Abstract. Introduction. RBMOnline - Vol 17. No Reproductive BioMedicine Online; on web 17 July 2008

Transcription

1 RBMOnline - Vol 17. No Reproductive BioMedicine Online; on web 17 July 2008 Article Optimal follicle and oocyte numbers for cryopreservation of all embryos in IVF cycles at risk of OHSS Gideon Verwoerd graduated with honours in medicine at the University of Stellenbosch, South Africa in He completed specialization in obstetrics and gynaecology at the Tygerberg campus of the University of Stellenbosch in He was admitted as a member of the Royal College of Obstetricians and Gynaecologists in London in He joined Bourn Hall Clinic the same year, where he is currently working as an IVF specialist. He has a special interest in ovarian hyperstimulation syndrome and the use of donor gametes. Dr Gideon Verwoerd Gideon R Verwoerd 1,2, Thomas Mathews 1, Peter R Brinsden 1 1 Assisted Conception Unit, Bourn Hall Clinic, Bourn, Cambridge, UK 2 Correspondence: Tel: ; Fax: ; gideon.verwoerd@lcg-bourn.co.uk Abstract Ovarian hyperstimulation syndrome (OHSS) is a rare but potentially fatal complication of IVF treatment. The risk of OHSS increases with increasing numbers of follicles aspirated and oocytes retrieved, but there is little evidence to support whether threshold values of either can be used to correctly predict OHSS. Since the most severe forms of OHSS are usually associated with pregnancy, cryopreservation of all embryos may prevent this. The authors attempted to find thresholds of follicle and oocyte numbers that would optimally predict OHSS, through a retrospective analysis of 2253 consecutive cycles of IVF/ intracytoplasmic sperm injection treatment reaching oocyte retrieval, between 1 January 2003 and 31 March Receiver operator characteristic (ROC) curves were calculated for both parameters, to determine threshold values that might predict OHSS in women with 20 oocytes. For the prediction of early onset OHSS, ROC curves showed that an optimal balance between sensitivity and specificity was achieved using thresholds of 24 oocytes (79%, 60%) and 29 follicles (82%, 65%) respectively. Using these thresholds, cryopreservation of all embryos may be offered as an alternative to cancellation of a treatment cycle due to excessive ovarian response, thus minimizing the number of unnecessary interventions while still correctly predicting most cases of early onset OHSS. Keywords: cryopreservation all embryos, follicles, IVF, OHSS, oocytes, threshold values Introduction 312 Ovarian hyperstimulation syndrome (OHSS) is a rare but potentially life threatening complication of ovarian stimulation treatment, and is notoriously difficult to predict accurately. It has been established that the risk of OHSS, and particularly the early onset type, increases proportionally with an increase in ovarian response to stimulation, as measured by the number of ovarian follicles, the number of oocytes retrieved and serum oestradiol concentration on the day of human chorionic gonadotrophin (HCG) administration (Asch et al., 1991; Danninger et al., 1996; Mathur et al., 2000). However, late onset OHSS has a weaker association with ovarian response compared with early onset OHSS, and is, therefore, more difficult to predict using ovarian response parameters (Mathur et al., 2000; Papanikolaou et al., 2006). Although others (Papanikolaou et al., 2006) have shown the optimal threshold value for prediction of OHSS to be 18 follicles, such a low value is not practical when a major intervention such as the elective cryopreservation of all embryos (freeze all embryos, FAE) is considered, as it would lead to an unacceptably high elective cryopreservation rate. Other interventions around the time of oocyte retrieval have been evaluated to reduce the risk of OHSS in high-risk women. Intravenous albumin infusions, coasting, and the elective cryopreservation of all embryos have not been shown to reduce significantly the risk of severe OHSS (Aboulghar et al., 2002, 2003; D Angelo and Amso, 2002, 2007; Delvigne and Rozenberg, 2002; Chen et al., 2003). In two recent proof-of-concept studies (Griesinger et al., 2007a), the use of gonadotrophin-releasing 2008 Published by Reproductive Healthcare Ltd, Duck End Farm, Dry Drayton, Cambridge CB3 8DB, UK

