Accuracy of ultrasound subjective pattern recognition for the diagnosis of borderline ovarian tumors

Ultrasound Obstet Gynecol 2007; 29: 489 495 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/uog.4002 Accuracy of ultrasound subjective pattern recognition for the diagnosis of borderline ovarian tumors J. YAZBEK*, K. S. RAJU, J. BEN-NAGI*, T. HOLLAND*, K. HILLABY* and D. JURKOVIC* *Early Pregnancy and Gynaecology Assessment Unit, King s College Hospital and South East Gynaecological Cancer Centre, Guy s and St Thomas Foundation Trust, London, UK KEYWORDS: benign; borderline; malignant; ovarian tumors; ultrasound ABSTRACT Objectives To assess the value of pattern recognition for the preoperative ultrasound diagnosis of borderline ovarian tumors (BOTs). Methods This was a prospective study of women who were referred to our regional cancer center with the diagnosis of an adnexal mass on a Level II (routine) gynecological ultrasound scan. Women with lesions of uncertain nature were referred for a Level III (expert) ultrasound scan in our tertiary center. The tumor pattern recognition method was used to differentiate between various types of ovarian tumors. Morphological features suggestive of BOTs were: unilocular cyst with a positive ovarian crescent sign and extensive papillary projections arising from the inner wall, or a cyst with a well defined multilocular nodule. The ultrasound findings were compared with the final histological diagnosis. Results A total of 224 women with an adnexal mass of uncertain nature were referred for an expert scan, 166 (74.1%) of whom underwent surgery. In this group of women the final histological diagnoses were: 99 (60%) benign lesions, 32 (19%) invasive ovarian cancer and 35 (21%) BOTs. Using pattern recognition combining the different morphological features, a correct preoperative diagnosis of BOT was made in 24/35 (68.6%) women: area under the receiver operating characteristics curve 0.812 (standard error 0.049; 95% CI, 0.716 0.908), sensitivity 0.69 (95% CI, 0.52 0.81), specificity 0.94 (95% CI, 0.88 0.97), positive likelihood ratio 11.3 (95% CI, 5.53 22.8) and negative likelihood ratio 0.34 (95% CI, 0.21 0.55). Conclusions Ultrasound diagnosis of BOTs is highly specific. However, typical features are absent in onethird of cases, which are typically misdiagnosed as benign lesions. Copyright 2007 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION An accurate preoperative differential diagnosis of ovarian tumors is of major clinical significance because of the implications for further management and survival. In the past ovarian tumors used to be classified only as benign or malignant. Since 1971, the International Federation of Gynecology and Obstetrics (FIGO) has classified borderline ovarian tumors (BOTs) as an entity separate from benign and invasive ovarian tumors 1,2. However, until recently, BOTs have been treated in the same way as invasive malignant tumors. BOTs are more prevalent in women of childbearing age 3 5 and their prognosis is generally good 6,7. They often follow a relatively benign course 6,7 and therefore fertilitysparing conservative surgery is often contemplated in women with a BOT who wish to preserve their reproductive potential 8. BOTs are divided into two major histological groups: serous (two subtypes: typical serous and micropapillary) and mucinous (two subtypes: gastrointestinal and endocervical) 6. In the past decade, ultrasound examination has been accepted as the optimal diagnostic modality for the noninvasive assessment of ovarian tumors. However, the differentiation between various types of ovarian tumors on ultrasonography is sometimes difficult. Although, several ultrasound-based diagnostic algorithms have been proposed to facilitate differentiation between benign and malignant tumors 9 15, the pattern recognition method remains the best way of assessing the nature of ovarian tumors 16. However, the accuracy of pattern recognition is dependent on the operator s skill and experience 17. Several Correspondence to: Dr J. Yazbek, Suite 8, 3 rd Floor, Golden Jubilee Wing, Department of Obstetrics and Gynaecology, King s College Hospital, Denmark Hill, London SE5 8RX, UK (e-mail: joseph.yazbek@gmail.com) Accepted: 25 January 2007 Copyright 2007 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPER

490 Yazbek et al. recent studies examined the gray-scale sonographic and color Doppler features of BOTs 18 20. There were also attempts to describe the typical sonographic features of different histopathological subtypes of BOTs 18,21,22. However, the diagnostic accuracy of pattern recognition for the differential diagnosis between benign, borderline and invasive ovarian tumors has not been tested as yet. The aim of this study was to assess prospectively the value of pattern recognition for the differential diagnosis of adnexal tumors and in particular its accuracy in establishing the specific diagnosis of histological subtypes of BOTs. The presence of healthy ovarian tissue adjacent to the tumor, the ovarian crescent sign 24, was used to exclude invasive ovarian cancer 25.Theovariancrescent sign was defined as visible hypoechogenic tissue with or without ovarian follicles enclosed within the ovarian capsule encircling the tumor and located adjacent to the cyst wall. This tissue would not be separated from the cyst when applying a moderate amount of pressure. Women who were diagnosed with benign ovarian tumors underwent conservative surgery. Women of METHODS This was a prospective observational study of women who were referred to our regional gynecological cancer center with the diagnosis of an adnexal mass. They were all assessed clinically and had a Level II 23 gynecological ultrasound scan. Women with conclusive features of ovarian cancer or benign tumors were managed according to the center s protocols. Those women in whom the nature of the lesion was uncertain during the Level II gynecological ultrasound scan were referred for a Level III 23 ultrasound scan at our center, which is a tertiary referral gynecological ultrasound unit. All women referred to the unit underwent a pelvic ultrasound scan, which involved a detailed examination of the uterus and the adnexa. All examinations were performed by gynecologists with a special interest in gynecological ultrasound using an Aloka SSD-5000 machine (Aloka Co., Tokyo, Japan). Transvaginal and transabdominal scans were subsequently performed on each patient to ensure a complete examination of the entire abdominal cavity. Age and menopausal status were recorded in all cases. Menopausal status was defined as more than 1 year of amenorrhea or age > 50 years in the case of previous hysterectomy. The following morphological ultrasound information was recorded in each case: site and volume of cyst, presence of septations and solid areas within the cyst, metastases and ascites. The tumor pattern recognition method was used prospectively to differentiate between various types of ovarian tumors. Morphological features suggestive of a BOT were: unilocular cyst with a positive ovarian crescent sign and extensive papillarities arising from the inner wall, or a cyst with a well defined multilocular nodule. Solid papillary projections were defined as any solid projections into the cyst cavity arising from the cyst wall with a height greater than or equal to 3 mm. A multilocular nodule ( honeycomb nodule ) was defined as a predominantly solid nodule with cystic areas arising from the inner cyst wall. The presence of multiple papillary projections and ovarian crescent sign were suggestive of a seroustype (Figure 1) or an endocervical-type (Figure 2) BOT, whereas the presence of a honeycomb nodule and thick echogenic fluid content was indicative of a gastrointestinal (GI)-type BOT (Figure 3). Figure 1 Ultrasound image showing a serous borderline ovarian tumor with extensive papillary projections arising from the cyst wall. Note the presence of healthy ovarian tissue, the ovarian crescent sign (arrow). Figure 2 Ultrasound image showing a mucinous endocervical-type borderline ovarian tumor (BOT). Note the resemblance to a serous BOT, but more organized papillary projections. The ovarian crescent sign is also a common feature (arrow).

