Can quantitative three-dimensional power Doppler angiography be used to predict ovarian hyperstimulation syndrome?

Transcription

1 Ultrasound Obstet Gynecol 2009; 33: Published online in Wiley InterScience ( DOI: /uog.6373 Can quantitative three-dimensional power Doppler angiography be used to predict ovarian hyperstimulation syndrome? K. JAYAPRAKASAN, R. JAYAPRAKASAN, H. A. AL-HASIE, J. S. CLEWES, B. K. CAMPBELL, I. R. JOHNSON and N. J. RAINE-FENNING Nottingham University Research and Treatment Unit in Reproduction (NURTURE), Division of Human Development, School of Clinical Sciences, Queen s Medical Centre, University of Nottingham, Nottingham, UK KEYWORDS: antral follicle count; in-vitro fertilization; ovarian blood flow; ovarian hyperstimulation syndrome (OHSS); three-dimensional power Doppler ultrasound ABSTRACT Objective To test the hypothesis that ovarian vascularity is increased in women developing ovarian hyperstimulation syndrome (OHSS) and to assess its value as a predictor of OHSS during in-vitro fertilization (IVF). Methods 118 subjects undergoing their first cycle of IVF had a three-dimensional (3D) transvaginal ultrasound scan in the early follicular phase of the menstrual cycle preceding IVF treatment. 18 of them developed moderate or severe OHSS and 100 subjects had normal ovarian response. Antral follicle count, ovarian volume, and ovarian vascularity (vascularization index (VI), flow index (FI) and vascularization flow index (VFI)) were compared between OHSS and control groups. Multiple regression analysis was used to assess the predictive value of these variables against age, body mass index and basal folliclestimulating hormone level for the development of OHSS. Results The ovarian blood flow indices VI (11.1 ± 11.6 vs. 8.6 ± 7.3%; P = 0.23), FI (38.0 ± 4.8 vs.38.0 ± 5.5; P = 0.95) and VFI (4.2 ± 3.3 vs. 3.5 ± 3.1; P = 0.40) were similar in the OHSS group and the normal responders. While antral follicle count was significantly higher in women developing OHSS (33.0 ± 15.1) than in the control group (19.2 ± 9.9, P < 0.001), ovarian volume did not differ between the two groups (10.6 ± 3.8 vs. 8.9 ± 4.8 cm 3, respectively, P = 0.11). On multiple regression analysis, antral follicle count was the only significant predictor of OHSS (P < 0.01). Conclusions Women developing OHSS during IVF do not demonstrate an increased ovarian blood flow as measured by 3D ultrasound but do have a significantly higher antral follicle count, which is the only significant predictor of OHSS. Copyright 2009 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION Ovarian stimulation with gonadotropins remains a key part of assisted reproduction technology (ART), as it maximizes the chance of a successful outcome by increasing the number of oocytes for fertilization and therefore the number of embryos, allowing competitive selection of the best for transfer into the uterus 1.The response to gonadotropins is highly variable, however, and a certain proportion of women exhibit an unexpected, poor response to stimulation while others demonstrate an exaggerated response 2. The latter is tolerated by many patients but ovarian hyperstimulation syndrome (OHSS) can lead to significant morbidity and even mortality 3,4. While a majority of subjects undergoing controlled ovarian stimulation develop mild OHSS 5 with symptoms such as abdominal bloating and discomfort, the development of moderate and severe OHSS poses significant health risks to the affected women 6 8. Therefore, the identification of women developing moderate and severe OHSS during assisted reproduction treatment is important, as this allows modification of the ovarian stimulation protocol and dose of gonadotropin to reduce their risk. While there is no evidence to suggest that these changes will result in a reduced incidence of OHSS, there is a paucity of data on the prediction of OHSS as prospective Correspondence to: Dr K. Jayaprakasan, Nottingham University Research and Treatment Unit in Reproduction (NURTURE), Division of Human Development, School of Clinical Sciences, Queen s Medical Centre, University of Nottingham, Nottingham, NG7 2UH, UK ( k.jayaprakasan@nottingham.ac.uk) Accepted: 26 January 2009 Copyright 2009 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

2 584 Jayaprakasan et al. studies are difficult to design and conduct and this is compounded by the prevalence of the condition. The reported risk factors for the development of OHSS include young age, low body mass index (BMI), as a result of being tall and slim, and, most importantly, the presence of polycystic ovaries 6. The latter probably relates to the basal number of antral follicles within the ovary 9 but may also reflect the fact that these women tend to have an increased follicular sensitivity secondary to higher ovarian blood flow than their counterparts without polycystic ovaries, resulting in an increased delivery of gonadotropin to the follicles 10,11. Although there are many studies examining the correlation of ovarian blood flow with ovarian response during controlled ovarian stimulation in the literature, studies evaluating the relationship between pre-treatment ovarian blood flow and OHSS in particular have been limited. Using pulse-wave power Doppler, Agrawal et al. reported a higher peak systolic velocity within the ovarian stroma in women who developed OHSS, compared to those with a normal response, although the impedance to flow, as quantified by the pulsatility index and resistance index, was similar in both groups 16. In addition, there are other studies that have reported an increased ovarian stromal blood flow velocity in polycystic ovaries, the most important risk factor for OHSS 17,18. While pulsewave Doppler ultrasonography assesses the absolute flow velocity and resistance to the blood flow in a single vessel, three-dimensional (3D) power Doppler ultrasound offers a global evaluation of the total vascularization and blood flow within a volume of interest 19. To date there are no reported studies describing the pre-treatment evaluation of ovarian blood flow using 3D power Doppler in women who subsequently develop OHSS during in-vitro fertilization (IVF) treatment. This study used quantitative 3D power Doppler angiography to measure ovarian blood flow in women who develop OHSS and to compare the predictive value of such measures with more conventional clinical, ultrasonographic and