2 hormone (GnRH) agonists instead of HCG for oocyte maturation in antagonist cycles, and the use of very low dose HCG (2500 IU, Nargund et al., 2007), showed promise to reduce the risk of OHSS in high-risk women. However, a reduction in the risk of OHSS through the use of GnRH agonist triggering in antagonist cycles has also been associated with a reduction in pregnancy rates (Al-Inany et al., 2007). Nevertheless, patients are likely to find a lower chance of pregnancy preferable to cancellation of a treatment cycle to prevent OHSS. Although there is no direct evidence that elective cryopreservation of all embryos can reduce the severity of OHSS, severe cases of OHSS are far more likely to occur in conception cycles, and therefore it seems logical that avoidance of pregnancy may prevent this (Delvigne et al., 1993; Enskog et al., 1999). For this reason, the authors unit uses FAE as the primary preventative intervention for women who are considered at high risk of developing OHSS. Although it is true that the risk of developing OHSS cannot be defined by a single threshold value, this may at least form a basis for guidance in units where FAE is utilized as a preventative measure. Clinicians would nonetheless need to consider the patient s overall risk profile before making the decision to allow fresh embryo transfer. To the authors knowledge, this is the first study that attempts to provide such a guideline. The study population was a group of high-risk women ( 20 oocytes retrieved) in whom elective FAE would be feasible and advisable. Since late onset OHSS is impossible to predict using ovarian response parameters, it may be prudent to use either agonist triggering, or very low dose HCG during subsequent cycles in women with previous OHSS, in conjuction with elective cryopreservation of all their embryos. Materials and methods Patient population Data were retrospectively analysed from 2253 consecutive cycles of IVF or intracytoplasmic sperm injection (ICSI) treatment resulting in oocyte retrieval, carried out between 1 January 2003 and 31 March 2006 at Bourn Hall Clinic. Of these, 289 cycles resulted in the retrieval of 20 or more oocytes, and these were evaluated for inclusion in the study. After exclusion of treatment cycles in patients from satellite units (n = 50), cycles where coasting had occurred (n = 30), subsequent treatment cycles where more than one cycle had occurred in the same patient (n = 16), and cases of late onset OHSS (n = 10) a total of 183 patients were included in calculations to determine the optimal cut-off values of follicle and oocyte numbers, to predict early onset OHSS. Cases where FAE had occurred, as well as oocyte donation cycles were included in this analysis, because the development of early onset OHSS is independent of pregnancy. Since late onset OHSS only occurred in conception cycles and is unlikely to occur otherwise (Richter et al., 2004; Papanikolaou et al., 2006), only cycles resulting in pregnancy were included in calculations to predict late onset OHSS (72 non-ohss and 10 late onset OHSS cycles). Classification of OHSS OHSS was defined by using the criteria described by Rizk and Aboulghar (1999). This classification excluded cases with a mild degree of OHSS, as described in previous classifications by others (Schenker and Weinstein, 1978; Golan et al., 1989). Cases of moderate OHSS had nausea, vomiting, severe abdominal pain and distension, and ultrasound evidence of ascites. Cases of severe OHSS had additional features, including dyspnoea, haemoconcentration (haematocrit > 45%), clinical evidence of ascites or pleural effusion. This classification corresponds with that of moderate and severe OHSS suggested by the Royal College of Obstetricians and Gynaecologists (Jenkins and Mathur, 2006). Cases with critical OHSS, as described by them, did not occur in this series. In accordance with previous work (Mathur et al., 2000; Papanikolaou et al., 2006) late onset OHSS was defined as onset of clinical symptoms more than 9 days after oocyte retrieval. Stimulation protocol Throughout