Diagnosis of borderline ovarian tumors 491 RESULTS Figure 3 Ultrasound image showing a mucinous gastrointestinal-type borderline ovarian tumor. A honeycomb nodule is seen suspended within the cyst cavity (arrow). childbearing age with a suspected BOT who wished to preserve their fertility underwent fertility-sparing surgery such as cystectomy, oophorectomy or adnexectomy. Other women with BOTs who had completed their families and those with suspected invasive ovarian cancer underwent a laparotomy, total abdominal hysterectomy, bilateral salpingo-oophorectomy and omentectomy at this center. The ultrasound findings were compared with the histology, which was classified according to the World Health Organization guidelines 2. Ovarian malignancies were staged according to the classification of FIGO 1,26. The study was approved by the King s Research Ethics Committee and all patients gave informed consent. All statistical analyses were carried out using SPSS version 14 (SPSS Inc., Chicago, IL, USA). The statistical significance of differences in continuous variables was determined using Mann Whitney U-test, Kruskal Wallis test or Student t-test depending on data distribution. Proportions were compared using Yates -corrected Chisquare test. Two-tailed P < 0.05 was considered statistically significant. The diagnostic accuracy of the tests was assessed using sensitivity, specificity and likelihood ratio measures. A total of 224 women with an adnexal mass of uncertain nature were referred for an expert scan, 166 (74.1%) of whom underwent surgery. The remaining 58 (25.9%) had functional ovarian cysts, hydrosalpinges, peritoneal inclusion cysts or uterine fibroids diagnosed on ultrasound scan, and these did not require surgical intervention. Women with a suspected functional ovarian cyst had at least one follow-up ultrasound scan to confirm spontaneous resolution of the cyst. Histopathological results were available for all 166 women who underwent surgery and they were all included in the final data analyses. There were 99 (59.6%; 95% CI, 52 67) benign tumors, 35 (21.1%; 95% CI, 15 27) BOTs and 32 (19.3%; 95% CI, 13 25) invasive ovarian cancers. The patients mean age was 42 (range, 14 88) years and 45/166 (27%; 95% CI, 20 34) women were postmenopausal. Patients demographic data and clinical symptoms at presentation are listed in Table 1. There were significant differences in the women s age and clinical symptoms between those with benign, borderline and malignant ovarian tumors. FIGO staging and histological types of BOT and invasive cancers are listed in Table 2; 25/35 (71.4%; 95% CI, 56 86) BOTs underwent surgical staging. All the staged GI-type and the majority of the staged serous endocervical-type BOTs were FIGO Stage I. Only 2/16 (12.5%; 95% CI 0 29) serous endocervical-type BOTs were FIGO Stage III. Morphological ultrasonic characteristics and final histology of all ovarian tumors in the study population are summarized in Table 3. Papillary projections were present in approximately half of all borderline and invasive ovarian tumors, and in one-fifth of ovarian cystadenomas (χ 2 = 35.0, P < 0.001). Solid papillary projections were much more frequent in serous endocervical-type BOTs (80%; 95% CI, 63 98) than GI-type BOTs (7%; 95% CI, 0 20) (χ 2 = 18.5, P < 0.001). There were also significant differences in the presence of the ovarian crescent sign between benign, borderline and invasive ovarian tumors (χ 2 = 75.6, P < 0.001). In the subtypes of borderline tumors, the ovarian crescent sign was present in 75% (95% CI; 56 94) of serous endocervical-type BOTs and in 20% (95% CI; 0 40) of GI-type BOTs Table 1 Demographic data and clinical symptoms at presentation of 166 women with an adnexal mass Type of adnexal mass Benign (n = 99) Borderline (n = 35) Invasive (n = 32) P Age (years, mean (range)) 39 (14 83) 39 (21 65) 52 (17 88) 0.002* Postmenopausal ( [95% CI]) 19 (19) [11 27] 7 (20) [7 33] 19 (59) [42 76] <0.001 Pain ( [95% CI]) 41 (41) [31 51] 9 (26) [12 41] 5 (16) [3 29] <0.05 Abdominal distension ( [95% CI]) 8 (8) [3 13] 5 (14) [3 26] 18 (56) [39 73] <0.001 Incidental finding on ultrasonography or clinical examination ( [95% CI]) 50 (51) [41 61] 21 (60) [44 76] 9 (28) [12 44] <0.05 *Kruskal Wallis test; chi-square test.