biochemical parameters. The hypothesis was that women developing OHSS would have significantly higher pre-treatment ovarian vascularity indices as measured by 3D power Doppler and that these would be independent predictors of the development of OHSS. PATIENTS AND METHODS Experimental design In this prospective observational study we aimed to recruit subjects under the age of 40 years with an early follicular phase follicle-stimulating hormone (FSH) level below 12 IU/L, who developed clinical features suggestive of moderate or severe OHSS 8 in association with exaggerated ovarian response, as defined by the retrieval of 15 or more oocytes 20 in their first cycle of assisted reproduction treatment. The study was performed at the University of Nottingham s Assisted Conception Unit (NURTURE: Nottingham University Research and Treatment Unit in Reproduction) and included all subjects who met the above criteria between March 2005 and May consecutive subjects who had a normal ovarian response, as defined by the retrieval of four to 15 oocytes in the absence of OHSS during the same period, were enrolled as controls for comparative analysis. Only subjects who had undergone a 3D ultrasound assessment immediately prior to their IVF treatment were included. Subjects were excluded if they had a history of ovarian surgery or were found to have an ovarian cyst or follicle measuring 20 mm or more in diameter. The study was approved by the hospital trust s ethics committee and research and development group and informed, written consent was obtained prior to the enrollment of each subject. Ultrasound data acquisition All subjects had a transvaginal scan performed by one of two investigators (K. J. or J. C.) using a Voluson Expert 730 TM (GE Medical Systems, Zipf, Austria) and a four-dimensional 5 9-MHz transvaginal probe. Subjects were scanned with their legs supported by stirrups in a modified Lloyd Davies position to limit discomfort and ensure free manipulation of the transvaginal transducer. All ultrasound examinations were performed during the early follicular phase of the menstrual cycle (between days 2 and 5) immediately prior to the cycle in which down-regulation was commenced. Our technique for the acquisition of 3D volumetric and power Doppler data has been described in detail 23.Briefly this includes an initial two-dimensional (2D) ultrasound assessment of the pelvis to exclude any obvious pathology before the application of a region of interest over the ovary that defined the volume to be acquired. An automated mechanical sweep of this region through 90 was then undertaken using the slow-sweep mode and the resultant multiplanar display examined to ensure that the entire ovary had been captured. Power Doppler was then applied using predefined settings that offer the best compromise between small ovarian vessel detection and artifact 24, and these were kept constant for every patient for a meaningful comparison to be possible 25.The following settings were used: pulse repetition frequency 1.0, power 4.0, color gain 38.4, wall motion filter 75, rise 0.2, persistence 0.8 and reject 82 with the central frequency set to mid. The volume mode was entered once an adequate power Doppler signal had been obtained. The resultant truncated sector defining the area of interest was then moved and adjusted and the sweep angle set to 90 to ensure that a complete ovarian volume was obtained. Every effort was made to avoid movement artifact by asking the subjects to remain as still as possible and by limiting movements of the transducer by the ultrasonographer. If the acquired volume was complete and considered of sufficient quality, with no power Doppler artifact, the dataset was saved to the hard drive of the ultrasound machine. Two volume acquisitions for each ovary, one with gray scale and the other with power Doppler information, were obtained. The data were

3 3D power Doppler angiography in the prediction of OHSS 585 subsequently transferred to a personal computer via a Digital Video Disk (DVD) without any data compression. Ultrasound data measurement All measurements were made on a personal computer using 4D View (version 5.3; GE Medical Systems) by two investigators (R. J. and H. A.). The 3D gray-scale ovarian volume dataset was initially displayed in the multiplanar view and the antral follicles measuring 2 9 mm in diameter were counted as previously described 26.Virtual Organ Computer-aided AnaLysis (VOCAL ; GE Medical Systems) was used to measure ovarian volume through the manual delineation of the ovarian cortex in the B plane (transverse image) as the volume was rotated by 180 in 9 rotation steps (Figure 1) 27. Quantification of power Doppler information within the resultant 3D ovarian model (Figure 2) was performed using the histogram facility. Three indices of vascularity were generated: the vascularization index (VI), which represents the ratio of power Doppler information within the total dataset relative to both color and gray information; the flow index (FI), which is proportional to the power Doppler signal intensity; and the vascularization flow index (VFI), which reflects a combination of the two (Figure 3) 28. In addition, the mean signal intensity of the gray voxels is calculated automatically and reported as the mean gray (MG) value, which provides an objective representation of the mean tissue density and therefore its apparent echogenicity 29. Two measurements of each variable were made from each dataset and the mean value used for analysis. The mean intraclass correlation coefficient (ICC) and 95% confidence interval (CI) for measurement of the total number of antral follicles and ovarian volume were ( ) and ( ), respectively, between the observers indicating a high degree of reliability. 3D quantification of blood flow, through the vascular indices, and stromal echogenicity also showed good interobserver reliability, with mean ICCs (95% CI) of ( ), ( ), ( ) and ( ) for the VI, FI, VFI and MG value, respectively. Clinical parameters Clinical parameters including the subject s age, ovulation status, and presence or absence of polycystic ovaries Figure 1 Three-dimensional (3D) power Doppler assessment of ovarian vascularity. Ovarian volume is calculated using Virtual Organ Computer-aided AnaLysis by manually delineating the ovarian cortex in the B-plane while rotating the ovarian 3D data set through 180 in 9 rotation steps.