the study period, a low-start dose, step-up stimulation protocol with recombinant FSH (rfsh, Gonal-F, Serono, Middlesex, UK) was used. Down-regulation was routinely achieved with a GnRH agonist (buserelin, Suprecur, Aventis, Surrey, UK or nafarelin, Synarel, Pharmacia, Surrey, UK). Unless previous response suggested otherwise, the starting dose of rfsh was 150 IU for all women younger than 35 years or with risk factors for OHSS, 225 IU for women aged 35 to 39 years, and 300 IU for women 40 years or older. Oocyte maturation was routinely achieved with HCG (10,000 IU, Profasi, Serono, Middlesex, UK) during the first part of 2003, and subsequently with recombinant HCG (250 µg, Ovitrelle, Serono, Middlesex, UK). The proportion of patients who received 250 µg recombinant HCG was similar in cases with or without OHSS, and including the type of HCG in logistic regression analyses did not detect any association between the incidence of early onset OHSS and the two types of HCG used during the study period. HCG was administered once the lead follicles reached at least 18 mm in size. Follicles of 10 mm or larger were aspirated 36 h following HCG administration. Vaginal progesterone (Cyclogest, Alpharma, Devon, UK or Crinone, Serono, Middlesex, UK) was used for luteal support in all cases, which continued until HCG measurement 15 days later, or until day 77 in conception cycles. Couples were strongly advised to opt for cryopreservation of all their embryos if one of the following criteria were met: significant symptoms of OHSS at any time before embryo transfer, 30 or more oocytes retrieved, or serum oestradiol concentration on the day of HCG administration higher than 4000 pg/ml. Although patients were allowed a degree of autonomy, clinicians would not continue with fresh embryo transfer if the magnitude of the risk was considered unacceptably high. The decision to cryopreserve all embryos or not was made on the day of oocyte retrieval. Thus, most embryos were cryopreserved at the pronucleate stage. If all embryos had not been cryopreserved, the decision was reviewed if significant symptoms of OHSS appeared by the second or third day following oocyte retrieval for early stage embryo transfers, or by the fifth or sixth day for blastocyst embryo transfers. 313

3 Patients were provided with written information regarding the symptoms of OHSS, and asked to contact the unit if they experienced these at any time after oocyte retrieval. Symptomatic patients were assessed for severity and confirmation of the diagnosis by clinical, biochemical and ultrasound examination. Outcome measures The main outcomes were the threshold values of the numbers of follicles aspirated and oocytes retrieved, that provided the optimal balance between sensitivity and specificity to predict early and late onset OHSS respectively. Statistical analysis Continuous variables were investigated using either the unpaired -test or the Mann Whitney U-test, depending on the distribution. Normal probability plots and the Shapiro Wilk test were used to assess normal distribution of data. Categorical variables were compared using the two-tailed Fisher s exact test. Differences at a 95% confidence level were considered significant. Serum oestradiol concentration has been shown to be inferior to follicle number for the prediction of OHSS (Papanikolaou et al., 2006). Therefore, only the number of follicles aspirated and the number of oocytes retrieved were used to determine their discriminatory value to predict OHSS. Receiver operator characteristic (ROC) curves were used to determine the value of these parameters to predict OHSS. The positive and negative predictive values of each optimal threshold value were calculated, to evaluate their usefulness to predict the likelihood of OHSS developing in individual cases. Figure 1 shows this relationship on a continuous scale. Results Patient demographics The incidence of OHSS in the group selected for final analysis was 19.7% (38/193). Of these, 73.7% (28/38) were early onset and 26.3% (10/38) were late onset. The proportion of patients that had all their embryos cryopreserved was 16.6% (32/193), with 18/38 in the OHSS and 