492 Yazbek et al. Table 2 Histology and stages of borderline and invasive ovarian cancers FIGO Stage I FIGO Stage II FIGO Stage III FIGO Stage IV Serous endocervical-type BOT* (n = 16) 14 (87.5) 0 (0) 2 (12.5) 0 (0) GI-type BOT* (n = 9) 9 (100) 0 (0) 0 (0) 0 (0) Epithelial invasive (n = 24) 9 (37.5) 6 (25) 6 (25) 3 (12.5) Non-epithelial invasive (n = 8) 3 (37.5) 1 (12.5) 3 (37.5) 1 (12.5) *25/35 (71%) borderline tumors had formal surgical staging. BOT, borderline ovarian tumor; GI, gastrointestinal. Table 3 Gray-scale ultrasound characteristics and histology of the tumors of the study population Type of tumor Benign (n = 99) Borderline (n = 35) Invasive (n = 32) Dermoid Cystadenoma Endometrioma Thecoma or fibroma Simple or inflammatory cyst Serous endocervicaltype BOT GI-type BOT Epithelial cancer Nonepithelial cancer Patients (n) 35 36 17 7 4 20 15 24 8 Unilocular () 11 (31.4) 14 (38.9) 12 (70.1) 0 (0) 2 (50) 3 (15) 1 (6) 1 (4.2) 0 (0) Unilocular solid () 15 (42.9) 0 (0) 1 (5.9) 0 (0) 0 (0) 16 (80) 0 (0) 2 (8.3) 0 (0) Multilocular () 3 (8.6) 20 (55.6) 4 (23.5) 0 (0) 2 (50) 1 (5) 4 (27) 6 (25) 1 (12.5) Multilocular solid () 5 (14.3) 2 (5.6) 0 (0) 0 (0) 0 (0) 0 (0) 10 (67) 7 (29.2) 2 (25) Solid () 1 (2.9) 0 (0) 0 (0) 7 (100) 0 (0) 0 (0) 0 (0) 8 (33.3) 5 (62.5) Papillary projections () 0 (0) 7 (19.4) 0 (0) 0 (0) 0 (0) 16 (80) 1 (6.7) 13 (54.2) 1 (12.5) Thick fluid content () 0 (0) 3 (8.3) 16 (94.1) 0 (0) 1 (25) 4 (20) 15 (100) 5 (20.1) 0 (0) Honeycomb nodule () 0 (0) 1 (2.8) 0 (0) 0 (0) 0 (0) 0 (0) 8 (53.3) 0 (0) 0 (0) Ovarian crescent () 33 (94.3) 28 (77.8) 16 (94.1) 5 (71.4) 4 (100) 15 (75) 3 (20) 1 (4.2) 0 (0) BOT, borderline ovarian tumor; GI, gastrointestinal. Table 4 Ultrasound morphological appearance of false-positive and false-negative cases n Histology Ultrasound diagnosis Ultrasound morphology 7 Cystadenoma Serous endocervical-type BOT Unilocular solid cyst with numerous PPs 1 Cystadenoma GI-type BOT Unilocular cyst with a honeycomb nodule 3 Serous endocervical-type BOT Cystadenoma Unilocular cyst 1 Serous endocervical-type BOT Cystadenoma Unilocular cyst with a PP 1 Serous endocervical-type BOT Cystadenoma Multilocular cyst 3 GI-type BOT Cystadenoma Multilocular cyst 1 GI-type BOT Endometrioma Unilocular cyst containing fluid with ground glass appearance 1 GI-type BOT Dermoid Multilocular solid cyst 1 GI-type BOT Invasive cancer Multilocular solid cyst BOT, borderline ovarian tumor; GI, gastrointestinal; PP, papillary projection. (χ 2 = 10.38, P < 0.01). Thick fluid was present in all GI-type BOTs. However, this was not a specific finding as it was also found in most endometriomas as well as in some serous endocervical-type BOTs and occasionally in epithelial ovarian cancers. The honeycomb nodule was a highly specific feature of GI-type BOTs. However, its sensitivity was low as it was absent in nearly half of cases. The overall accuracy of ultrasound diagnosis of BOTs using pattern recognition was 68.6% (95% CI, 53 84). The accuracy was better in cases of serous endocervicaltype (75%; 95% CI, 56 94) than GI-type (60%; 95% CI, 35 85) BOTs. The false-negative and false-positive cases encountered in this study are summarized with their respective gray-scale morphological appearances in Table 4. The women s age and tumor volume were not significantly different in cases of true-positive, falsepositive or false-negative diagnoses in women with different subtypes of BOTs (Table 5). The median volume of the false-negative unilocular cysts was 1226 (range, 412 5661) ml. The area under the receiver operating characteristics curve, sensitivity, specificity, and positive and negative likelihood ratios of pattern recognition combining

Diagnosis of borderline ovarian tumors 493 Table 5 Age of women and tumor volume in true-positive, false-positive and false-negative cases of borderline ovarian tumor (BOT) Finding True positive False positive False negative P GI-type BOT n 8 1 6 Age (years, median (range)) 39 (22 64) 62 41 (31 63) NS Tumor volume (ml, median (range)) 1124 (603 6406) 2469 476 (90 5661) NS Serous endocervical-type BOT n 16 7 5 Age (years, median (range)) 36 (21 47) 62 (16 80) 45 (31 65) NS Tumor volume (ml, median (range)) 77 (1 2579) 613 (25 4657) 321 (45 1593) NS GI, gastrointestinal; NS, not significant. Table 6 Accuracy of pattern recognition for the diagnosis of different types of ovarian tumor Type of tumor Area under ROC curve Standard error Sensitivity Specificity LR+ LR Benign 0.872 (0.811 0.934) 0.031 0.91 (0.83 0.95) 0.86 (0.75 0.92) 6.45 (3.71 11.19) 0.11 (0.05 0.21) Borderline 0.812 (0.716 0.908) 0.049 0.69 (0.52 0.81) 0.94 (0.88 0.97) 11.3 (5.53 22.8) 0.34 (0.21 0.55) Malignant 0.977 (0.940 1.014) 0.019 0.97 (0.84 0.99) 0.99 (0.96 0.99) 129.8 (18.4 915.7) 0.03 (0.01 0.22) Sensitivities, specificities and positive and negative likelihood ratios (LR+ and LR ) were calculated separately for each of the three groups of tumors. ROC, receiver operating characteristics. different morphological features for the diagnosis of benign, borderline or malignant tumors are shown in Table 6. DISCUSSION Our study has confirmed that BOTs tend to occur in younger women and that they are detected incidentally during clinical or ultrasound examination in the majority of cases for indications other than a suspected ovarian tumor. The differences in age and clinical symptoms between women with benign cysts, borderline and invasive cancers were statistically significant, and should be taken into consideration when assessing an adnexal tumor by ultrasound examination. These observations were in agreement with the results of previous studies, which also showed that women diagnosed with a BOT were more likely to be asymptomatic than those with benign or invasive ovarian tumors 3 5,27. We were able to establish a correct histological diagnosis in more than two-thirds of all BOTs. The specificity of ultrasound diagnosis was very high, but the sensitivity was only 69%. This was mostly due to the relatively high proportion of cysts, which did not exhibit typical ultrasonic features of BOTs. Amongst 11 falsenegative cases encountered in our study, there were four (36%) cysts, which were unilocular with smooth inner walls and no papillary projections. In these four cases, there was a complete agreement between the ultrasound morphological appearances and the histopathological macroscopic description in misdiagnosed ovarian tumors. There were no cases in which papillary projections were missed on ultrasound scan when the results were compared with histological findings. Nonetheless, our data do not support surgical intervention in all cases of unilocular cysts, although it would be fair to say that the diagnosis of BOT needs to be considered in women with large and growing unilocular cysts. The proportion of unilocular cysts in our population of women with BOTs was 11.4%, which was much higher than the findings of a retrospective study by Fruscella et al., who found unilocular cysts in only 3.5% of their population of BOTs 21. However, we were able to confirm their observation that different BOT subtypes display different morphological features, which is helpful in the differential diagnosis 21. We also agree with their findings that serous endocervical-type BOTs are characterized by a higher rate of unilocular solid lesions, higher number of papillary projections and lower prevalence of multilocular lesions compared with GI-type BOTs. Both studies showed a statistically significant difference between the morphological ultrasound appearances of serous endocervical-type and GI-type BOTs. We were more accurate in diagnosing serous endocervical-type than GI-type BOTs. This is important in view of the relatively worse prognosis of serous endocervical-type BOTs and the higher recurrence rate 6. The presence of a honeycomb nodule was highly specific for GI-type BOTs. Although slightly more than half of the GI-type BOTs had a honeycomb nodule, this nodule was also present in a case of benign mucinous cystadenoma. On the other hand, all GI-type BOTs had thick fluid noted within the cyst, which makes this finding a highly

494 Yazbek et al. sensitive marker for this subtype of mucinous BOT. Thick fluid content was also a characteristic of the majority of ovarian endometriomas, a few serous endocervical-type BOTs and some epithelial malignant ovarian tumors. Our description of GI-type BOTs is not substantially different from the one provided by Fruscella et al. 21, who described the typical pattern of mucinous GItype BOT as being a multilocular cyst with a high number of locules. Using the International Ovarian Tumor Analysis (IOTA) 28 classification these tumors should indeed be described as multilocular cysts. The honeycomb nodule, however, is a specific feature of GItype BOTs, which represent a subgroup of multilocular cysts. This distinction is important, as many benign and invasive tumors are classified as multilocular cysts. Nearly two-thirds of benign cystadenomas in this study were described as multilocular cysts and, without performing a more detailed analysis, they could have all been wrongly classified as GI-type BOT. The ovarian crescent sign was present in 75% of serous endocervical-type BOTs, but only in 20% of GItype BOTs. This was mainly due to the large size of GI-type BOTs, which makes the visualization of healthy ovarian tissue more difficult. However, in the presence of a positive ovarian crescent sign, the diagnosis of an invasive tumor is very unlikely. There was