4 586 Jayaprakasan et al. Figure 2 Three-dimensional (3D) power Doppler assessment of ovarian vascularity. The power Doppler information within the defined 3D model of the ovary is quantified using the histogram facility. Three vascular indices vascularization index, flow index and vascularization flow index are demonstrated. (PCO) and polycystic ovarian syndrome (PCOS) were recorded. Body weight and height were measured and the BMI calculated from the formula BMI = weight in kilograms/(height in meters) 2. The diagnoses of PCO and PCOS were made from the clinical data and ultrasound assessment based on the consensus derived by the Rotterdam PCOS consensus workshop group 30. PCO were defined by the presence of 12 or more antral follicles measuring 2 9 mm and/or an ovarian volume of more than 10 cm 3. Subjects were diagnosed to have PCO in isolation or PCOS if they also had oligoamenorrhea or hyperandrogenism or both of the latter features but normal ovaries on ultrasound 30. The criteria for the diagnosis of moderate OHSS were evidence of nausea with or without vomiting, ultrasound evidence of ascites and ovarian size between 8 and 12 cm and for severe OHSS, clinical ascites, hydrothorax, oliguria, hematocrit > 45%, hypoproteinemia and ovarian size > 12 cm 6 8. Treatment protocol All the subjects underwent conventional ART treatment using a standard long protocol. This involved down-regulation with a gonadotropin-releasing hormone agonist (500 µg/day of Buserelin; Suprefact, Aventis Pharma, Kent, UK or 800 µg/day of Nafarelin; Synarel, Pharmacia, Milton Keynes, UK) commenced in the midluteal phase of the menstrual cycle 7 days prior to the earliest expected date of menstruation. Successful ovarian suppression was confirmed 2 weeks later through ultrasound evidence of a thin endometrium, measuring less than 5 mm at the junction of the upper third and lower two thirds in the longitudinal plane, absent ovarian activity, and a serum estradiol level below 200 pmol/l. Controlled ovarian stimulation was then commenced using either recombinant FSH (Gonal-F; Serono Pharmaceuticals Ltd, Feltham, UK) or purified urinary human menopausal gonadotropin (Menopur, Ferring Pharmaceuticals, Berks, UK). The starting dose of gonadotropin used for ovarian stimulation was determined on the basis of the subject s age (150 IU/day for women under 30 years of age, 225 IU/day for women aged between 30 and 38 years, and 300 IU/day for women aged 38 years or more) and presence or absence of PCO; subjects with PCO were started on 150 IU/day. The starting dose used for stimulation was adjusted according to the ovarian response, which was

5 3D power Doppler angiography in the prediction of OHSS 587 Figure 3 Three-dimensional (3D) power Doppler assessment of ovarian vascularity. (a) Both gray-scale and Doppler information are shown in a 3D glass body model of an ovary. The vascularization index represents the proportion of power Doppler information within this defined volume relative to both color and gray information, providing an indication of the degree of vascularity. (b) The same ovary is shown but with the gray-scale information removed. The flow index represents the mean power Doppler signal intensity within the dataset and reflects the volume flow rate. monitored daily by serial transvaginal ultrasound and serum estradiol measurements from the fifth day of stimulation. Human chorionic gonadotropin (hcg; 6500 IU of Ovitrelle; Serono Pharmaceuticals or IU of Pregnyl; Organon Laboratories Ltd, Cambs, UK) was administered when there were at least three follicles measuring 18 mm or more in diameter, and oocyte retrieval was performed 36 hours later. Treatment was stopped if subjects did not develop at least three follicles measuring 18 mm or more in diameter. Alternatively, it was converted to intrauterine insemination treatment, depending on other clinical factors including tubal patency and seminal fluid analysis. A maximum of two normally cleaved embryos were transferred into the uterus 2 3 days after oocyte retrieval and the level of serum hcg measured 16 days later to determine the outcome. If the test was positive (hcg > 50 IU/L; biochemical pregnancy), a transvaginal ultrasound scan was arranged 2 weeks later to confirm a clinical pregnancy, defined as presence of an intrauterine gestation with fetal heart activity. The presence of a viable fetus at a subsequent scan at 12 weeks gestation confirmed ongoing pregnancy. Statistical analysis The Statistical Package for the Social Sciences (SPSS version 16.0, Chicago, IL, USA) was used for statistical analysis. The primary outcome measure was the development of OHSS. The total antral follicle count for both ovaries was used for analysis and, as there were no statistical differences between the overall mean volume or vascularity of the right and left ovary, the mean values were used for each of the other ovarian variables. The ultrasound parameters and baseline characteristics of the group developing OHSS were compared to those of the controls using an unpaired