14/155 in the non-ohss group, respectively. The number of patients who received an embryo transfer, was 17/38 in the OHSS and 132/155 in the non-ohss group, respectively. The patient characteristics and treatment outcomes of early OHSS, late OHSS and non-ohss cycles are shown in Table 1. The mean ages, body mass index and proportions of women with polycystic ovaries were similar in all three groups. Compared with the non-ohss group, there was no difference in the total dose of gonadotrophin used in the early OHSS group, but the median number of follicles aspirated (40.0 versus 26.0, P < 0.001) and oocytes retrieved (31.5 versus 23.0, P < 0.001), were significantly higher among women who developed early OHSS, indicating an exaggerated ovarian response in these women. Optimal threshold values for the number of follicles aspirated and the number of oocytes retrieved ROC curve analysis using the number of follicles to predict early-onset OHSS, showed the threshold values with an optimal balance between sensitivity and specificity to be follicles (abnormals above cut-off). From these, the cut-off value of 28 follicles led to significant loss of specificity compared with 30 follicles (63% versus 75%, P = 0.04). Therefore, to maximize sensitivity without significant loss of specificity, 29 follicles is the optimal number and FAE should be advised if 30 follicles were aspirated. Using the number of oocytes retrieved, the threshold values with an optimal balance between sensitivity and specificity to predict early onset OHSS were oocytes (abnormals above cutoff). From these, the cut-off value of 23 oocytes led to significant loss of specificity compared with 26 oocytes (54% versus 71%, P = 0.002). Therefore, to maximize sensitivity without significant loss of specificity, 24 oocytes is the optimal number and FAE should be advised if 25 oocytes are retrieved. Tables 2 and 3 provide comparisons between early- and lateonset OHSS in terms of treatment outcome and severity of disease, (Table 2) and accuracy to predict disease using numbers of follicles and oocytes (Table 3). 314 Positive predictive value (%) Follicles aspirated Oocytes retrieved Number of follicles/oocytes Figure 1. Positive predictive value using numbers of follicles or oocytes to predict early-onset ovarian hyperstimulation syndrome among women with 20 or more oocytes retrieved.

4 Table 1. Patient characteristics, treatment and outcome in early-onset ovarian hyperstimulation syndrome (OHSS) and late-onset OHSS, compared with non-ohss. Variable Non-OHSS Early OHSS P-value Late OHSS P-value (n = 155) (n = 28) (n = 10) Age (years) 32.9 ± ± 0.8 NS 30.2 ± 1.3 NS BMI (kg/m 2 ) 24.1 ± ± 0.6 NS 23.4 ± 1.0 NS PCO on TVS (%) 61 (39) 16 (57) NS 5 (50) NS Total rfsh administered (IU) Median NS Range Number of aspirated follicles Median < NS Range Number of retrieved oocytes Median < NS Range Clinical pregnancy (%) 48 (31) 6 (21) 10 (100) BMI = body mass index; NS = not statistically significant; PCO = polycystic ovaries; rfsh = recombinant FSH; TVS = transvaginal scan. Table 2. Comparison of treatment outcome and severity of disease between women with early-onset and late-onset ovarian hyperstimulation syndrome (OHSS). Variable Early OHSS Late OHSS P-value (n = 28) (n = 10) Number of aspirated follicles Median Range Number of retrieved oocytes Median Range Severity of OHSS Severe OHSS (%) 11 (39) 5 (50) NS Moderate OHSS (%) 17 (61) 5 (50) Pregnancy outcome No. pregnant (%) 6 (21) 10 (100) Multiple pregnancy (%) 4 (67) 2 (20) NS NS = not statistically significant. Table 3. The accuracy of optimal thresholds of the numbers of follicles aspirated and oocytes retrieved, to predict early and late onset ovarian hyperstimulation syndrome (OHSS), in a group of women with 20 oocytes retrieved. Comparison Sensitivity Specificity PPV NPV Follicles aspirated (cut-off 30) Early OHSS versus non-ohss (%) a Late OHSS versus non-ohss (%) b Oocytes retrieved (cut-off 25) Early OHSS versus non-ohss (%) a Late OHSS versus non-ohss (%) b PPV = positive predictive value; NPV = negative predictive value. a n = 28 early onset OHSS and 155 non-ohss. b n = 10 late onset OHSS and 72 non-ohss (conception cycles only). 315