only one case of an invasive tumor with a positive crescent sign in this series, which was a mature cystic teratoma with a focus of squamous cell carcinoma (Stage I). In practical terms the ovarian crescent sign can be used to exclude an invasive ovarian cancer, but its absence is not diagnostic of invasive ovarian disease. In this study, the overall sensitivity of pattern recognition in differentiation between benign and malignant tumors (including both borderline and invasive lesions) was 91% (95% CI; 83 95), similar to the findings of Valentin et al. 16 who reported a sensitivity of 83% (95% CI; 67 94). However, the specificity in our study was less (85% (95% CI; 75 92) vs. 91% (95% CI; 84 96) 16 ), which may be explained by a higher proportion of BOTs in our study population. Our data show that the preoperative ultrasound diagnosis of BOT using the pattern recognition method is highly specific. Thus, women who are in the reproductive age group could benefit from fertility-sparing surgery in a regional gynecological cancer center. On the other hand, when reproductive potential is not an issue, accuracy in preoperative diagnosis would allow better preoperative counseling and a more conservative surgical approach. All the correctly diagnosed BOTs and one false-negative case (thought to be an invasive cancer) underwent surgical staging. Staging remains important in cases of serous endocervical-type BOT, because of the possibility of advanced stage disease at diagnosis 7, especially in the case of micropapillary-subtype serous BOT. The importance of staging may be less in GI-type BOT and the impact of complete staging on prognosis of BOTs is far from clear 4,29. It is also unclear whether spillage of the cyst content adversely affects the prognosis of women with BOTs. This is a particularly important issue in the context of laparoscopic surgery for ovarian tumors. Laparoscopic surgery offers many advantages to women with benign cysts because of the shorter hospital stay, reduced postoperative pain, and faster recovery and return to normal daily activities 30 32. However, the risk of cyst spillage during laparoscopy is much greater than during open surgery. Our data show that, even in optimal circumstances, a number of borderline tumors will be misclassified as benign on ultrasound scan and therefore be considered suitable for minimally invasive surgery. Thus, there is a great need to continue improving noninvasive diagnosis of borderline tumors and, at the same time, to assess any effects of BOT spillage on survival and recurrence rates 8,33. Until such data are available, every effort should be made to minimize the risk of spillage of cyst content during surgery for all types of ovarian cyst. This study only assessed the value of pattern recognition in diagnosing BOTs and differentiating them from benign and invasive ovarian tumors. Doppler studies were not used in the differentiation between different types of ovarian tumors. There is no proven benefit of Doppler studies in the differentiation of ovarian tumors 34 and, perhaps, the value of Doppler studies in aiding the diagnosis of BOT should be evaluated in a prospective randomized study. Our dataset included a significant number of functional and dermoid cysts, which are usually simple to diagnose on a routine ultrasound examination. They were often referred to the regional cancer center because of raised serum CA125 levels in association with the cyst. In such situations it is not unusual for gynecological oncologists to refer these patients for an expert scan even if non-specialist ultrasound findings were reassuring. In conclusion, our study has shown that the correct ultrasound diagnosis of BOT can be achieved in twothirds of cases. Further work is required to identify morphological and biochemical features that may help to increase the accuracy of preoperative detection of such tumors. REFERENCES 1. International Federation of Gynecology and Obstetrics. Classification and staging of malignant tumors in the female pelvis. Acta Obstet Gynecol Scand 1971; 50: 1 7. 2. Serov SF, Scully RE, Sobin LH. The World Health Organization International histological classification of ovarian tumours: histological typing of ovarian tumours. World Health Organization: Geneva, 1973. 3. Gotlieb WH, Chetrit A, Menczer J, Hirsh-Yechezkel G, Lubin F, Friedman E, Modan B, Ben-Baruch G. Demographic and genetic characteristics of patients with borderline ovarian tumors as compared to early stage invasive ovarian cancer. Gynecol Oncol 2005; 97: 780 783. 4. Gotlieb WH, Flikker S, Davidson B, Korach Y, Kopolovic J, Ben-Baruch G. Borderline tumors of the ovary. Fertility treatment, conservative management, and pregnancy outcome. Cancer 1998; 82: 141 146.