t-test or Mann Whitney U-test, depending on the normality of distribution, which was assessed by applying a normal probability plot. The chi-square test was used for binomial variables, and P < 0.05 was considered statistically significant. Multiple logistic regression analysis was used to assess the effect of each variable on the prediction of OHSS. Receiver operating characteristics (ROC) curve analysis was performed to quantify the capacity of the different parameters to discriminate between subjects with OHSS and those exhibiting a normal response. The sensitivity, specificity, and positive and negative predictive value of the most significant predictor were defined, and the positive likelihood ratio for an optimal cut-off value and the area under the ROC curve (AUC) were then calculated. RESULTS The incidence of moderate and severe OHSS within the assisted conception unit during the study period was 3% (20/656). Two subjects were excluded as they had an ovarian cyst > 20 mm in diameter, leaving 18 subjects in the OHSS group. 109 controls were recruited but nine had to be excluded: five subjects with a history of ovarian surgery, three having ovarian cyst and one in whom only a single ovary could be identified. A final study group of 118 subjects 18 having moderate or severe OHSS and 100 controls was therefore obtained for analysis. Various causative factors for subfertility were identified in these subjects including tubal disease (18.6%), male factor (30.5%), unexplained (22.9%), mixed (14.4%), endometriosis (9.3%), and PCOS (4.2%). Treatment

6 588 Jayaprakasan et al. was canceled in two subjects (1.7%) owing to the development of severe OHSS and they were offered freezing of all of their embryos at the pro-nucleate stage or of a selected number of higher quality embryos 48 h following fertilization. A further six subjects experienced failed fertilization (5.1%) such that embryo transfer was performed in 110 subjects (93.2%). This resulted in 63 pregnancies, giving a pregnancy rate of 53.4% per cycle started and 57.3% per embryo transfer completed. Six of these (9.5%) were biochemical pregnancies as there was no evidence of a pregnancy on a scan 2 weeks later at 6 weeks of gestation, giving a revised clinical pregnancy rate of 48.3% per cycle and 51.8% per embryo transferred. The ongoing pregnancy rate was 44.9% per cycle and 48.2% per embryo transferred, as four pregnancies miscarried during the next 6 weeks. The baseline clinical characteristics of the OHSS group (n = 18) and the control group (n = 100) are shown in Table 1. While there was no difference in the mean age and BMI of the women between the two groups, the incidence of PCOS and PCO was significantly higher (P < 0.001) in the group that developed OHSS. Serum FSH and luteinizing hormone levels were comparable between the groups (Table 2). The total antral follicle count, an important ultrasound determinant of PCO, was significantly higher in women with OHSS, but the other ultrasound parameters, including the 3D measures of ovarian volume, blood flow, vascularity and echogenicity, were similar between the groups. Both the starting and total dose of gonadotropin were significantly lower in the OHSS group (192 ± 46 vs. 239 ± 64 IU; P < 0.01 and 2044 ± 705 vs ± 764 IU; P < 0.01, respectively), while the number of days of ovarian stimulation was similar (10.6 ± 1.4 vs. Table 1 Comparison of baseline clinical characteristics between control and ovarian hyperstimulation syndrome (OHSS) groups Parameter OHSS group (n = 18) Control group (n = 100) P Age (years) 33.0 ± ± Subjects with primary 6 (33.3) 49 (49) 0.31 subfertility Principal cause of subfertility: Tubal factor 0 (0) 22 (22) < 0.05 Male factor 7 (38.9) 29 (29) 0.40 Unexplained 2 (11.1) 25 (25) 0.24 Mixed 5 (27.8) 12 (12) 0.14 Endometriosis 2 (11.1) 9 (9) 0.67 Anovulation due to 2 (11.1) 3 (3) 0.16 PCOS Subjects with PCOS 7 (38.9) 3 (3) < Subjects with ultrasound 13 (72.2) 28 (28) < evidence of PCO Duration of subfertility 48.6 ± ± (months) Body mass index (kg/m 2 ) 23.2 ± ± Smoker 1 (5.6) 12 (12) 0.69 Data are given as n (%) or mean ± SD. PCO, polycystic ovary; PCOS, polycystic ovary syndrome. Table 2 Comparison of baseline endocrine and ultrasound characteristics of patients in control and ovarian hyperstimulation syndrome (OHSS) groups Parameter OHSS group (n = 18) Control group (n = 100) P Basal FSH level (IU/L) 5.8 ± ± Basal LH level (IU/L) 5.7 ± ± Total antral follicle count 33.0 ± ± 9.9 < Ovarian volume (cm 3 )* 10.6 ± ± Vascularization index 11.1 ± ± (%)* Flow index (0 100)* 38.0 ± ± Vascularization flow 4.2 ± ± index (0 100)* Mean gray value (0 100)* 29.7 ± ± Data given as mean ± SD. *Mean of left and right ovarian values. FSH, follicle-stimulating hormone; LH, luteinizing hormone ± 1.4; P = 0.22). Estradiol levels were higher in the OHSS group on the fifth day of stimulation (1422 ± 1291 vs. 769 ± 686 pmol/l; P < 0.01) and on the day of hcg administration (13576 ± 5056 vs ± 4155 pmol/l; P < 0.001) in keeping with the development of a significantly higher