5 316 Discussion The primary aim of the present study was to determine the optimal cut-off values of the numbers of aspirated follicles and retrieved oocytes, to aid the clinical decision on when to advise patients that fresh embryo transfer should be avoided through cryopreservation of all their embryos. These new threshold values would then replace those that were in use during the study period. Ideally, such a study should be carried out prospectively in a group of women without any preventative interventions, but this would be dangerous and unethical. Although FAE was used as an intervention in the study population, it is unlikely to have affected the incidence of early onset OHSS, as this occurs independently of pregnancy. The ultimate purpose is to identify correctly those women who are likely to develop early onset OHSS, on the assumption that the secondary exacerbation of OHSS following pregnancy, may then be prevented in these women, through avoidance of fresh embryo transfer. The lowest cut-off values for numbers of follicles and oocytes (maximizing sensitivity) without significant loss of specificity, were considered the optimal values to serve as a guideline when FAE should be advised. The effect of FAE on the incidence of late onset or secondary exacerbation of early onset OHSS falls outside the scope of this study. Although sensitivity and specificity are useful to evaluate optimal cut-off values for the population as a whole, positive predictive value (PPV) is of greater practical value to an individual patient, as this provides an indication of her likelihood to develop OHSS. An analysis of the relationship between follicle or oocyte numbers and OHSS on a continuous scale provides further information and may be more useful in individual cases than a simple threshold (Figure 1). Owing to the low incidence of OHSS, positive predictive values remained low, ranging from 15 55% when oocytes were retrieved. It should be noted that OHSS may also occur when fewer than 20 oocytes are retrieved. However, the aim of the study was not to predict all possible cases of OHSS in a general population of IVF patients, but instead to investigate the feasibility of advising FAE when the risk of OHSS is a concern. The intention was to determine the level of ovarian response that would correctly predict the maximum number of OHSS cases while keeping the number of false predictions to a minimum. Owing to the drastic nature and lack of proven benefit of this intervention, it is undesirable to advise FAE to large numbers of women if there will be no benefit to the majority. In a retrospective analysis of 637 cycles of ovarian stimulation (Asch et al., 1991), no patient developed severe OHSS when fewer than 20 oocytes were collected, and 20 oocytes was recently shown to be the optimal cut-off point to predict OHSS in a general population of IVF patients (Mittal et al., 2006). The potential reduction of pregnancy potential due to cryopreservation of embryos also needs to be considered within the context of defining threshold values. The authors previously demonstrated that this need not be a concern in a unit that has a successful programme of embryo cryopreservation (Verwoerd et al., 2007). The clinical pregnancy rate for the first frozen embryo treatment, following FAE in the context of excessive ovarian response, was similar to fresh embryo treatment (33.3%, 24/72 versus 27.9%, 944/3385). This is probably due to the young age of the women who had an excessive ovarian response (mean age, 32.2 years), as well as a relatively high proportion of embryo transfers at the blastocyst stage (37.5% of embryo transfers) during their subsequent frozen embryo treatment cycle. Although the use of GnRH agonists to trigger oocyte maturation in GnRH antagonist cycles has been shown to reduce the risk of OHSS, it is of some concern that this is also associated with a reduction in live birth rate (Al-Inany et al., 2007). However, a recent multicentre study comparing the outcome of frozen embryo treatment following FAE in high-risk women, between agonist and antagonist cycles, showed no difference in cumulative live