10, 227–232.[CrossRef][ISI][Medline]

Ben-Rafael Z, Levy T and Schoemaker J (1995) Pharmacokinetics of follicle stimulating hormone: clinical significance. Fertil Steril 63, 689–700.[ISI][Medline]

Bland JM and Altman DG (2000) The odds ratio. Br Med J 320, 1468.[Free Full Text]

Bukman A and Heineman MJ (2001) Ovarian reserve testing and the use of prognostic models in patients with subfertility. Hum Reprod Update 7, 581–590.[Abstract/Free Full Text]

Cramer DW (1994) Case–control studies of infertility. Infertil Reprod Med Clin N Am 5, 321–335.

Chuang C, Chen C, Chao K, Chen S, Ho H and Yang Y (2003) Age is a better predictor of pregnancy potential than basal follicle-stimulating hormone levels in women undergoing in vitro fertilization. Fertil Steril 79, 63–68.[CrossRef][ISI][Medline]

Crosignani PG, Ragni G, Parazzini F, Wyssling H, Lombroso G and Perotti L (1994) Anthropometric indicators and response to gonadotrophin for ovulation induction. Hum Reprod 9, 420–423.[Abstract]

Davis OK and Rosenwaks Z (1996) In vitro fertilization. In Adashi EY, Rock JA and Rosenwaks Z (eds), Reproductive Endocrinology, Surgery, and Technology, vol. 2. Philadelphia: Lippincott-Raven, pp. 2319–2334.

ESHRE Capri Workshop Group (2000) Multiple gestation pregnancy. Hum Reprod 15, 1856–1864.[Abstract/Free Full Text]

Farhi J, Homburg R, Ferber A, Orvieto R and Ben Rafael Z (1997) Non-response to ovarian stimulation in normogonadotrophic, normogonadal women: a clinical sign of impending onset of ovarian failure pre-empting the rise in basal follicle stimulating hormone levels. Hum Reprod 12, 241–243.[Abstract]

Fasouliotis SJ, Simon A and Laufer N (2000) Evaluation and treatment of low responders in assisted reproductive technology: a challenge to meet. J Assist Reprod Genet 17, 357–373.[CrossRef][ISI][Medline]

Garcia J, Jones GS, Acosta AA and Wright GL Jr (1983) Human menopausal gonadotropin/human chorionic gonadotropin follicular maturation for oocyte aspiration: phase II, 1981. Fertil Steril 39, 174–179.[ISI][Medline]

Grimes DA and Schulz K (2002) Cohort studies: marching towards outcomes. Lancet 359, 341–345.[CrossRef][ISI][Medline]

Hanley JA and McNeil BJ (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143, 29–36.[Abstract]

Karande V and Gleicher N (1999) A rational approach to the management of low responders in in-vitro fertilization. Hum Reprod 14, 1744–1748.[Free Full Text]

Kim MR (1998) Screening for ovarian reserve. Assist Reprod Rev 8, 17–22.

Kligman I and Rosenwaks Z (2001) Differential clinical profiles: predicting good responders, poor responders, and hyperresponders. Fertil Steril 76, 1185–1190.[CrossRef][ISI][Medline]

Klinkert ER, Broekmans FJM, Looman CWN and te Velde ER (2004) A poor response in the first in vitro fertilization cycles is not necessarily related to a poor prognosis in subsequent cycles. Fertil Steril 81, 1247–1253.[ISI][Medline]

Lass A (2001) Assessment of ovarian reserve—is there a role for ovarian biopsy? Hum Reprod 16, 1055–1057.[Abstract/Free Full Text]

Loh S, Wang JX and Matthews CD (2002) The influence of bosy mass index, basal FSH and age on the response to gonadotrophin stimulation in non-polycystic ovarian syndrome patients. Hum Reprod 17, 1207–1211.[Abstract/Free Full Text]

Lolis DE, Tsolas O and Messinis IE (1995) The follicle-stimulating hormone threshold level for follicle maturation in superovulated cycles. Fertil Steril 63, 1272–1277.[ISI][Medline]

Mahutte NG and Arici A (2002) Poor responders: does the protocol make a difference? Curr Opin Obstet Gynecol 14, 275–281.[CrossRef][ISI][Medline]

Messinis IE and Templeton AA (1990) The importance of follicle-stimulating hormone increase for folliculogenesis. Hum Reprod 5, 153–156.[Abstract]

Peñarrubia J, Fábregues F, Creus M, Manau D, Casamitjana R, Guimerá M, Carmona F, Vanrell JA and Balasch J (2003) LH serum levels during ovarian stimulation as predictors of ovarian response and assisted reproduction outcome in down-regulated women stimulated with recombinant FSH. Hum Reprod 18, 2689–2697.[Abstract/Free Full Text]

Porchet HC, Le Cotonnec JY and Loumaye E (1994) Clinical pharmacology studies of recombinant human follicle-stimulating hormone. III. Pharmacokinetic-pharmacodynamic modeling after repeated subcutaneous administration. Fertil Steril 61, 687–695.[ISI][Medline]

Roseboom TJ, Vermeiden JPW, Schoute E, Lens JW and Schats R (1995) The probability of pregnacy after embryo transfer is affected by the age of the patient, cause of infertility, number of embryos trasnsferred and the average morphology score, as revealed by multiple regression analysis. Hum Reprod 10, 3035–3041.[Abstract]

Scott RT and Hofmann GE (1995) Prognostic assessment of ovarian reserve. Fertil Steril 63, 1–11.[ISI][Medline]

Scheffer GJ, Broekmans FJ, Dorland M, Habbema JD, Looman CW and te Velde ER (1999) Antral follicle counts by transvaginal ultrasonography are related to age in women with proven natural fertility. Fertil Steril 72, 845–851.[CrossRef][ISI][Medline]

Sharara FI and Scott RT (1997) Assessment of ovarian reserve and treatment of low responders. Infertil Reprod Med Clin N Am 8, 501–522.

Tarlatzis BC, Zepiridis L, Grimbizis G and Bontis J (2003) Clinical management of low ovarian response to stimulation for IVF: a systematic review. Hum Reprod Update 9, 61–76.[Abstract/Free Full Text]

Tinkanen H, Bläuer M, Laippala P, Tuohimaa P and Kujansuu E (1999) Prognostic factors in controlled ovarian hyperstimulation. Fertil Steril 72, 932–936.[CrossRef][ISI][Medline]

Toner JP (2003) Age=egg quality, FSH level=egg quantity. Fertil Steril 79, 491.[CrossRef][ISI][Medline]

van Loenen ACD, Huirne JAF, Schats R, Hompes PGA and Lambalk CB (2002) GnRH agonists, antagonists, and assisted conception. Sem Reprod Med 20, 349–364.[CrossRef][ISI]

van Rooij IAJ, Bancsi LFJMM, Broekmans FJM, Looman CWN, Habbema JDF and te Velde ER (2003) Women older than 40 years of age and those with elevated follicle-stimulating hormone levels differ in poor response rate and embryo quality in in vitro fertilization. Fertil Steril 79, 482–488.[CrossRef][ISI][Medline]

van Santbrink EJP, Hop WC, van Dessel TJHM, de Jong FH and Fauser BCJM (1995) Decremental follicle-stimulating hormone and dominant follicle development during the normal menstrual cycle. Fertil Steril 64, 37–43.[ISI][Medline]

Veeck LL (1999) An Atlas of Human Gametes and Conceptuses: An Illustrated Reference for Assisted Reproductive Technology. CRC Press-Parthenon Publishers, New York.

Zweig MH and Campbell G (1993) Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 39, 561–577.[Abstract/Free Full Text]

Submitted on June 22, 2004; resubmitted on September 20, 2004; accepted on November 25, 2004.