number of follicles. Women with OHSS had significantly more follicles aspirated on the day of egg retrieval (20.9 ± 7.3vs.11.3 ± 4.4; P < 0.001) and significantly more oocytes were obtained (18.4 ± 6.3 vs. 9.5 ± 3.5; P < 0.001) as a result. Despite the higher incidence of PCO and PCOS in women with OHSS, the majority of the oocytes recovered were mature in both the OHSS (85%) and control (82.8%) groups. The clinical pregnancy rate was significantly higher in the OHSS group (14/18 (77.8%) vs. 43/100 (43%); P < 0.01). However, the ongoing pregnancy rate was statistically similar in both groups (12/18 (66.7%) vs. 41 (41%); P = 0.06), although women who developed OHSS were more likely to be pregnant, as expected. Multiple logistic regression analysis showed that the total antral follicle count was the only significant predictor of OHSS (Table 3). None of the other ultrasound parameters, including the 3D vascular indices, was predictive of OHSS. On ROC curve analysis, the antral follicle count had the best discriminative potential for the prediction of OHSS, expressed by the largest AUC (0.809), which was statistically significantly different from the corresponding values for age (0.556), BMI (0.610) FSH (0.613) and other ultrasound variables, ovarian volume (0.675), ovarian vascularity indices VI (0.571), FI (0.536) and VFI (0.591) and ovarian echogenicity (0.644) (Figure 4). The optimum cut-off value for the antral follicle count for the prediction of OHSS was > 21 with a sensitivity, specificity and positive likelihood ratio of 83%, 73% and 3.1, respectively. Subgroup analysis was performed to compare the 3D ultrasound variables between the moderate OHSS group (n = 13) and severe OHSS group (n = 5). The ovarian vascularity indices (VI: 13.0 ± 13.2 vs. 6.3 ± 3.0, P = 0.09; FI: 38.4 ± 5.5 vs.37.2 ± 2.5, P = 0.46; and VFI:

7 3D power Doppler angiography in the prediction of OHSS 589 Table 3 Multiple logistic regression analysis evaluating the value of age, basal follicle-stimulating hormone (FSH), body mass index and three-dimensional ultrasound parameters on the prediction of ovarian hyperstimulation syndrome Parameter Odds ratio (95% CI) P Age ( ) 0.35 Basal FSH ( ) 0.74 Body mass index ( ) 0.38 Total antral follicle count ( ) < 0.01 Ovarian volume* ( ) 0.25 Vascularization index* ( ) 0.43 Flow index* ( ) 0.34 Vascularization flow index* ( ) 0.44 Mean gray value* ( ) 0.15 *Using mean of left and right ovarian values. Sensitivity (%) Antral follicle count Mean ovarian volume Mean MG Mean VFI Mean VI Mean FI Specificity (%) Figure 4 Receiver operating characteristics curve analysis showing the performance of three-dimensional ultrasound ovarian parameters for the prediction of moderate and severe ovarian hyperstimulation syndrome. Antral follicle count is a significantly better predictor than ovarian volume, ovarian blood flow indices and ovarian echogenicity. FI, flow index; MG, mean gray value; VFI; vascularization flow index; VI, vascularization index. 4.9 ± 3.6 vs.2.5 ± 1.2, P = 0.09), ovarian echogenicity (30.8 ± 5.5 vs ± 5.8, P = 0.30), antral follicle count (32.4 ± 17.2 vs.34.6 ± 8.5, P = 0.52) and ovarian volume (9.8 ± 3.7 vs.12.6 ± 3.7, P = 0.13) were similar between the two groups. DISCUSSION This is the first study to examine the role of quantitative 3D power Doppler angiography preceding ovarian stimulation in the prediction of OHSS. The data suggest that the total antral follicle count is the only ultrasound parameter that significantly predicts the development of OHSS and ovarian hyper-response in women undergoing conventional IVF treatment using the long protocol. Ovarian blood flow, as measured by quantitative 3D power Doppler angiography, was not increased in women who subsequently developed moderate or severe OHSS. Pre-treatment assessment of ovarian vascularity using 3D ultrasonography does not appear to have a role in the prediction of OHSS. Our findings are in contrast to those of most studies, which used pulse-wave Doppler to assess ovarian stromal blood flow 16,18,31. The evidence from these studies is that ovarian stromal peak systolic velocity is increased in the early follicular phase in women who are at risk of OHSS and that these velocities correlate with the levels of vascular endothelial growth factor, an important mediator of OHSS 16. This indicates that 2D pulse-wave Doppler measures a different index of ovarian blood flow compared to 3D power Doppler, which has limitations in detecting the absolute velocity of flow within the vessels. The VI quantifies the number or size of vessels by calculating the total amount of power Doppler information within the defined volume 32. The presence of more vessels or their dilatation would, therefore, be expected to increase the VI. If the increased blood flow is solely related to higher flow velocities through the tissue without any reduction in the resistance to flow this would not be expected to have an effect on the VI. The FI estimates the mean intensity of the power Doppler signal and therefore provides a different