birth rate per patient starting a frozen embryo treatment cycle (Griesinger et al., 2007b). Antagonist cycles therefore remain an attractive option to reduce the risk of OHSS in high-risk women. The optimal thresholds identified in the present study may be useful to decide which of these high-risk women are most likely to obtain additional benefit from FAE, which may prevent secondary exacerbation of OHSS following pregnancy. This is the first study to provide evidence regarding the optimal threshold values of follicles and oocytes in a population at high risk of OHSS, specifically to serve as a guideline for advising FAE as an alternative to cancellation of a treatment cycle. Previous studies have provided optimal threshold values to predict OHSS in lower risk groups, but these are not practical to use in the context of excessive ovarian response, when FAE or cancellation of a cycle is considered. In a general IVF population, an oocyte number of 10 was shown to be the optimal threshold to predict early OHSS, with a sensitivity of 81% and specificity of 61% (Mathur et al., 2000). A large cohort study (Papanikolaou et al., 2006) showed that the optimal cut-off value to predict early OHSS was 18 follicles (sensitivity 79% and specificity 84%). These relatively low threshold values of follicle and oocyte numbers may be acceptable if used to select women for less drastic interventions than FAE. However, in the present population this would have resulted in many more women than those included in the entire study group being faced unnecessarily with a difficult and distressing decision. Since late-onset OHSS occurs after milder ovarian response than early-onset OHSS, it is more difficult to predict. Using the same cut-off values for follicles and oocytes, 79 82% of early-onset OHSS cases were predictable, but at most 50 60% of late-onset cases were predicted in this high risk group. In a general population of IVF patients, less than half of late OHSS cases would be predictable using ovarian response parameters. In fact, many cases of late-onset OHSS occur after fewer than 20 oocytes have been retrieved, which may explain why a relatively small proportion of OHSS cases in this series was of the late onset type (26% as opposed to 39% and 42% in the studies by Mathur et al. (2000) and Papanikolaou et al., (2006). Although a false-positive rate of 35% may seem unacceptably high, the study population was a relatively small, selected group of women at very high risk to develop OHSS. The actual number of false-positive predictions seen against the size of a background population of low risk IVF patients, would therefore be very small. Since patient safety is of paramount importance, it is preferable to use the lowest possible threshold

6 values, thereby maximizing sensitivity, while keeping falsepositive predictions to a minimum that is practically attainable for a condition that is very difficult to predict accurately. While threshold values with an optimal balance between sensitivity and specificity provide information regarding the population as a whole, the risk of OHSS as depicted on a continuous scale in Figure 1, provides perspective regarding the level of risk for an individual patient. Clinicians can use this information to decide whether FAE should be advised to patients, either as an intervention on its own to prevent exacerbation of early-onset and late-onset OHSS, or in combination with other measures such as very low dose HCG, or the use of GnRH agonists for oocyte maturation in GnRH antagonist cycles. Acknowledgements The authors would like to thank the staff at Bourn Hall Clinic for their assistance in collection of data for the study, and Dr E Walters from IVF Statistics Advice, Cambridge, for assistance in statistical analysis of the data. References Aboulghar MA, Mansour RT 2003 Ovarian hyperstimulation syndrome: classifications and critical analysis of preventive measures. Human Reproduction Update 9, Aboulghar M, Evers JH, Al-Inany H 2002 Intra-venous albumin for preventing severe ovarian hyperstimulation syndrome. Cochrane Database Systematic Reviews 2, CD Al-Inany HG, Abou-Setta AM, Aboulghar M 2007 