Diagnosis of borderline ovarian tumors 495 5. Boran N, Cil AP, Tulunay G, Ozturkoglu E, Koc S, Bulbul D, Kose MF. Fertility and recurrence results of conservative surgery for borderline ovarian tumors. Gynecol Oncol 2005; 97: 845 851. 6. Ahmed ASM, Lawton FG. Borderline ovarian tumours: current concepts and management. Reviews Gynaecol Pract 2005; 5: 139 151. 7. Silverberg SG, Bell DA, Kurman RJ, Seidman JD, Prat J, Ronnett BM, Copeland L, Silva E, Gorstein F, Young RH. Borderline ovarian tumors: key points and workshop summary. Hum Pathol 2004; 35: 910 917. 8. Maneo A, Vignali M, Chiari S, Colombo A, Mangioni C, Landoni F. Are borderline tumors of the ovary safely treated by laparoscopy? Gynecol Oncol 2004; 94: 387 392. 9. Granberg S, Norstrom A, Wikland M. Tumors in the lower pelvis as imaged by vaginal sonography. Gynecol Oncol 1990; 37: 224 229. 10. Sassone AM, Timor-Tritsch IE, Artner A, Westhoff C, Warren WB. Transvaginal sonographic characterization of ovarian disease: evaluation of a new scoring system to predict ovarian malignancy. Obstet Gynecol 1991; 78: 70 76. 11. Lerner JP, Timor-Tritsch IE, Federman A, Abramovich G. Transvaginal ultrasonographic characterization of ovarian masses with an improved, weighted scoring system. Am J Obstet Gynecol 1994; 170: 81 85. 12. Jacobs I, Oram D, Fairbanks J, Turner J, Frost C, Grudzinskas JG. A risk of malignancy index incorporating CA125, ultrasound and menopausal status for the accurate pre-operative diagnosis of ovarian cancer. BrJObstetGynaecol1990; 97: 922 929. 13. Tailor A, Jurkovic D, Bourne TH, Collins WP, Campbell S. Sonographic prediction of malignancy in adnexal masses using multivariate logistic regression analysis. Ultrasound Obstet Gynecol 1997; 10: 41 47. 14. Timmerman D, Verrelst H, Bourne TH, De Moor B, Collins WP, Vergote I, Vandewalle J. Artificial neural network models for the preoperative discrimination between malignant and benign adnexal masses. Ultrasound Obstet Gynecol 1999; 13: 17 25. 15. Aslam N, Tailor A, Lawton F, Carr J, Savvas M, Jurkovic D. Prospective evaluation of three different models for the preoperative diagnosis of ovarian cancer. BrJObstetGynaecol 2000; 107: 1347 1353. 16. Valentin L, Hagen B, Tingulstad S, Eik-Nes S. Comparison of pattern recognition and logistic regression models for discrimination between benign and malignant pelvic masses: a prospective cross validation. Ultrasound Obstet Gynecol 2001; 18: 357 365. 17. Timmerman D, Schwarzler P, Collins WP, Claerbout F, Coenen M, Amant F, Vergote I, Bourne TH. Subjective assessment of adnexal masses with the use of ultrasonography: an analysis of interobserver variability and experience. Ultrasound Obstet Gynecol 1999; 13: 11 16. 18. Exacoustos C, Romanini ME, Rinaldo D, Amoroso C, Szabolcs B, Zupi E, Arduini D. Preoperative sonographic features of borderline ovarian tumors. Ultrasound Obstet Gynecol 2005; 25: 50 59. 19. Pascual MA, Tresserra F, Grases PJ, Labastida R, Dexeus S. Borderline cystic tumors of the ovary: gray-scale and color Doppler sonographic findings. J Clin Ultrasound 2002; 30: 76 82. 20. Wu CC, Lee CN, Chen TM, Shyu MK, Hsieh CY, Chen HY, Hsieh FJ. Incremental angiogenesis assessed by color Doppler ultrasound in the tumorigenesis of ovarian neoplasms. Cancer 1994; 73: 1251 1256. 21. Fruscella E, Testa AC, Ferrandina G, De Smet F, Van Holsbeke C, Scambia G, Zannoni GF, Ludovisi M, Achten R, Amant F, Vergote I, Timmerman D. Ultrasound features of different histopathological subtypes of borderline ovarian tumors. Ultrasound Obstet Gynecol 2005; 26: 644 650. 22. Gotlieb WH, Soriano D, Achiron R, Zalel Y, Davidson B, Kopolovic J, Novikov I, Ben-Baruch G. CA 125 measurement and ultrasonography in borderline tumors of the ovary. Am J Obstet Gynecol 2000; 183: 541 546. 23. EFSUMB Newsletter. Minimum training recommendations for the practice of medical ultrasound. Ultrashall Med 2005; 26: 79 105. 24. Hillaby K, Aslam N, Salim R, Lawrence A, Raju KS, Jurkovic D. The value of detection of normal ovarian tissue (the ovarian crescent sign ) in the differential diagnosis of adnexal masses. Ultrasound Obstet Gynecol 2004; 23: 63 67. 25. Yazbek J, Aslam N, Tailor A, Hillaby K, Raju KS, Jurkovic D. A comparative study of the risk of malignancy index and the ovarian crescent sign for the diagnosis of invasive ovarian cancer. Ultrasound Obstet Gynecol 2006; 28: 320 324. 26. Shepherd JH. Revised FIGO staging for gynaecological cancer. Br J Obstet Gynaecol 1989; 96: 889 892. 27. Webb PM, Purdie DM, Grover S, Jordan S, Dick ML, Green AC. Symptoms and diagnosis of borderline, early and advanced epithelial ovarian cancer. Gynecol Oncol 2004; 92: 232 239. 28. Timmerman D, Valentin L, Bourne TH, Collins WP, Verrelst H, Vergote I. Terms, definitions and measurements to describe the sonographic features of adnexal tumors: a consensus opinion from the International Ovarian Tumor Analysis (IOTA) group. Ultrasound Obstet Gynecol 2000; 16: 500 505. 29. Fauvet R, Boccara J, Dufournet C, David-Montefiere E, Poncelet C, Darai E. Restaging surgery for women with borderline ovarian tumors: results of a French multicenter study. Cancer 2004; 100: 1145 1151. 30. Lin P, Falcone T, Tulandi T. Excision of ovarian dermoid by laparoscopy and by laparotomy. Am J Obstet Gynecol 1995; 173: 769 771. 31. Howard FM. Surgical management of benign cystic teratoma: laparoscopy vs. laparotomy. J Reprod Med 1995; 40: 495 499. 32. Mais V, Ajossa S, Piras B, Marongiu D, Guerriero S, Melis GB. Treatment of nonendometriotic benign adnexal cysts: a randomized comparison of laparoscopy and laparotomy. Obstet Gynecol 1995; 86: 770 774. 33. Fauvet R, Boccara J, Dufournet C, Poncelet C, Darai E. Laparoscopic management of borderline ovarian tumors: results of a French multicenter study. Ann Oncol 2005; 16: 403 410. 34. Valentin L. Comparison of Lerner score, Doppler ultrasound examination, and their combination for discrimination between benign and malignant adnexal masses. Ultrasound Obstet Gynecol 2000; 15: 143 147.