index of blood flow. However, power Doppler does not provide any information on the velocity of blood flow, this being sacrificed in favor of an improved sensitivity to low flow. The exact relationship of this index to true blood flow characteristics needs to be investigated further 24,32.Itmay provide information on volume flow as it is a 3D measure but though the data acquisition process involves time, it is not a measure of perfusion, as the volume is reconstructed from 2D image planes. Thus, an isolated increase in the velocity of blood flow within individual ovarian stromal vessels would not be expected to affect the 3D indices of blood flow and vascularity. Our results suggest, therefore, that there are no more vessels, or no larger vessels, in the ovaries of women who subsequently develop OHSS. Pan et al. 20 demonstrated a significantly increased ovarian blood flow measured using 3D power Doppler angiography in women who demonstrated ovarian hyperresponse, defined as the retrieval of 15 or more oocytes or a peak estradiol level of over 3000 pg/ml, on the day of hcg administration, 36 hours before oocyte retrieval during IVF. However, the clinical application of the data from this study is limited as it does not provide any predictive information regarding the development of OHSS prior to the start of IVF treatment, which helps the clinician to tailor the stimulation protocol to optimize the ovarian response. It is likely that ovarian vascularity is similar in both hyper- and normal responders at the outset and that it increases during the course of ovarian stimulation significantly more in hyper-responders owing to exaggerated vasodilatation or neovascularization secondary to the significant rise in the serum level of estradiol 20 and of angiogenic substances

8 590 Jayaprakasan et al. such as VEGF 16. However, the finding of a similar pretreatment ovarian blood flow among the OHSS group and normal responders in our study may have also been due to the potential limitation of 3D power Doppler to detect a subtle difference in blood flow between the two groups if, indeed, it is present. Our findings are consistent with previous reports of an increased incidence of OHSS in women who have PCO 33 and PCOS 34. One of the possible explanations for the increased incidence of exaggerated ovarian response and OHSS among the PCO women is an increased sensitivity of the receptors to FSH in the ovarian follicles secondary to increased intra-ovarian estrogen resulting from the conversion of large amount of ovarian androstenedione and testosterone 33. In contrast to the normal responders (28%; 28 out of 100), the majority of women who developed moderate or severe OHSS in this study (72.2%; 13 out of 18, P < 0.001) had a PCO appearance of the ovaries on 3D ultrasound. The finding of PCO as a risk factor for OHSS is not surprising, but a significant proportion of subjects who had PCO (28%) did not develop OHSS in this study. Therefore, increased ovarian sensitivity is an important factor in addition to the size of the recruitable antral follicle population. On subgroup analysis, the ovarian blood flow did not differ even between the PCO women who developed OHSS and those who did not (VI: 11.2 ± 3.8 vs.10.1 ± 2.1, P = 0.76; FI: 37.9 ± 1.4 vs.38.7 ± 1.0, P = 0.66; VFI: 4.1 ± 1.0 vs.3.9 ± 0.6, P = 0.86). This suggests that 3D ovarian vascularity does not appear to be reflective of ovarian sensitivity. However, this may relate to follicular microcirculation as in-vitro studies suggest that follicular microvascularity relates to the gonadotropin receptor concentration, gonadotropin concentration and access to nutrients 35,36. It is likely that follicular microcirculation is below the resolution limit of conventional ultrasound and therefore, ovarian blood flow as measured using 3D power Doppler may not be a surrogate marker of the capillary blood flow around recruitable antral follicles. Future work should be focused on developing a noninvasive method capable of detecting and quantifying the perifollicular microvasculature. Our study agrees with two previous reports that suggest that antral follicle count is a significant predictor of ovarian hyper-response and OHSS 37,38. However, both these studies did not compare the predictive value of antral follicle count with that of ovarian vascular indices. Further, the cut-off values of the antral follicle count described in these studies 9 and 14, respectively seems low in contrast to that in our study which, at 22, seems more in keeping with the normal ranges defined by the Rotterdam consensus 30 on the ultrasonographic diagnostic criteria for the PCO, which by definition represents the upper extreme of normality. Ovarian volume, a determinant of PCO, on its own is not a significant predictor of OHSS as demonstrated on multiple logistic regression analysis in our study. Moreover, both the OHSS group and the control group demonstrated a similar ovarian volume. This is contrary to the report by Danninger et al., who used 3D transvaginal ultrasonography to measure