Gonadotrophinreleasing hormone antagonists for assisted conception: a Cochrane review. Reproductive BioMedicine Online 14, Asch RH, Li HP, Balmaceda JP et al Severe ovarian hyperstimulation syndrome in assisted reproductive technology: definition of high risk groups. Human Reproduction 6, Chen D, Burmeister L, Goldschlag D et al Ovarian hyperstimulation syndrome: strategies for prevention. Reproductive BioMedicine Online 7, D Angelo A, Amso N 2007 Embryo freezing for preventing ovarian hyperstimulation syndrome. Cochrane Database Systematic Reviews 3, CD D Angelo A, Amso N 2002 Coasting (withholding gonadotrophins) for preventing ovarian hyperstimulation syndrome. Cochrane Database Systematic Reviews 3, CD Danninger B, Brunner M, Obruca A, Feichtinger W 1996 Prediction of ovarian hyperstimulation syndrome by ultrasound volumetric assessment [corrected] of baseline ovarian volume prior to stimulation. Human Reproduction 11, Delvigne A, Rozenberg S 2002 Epidemiology and prevention of ovarian hyperstimulation syndrome (OHSS): a review. Human Reproduction Update 8, Delvigne A, Demoulin A, Smitz J et al The ovarian hyperstimulation syndrome in in-vitro fertilization: a Belgian multicentric study. I. Clinical and biological features. Human Reproduction 8, Enskog A, Henriksson M, Unander M et al Prospective study of the clinical and laboratory parameters of patients in whom ovarian hyperstimulation syndrome developed during controlled ovarian hyperstimulation for in vitro fertilization. Fertility and Sterility 71, Golan A, Ron-El R, Herman A et al Ovarian hyperstimulation syndrome: an update review. Obstetrics and Gynecology Survey 44, Griesinger G, Von Otte S, Schroer A et al. 2007a Elective cryopreservation of all pronuclear oocytes after GnRH agonist triggering of final oocyte maturation in patients at risk of developing OHSS: a prospective, observational proof-of-concept study. Human Reproduction 22, Griesinger G, Kolibianakis EM, Papanikolaou EG et al. 2007b Triggering of final oocyte maturation with gonadotropin-releasing hormone agonist or human chorionic gonadotropin. Live birth after frozen-thawed embryo replacement cycles. Fertility and Sterility 88, Jenkins JM, Mathur RS 2006 The management of ovarian hyperstimulation syndrome. Royal College of Obstetrics and Gynaecology Green-top Guideline No. 5. Mathur RS, Akande AV, Keay SD et al Distinction between early and late ovarian hyperstimulation syndrome. Fertility and Sterility 73, Mittal S, Gaur R, Majumdar G, Majumdar A 2006 Combination of early estradiol levels and oocyte number as a strong prognostic tool for ovarian hyperstimulation syndrome in IVF. In: 22nd Annual Meeting of the European Society of Human Reproduction and Embryology, Prague, Czech Republic, June Human Reproduction 21 (Suppl. 1), P-377 [poster]. Nargund G, Hutchison L, Scaramuzzi R et al Low-dose HCG is useful in preventing OHSS in high-risk women without adversely affecting the outcome of IVF cycles. Reproductive BioMedicine Online 14, Papanikolaou EG, Pozzobon C, Kolibianakis EM et al Incidence and prediction of ovarian hyperstimulation syndrome in women undergoing gonadotrophin releasing hormone antagonist in vitro fertilization cycles. Fertility and Sterility 85, Richter KS, Van Nest RL, Stillman RJ 2004 Late presentation with severe ovarian hyperstimulation syndrome is diagnostic of clinical in vitro fertilization pregnancy. Fertility and Sterility 82, Rizk B, Aboulghar MA 1999 Classification, pathophysiology and management of ovarian hyperstimulation syndrome. In: Brinsden P (ed.) In Vitro Fertilization and Assisted Reproduction. Parthenon Publishing Group, New York, Schenker JG, Weinstein D 1978 Ovarian hyperstimulation syndrome: a current survey. Fertility and Sterility 30, Verwoerd GR, Marikinti K, Mathews T, Macnamee MC 2007 Pregnancy outcome after frozen/thawed embryo treatment following cryopreservation of all embryos in cycles with excessive ovarian response. P-44 from the 63 rd Annual Meeting of the ASRM, Washington, USA, October Declaration: The authors report no financial or commercial conflicts of interest. Received 24 September 2007; refereed 26 October 2007; accepted 16 April