ovarian volume on day 1 of stimulation, and investigated the correlation between this and the development of OHSS 39. This group found a significantly higher mean ovarian volume in patients who developed OHSS than in those who demonstrated a normal response (13.2 vs. 8.9 cm 3, respectively). Age and weight are also reported to be important predictors of OHSS, but both variables were similar in the OHSS group and normal responders in the present study. Moreover, age and BMI were not significant predictors of OHSS on multiple regression analysis. This may be because our subjects in both the study and control groups were relatively young and had low BMI. While age is possibly an important factor, as the antral follicular population is larger in younger women, the relationship between weight and OHSS is unclear, with some authors suggesting that lean women are more prone to developing OHSS 40 and others failing to find any correlation 34. This study is limited by the use of different starting doses of gonadotropins for ovarian stimulation. However, though ovarian response and the development of OHSS may have been influenced by the dose of gonadotropins, the mean starting dose, the duration of stimulation and the total dose were significantly lower in the OHSS group compared to controls as has been reported in previous studies 33,34. Moreover, the protocol we used was reflective of true clinical practice as a majority of IVF units use variable starting doses of gonadotropins and modify the daily dose based on ultrasonographic or endocrine response to stimulation 5. In conclusion, women developing moderate or severe OHSS during ART do not demonstrate increased pretreatment ovarian blood flow as measured by 3D power Doppler ultrasonography, but do have a significantly higher antral follicle count. Among potential clinical, endocrine and ultrasound factors, total antral follicle count is the only significant predictor of OHSS, and a cut-off value of > 21 provides the optimum sensitivity and specificity as a screening test. Ovarian vascularity measured using 3D ultrasound does not appear to have a role in the prediction of OHSS or its severity. REFERENCES 1. Arslan M, Bocca S, Mirkin S, Barroso G, Stadtmauer L, Oehninger S. Controlled ovarian hyperstimulation protocols for in vitro fertilization: two decades of experience after the birth of Elizabeth Carr. Fertil Steril 2005; 84: Broekmans FJ, Kwee J, Hendriks DJ, Mol BW, Lambalk CB. A systematic review of tests predicting ovarian reserve and IVF outcome. Hum Reprod Update 2006; 12: Asch RH, Li HP, Balmaceda JP, Weckstein LN, Stone SC. Severe ovarian hyperstimulation syndrome in assisted reproductive technology: definition of high risk groups. Hum Reprod 1991; 6: Rizk B, Smitz J. Ovarian hyperstimulation syndrome after superovulation using GnRH agonists for IVF and related procedures. Hum Reprod 1992; 7: Delvigne A, Rozenberg S. Epidemiology and prevention of ovarian hyperstimulation syndrome (OHSS): a review. Hum Reprod Update 2002; 8:

9 3D power Doppler angiography in the prediction of OHSS Mathur R, Kailasam C, Jenkins J. Review of the evidence base of strategies to prevent ovarian hyperstimulation syndrome. Hum Fertil (Camb) 2007; 10: Mathur R, Evbuomwan I, Jenkins J. Prevention and management of ovarian hyperstimulation syndrome. Current Obstet Gynaecol 2005; 15: Royal College of Obstetricians and Gynaecologists (RCOG). The management of ovarian hyperstimulation syndrome. RCOG green-top guideline number 5. RCOG Press: London, 2006; Navot D, Bergh PA, Laufer N. Ovarian hyperstimulation syndrome in novel reproductive technologies: prevention and treatment. Fertil Steril 1992; 58: Sladkevicius P, Valentin L, Marsal K. Blood flow velocity in the uterine and ovarian arteries during the normal menstrual cycle. Ultrasound Obstet Gynecol 1993; 3: Zaidi J, Barber J, Kyei-Mensah A, Bekir J, Campbell S, Tan SL. Relationship of ovarian stromal blood flow at the baseline ultrasound scan to subsequent follicular response in an in vitro fertilization program. Obstet Gynecol 1996; 88: Kupesic S, Kurjak A. Predictors of IVF outcome by threedimensional ultrasound. Hum Reprod 2002; 17: Mercé LT, Barco MJ, Bau S, Troyano JM. Prediction of ovarian response and IVF/ICSI outcome by three-dimensional ultrasonography and power Doppler angiography. Eur J Obstet Gynecol Reprod Biol 2007; 132: Ng EH, Tang OS, Chan CC, Ho PC. Ovarian stromal blood flow in the prediction of ovarian response during in vitro fertilization treatment. Hum Reprod 2005; 20: Ng EH, Tang OS, Chan CC, Ho PC. Ovarian stromal vascularity is not predictive of ovarian response and pregnancy. Reprod Biomed Online 2006; 12: Agrawal R, Conway G, Sladkevicius P, Tan SL, Engmann L, Payne N, Bekir J, Campbell S, Jacobs H. Serum vascular endothelial growth factor and Doppler blood flow velocities in in vitro fertilization: relevance to ovarian hyperstimulation syndrome and polycystic ovaries. Fertil Steril 1998; 70: Zaidi J, Collins W, Campbell S, Pittrof R, Tan SL. Blood flow changes in the intraovarian arteries during the periovulatory period: relationship to the time of day. Ultrasound Obstet Gynecol 1996; 7: Agrawal R, Sladkevicius P, Engmann L, Conway GS, Payne NN, Bekis J, Tan SL, Campbell S, Jacobs HS. Serum vascular endothelial growth factor concentrations and ovarian stromal blood flow are increased in women with polycystic ovaries. Hum Reprod 1998; 13: Raine-Fenning NJ, Campbell BK, Clewes JS, Kendall NR, Johnson IR. The interobserver reliability of three-dimensional power Doppler data acquisition within the female pelvis. Ultrasound Obstet Gynecol 2004; 23: Pan HA, Wu MH, Cheng YC, Wu LH, Chang FM. Quantification of ovarian Doppler signal in hyperresponders during in vitro fertilization treatment using three-dimensional power Doppler ultrasonography. Ultrasound Med Biol 2003; 29: Bancsi LF, Broekmans FJ, Eijkemans MJ, de Jong FH, Habbema JD, te Velde ER. Predictors of poor ovarian response in in vitro fertilization: a prospective study comparing basal markers of ovarian reserve. Fertil Steril 2002; 77: Pan HA, Wu MH, Cheng YC, Wu LH, Chang FM. Quantification of ovarian stromal Doppler signals in poor responders undergoing in vitro fertilization with three-dimensional power Doppler ultrasonography. Am J Obstet Gynecol 2004; 190: Raine-Fenning NJ, Campbell BK, Clewes JS, Kendall NR, Johnson IR. The reliability of virtual organ computer-aided analysis (VOCAL) for the semiquantification of ovarian, endometrial and subendometrial perfusion. Ultrasound Obstet Gynecol 2003; 22: Raine-Fenning NJ, Campbell BC, Johnson IR. The semiquantification of three-dimensional power Doppler angiography of the endometrium: preliminary experience. Abstracts from the 33rd British Medical Ultrasound Society Annual Scientific Meeting, Edinburgh, UK, December Eur J Ultrasound 2002; 15(Suppl. 1): S Raine-Fenning NJ, Nordin NM, Ramnarine KV, Campbell BK, Clewes JS, Perkins A, Johnson IR. Evaluation of the effect of machine settings on quantitative three-dimensional power Doppler angiography: an in-vitro flow phantom experiment. Ultrasound Obstet Gynecol 2008; 32: Jayaprakasan K, Hilwah N, Kendall NR, Hopkisson JF, Campbell BK, Johnson IR, Raine-Fenning NJ. Does 3D ultrasound offer any advantage in the pretreatment assessment of ovarian reserve and prediction of outcome after assisted reproduction treatment? Hum Reprod 2007; 22: Raine-Fenning NJ, Campbell BK, Clewes JS, Johnson IR. The interobserver reliability of ovarian volume measurement is improved with three-dimensional ultrasound, but dependent upon technique. Ultrasound Med Biol 2003; 29: Raine-Fenning NJ, Campbell BK, Kendall NR, Clewes JS, Johnson IR. Quantifying the changes in endometrial vascularity throughout the normal menstrual cycle with three-dimensional power Doppler angiography. Hum Reprod 2004; 19: Lam PM, Johnson IR, Raine-Fenning NJ. Three-dimensional ultrasound features of the polycystic ovary and the effect of different phenotypic expressions on these parameters. Hum Reprod 2007; 22: Balen AH, Laven JS, Tan SL, Dewailly D. Ultrasound assessment of the polycystic ovary: international consensus definitions. Hum Reprod Update 2003; 9: Zaidi J, Campbell S, Pittrof R, Kyei-Mensah A, Shaker A, Jacobs HS, Tan SL. Ovarian stromal blood flow in women with polycystic ovaries a possible new marker for diagnosis? Hum Reprod 1995; 10: Raine-Fenning NJ, Nordin NM, Ramnarine KV, Campbell BK, Clewes JS, Perkins A, Johnson IR. Determining the relationship between three-dimensional power Doppler data and true blood flow characteristics: an in-vitro flow phantom experiment. Ultrasound Obstet Gynecol 2008; 32: MacDougall MJ, Tan SL, Jacobs HS. In-vitro fertilization and the ovarian hyperstimulation syndrome. Hum Reprod 1992; 7: Delvigne A, Demoulin A, Smitz J, Donnez J, Koninckx P, Dhont M, Englert Y, Delbeke L, Darcis L, Gordts S, Puttemans P, Gerris J, Schoysman R, Leroy F. The ovarian hyperstimulation syndrome in in-vitro fertilization: a Belgian multicentric study. I. Clinical and biological features. Hum Reprod 1993; 8: Hay MF, Moor RM, Cran DG, Dott HM. Regeneration of atretic sheep ovarian follicles in vitro. J Reprod Fertil 1979; 55: Moor RM, Seamark RF. Cell signaling, permeability, and microvasculatory changes during antral follicle development in mammals. J Dairy Sci 1986; 69: Kwee JJ, Elting MM, Schats RR, McDonnell JJ, Lambalk NC. Ovarian volume and antral follicle count for the prediction of low and hyper responders with in vitro fertilization. Reprod Biol Endocrinol 2007; 5: Ng EH, Tang OS, Ho PC. The significance of the number of antral follicles prior to stimulation in predicting ovarian responses in an IVF programme. Hum Reprod 2000; 15: Danninger B, Brunner M, Obruca A, Feichtinger W. Prediction of ovarian hyperstimulation syndrome by ultrasound volumetric assessment [corrected] of baseline ovarian volume prior to stimulation. Hum Reprod 1996; 11: Navot D, Relou A, Birkenfeld A, Rabinowitz R, Brzezinski A, Margalioth EJ. Risk factors and prognostic variables in the ovarian hyperstimulation syndrome. Am J Obstet Gynecol 1988; 159: