Sonographic assessment of non-malignant ovarian cysts: does sonohistology exist?

Transcription

1 Human Reproduction Vol.19, No.9 pp , 2004 Advance Access publication June 10, 2004 DOI: /humrep/deh353 Sonographic assessment of non-malignant ovarian cysts: does sonohistology exist? C.D.de Kroon 1, H.A.G.M.van der Sandt 1, J.C.van Houwelingen 2 and F.W.Jansen 1,3 1 Department of Gynaecology and 2 Department of Medical Statistics, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands 3 To whom correspondence should be addressed. f.w.jansen@lumc.nl BACKGROUND: Transvaginal ultrasound (TVU) is feasible and accurate in the differentiation between nonmalignant and malignant ovarian abnormalities. However, despite the clinical relevance, the accuracy of TVU in the differentiation between the many different non-malignant cysts is unknown. METHODS: Between 1992 and 2002, all women who had surgery at our centre because of a non-malignant ovarian cyst were included prospectively in this study. The sonographic characteristics as well as the expected histological diagnosis (the sonohistological diagnosis ) were evaluated pre-operatively. This diagnosis was compared with the histopathological diagnosis, and diagnostic parameters [with 95% confidence interval (CI)] of the sonohistological diagnosis were calculated. Logistic models, with the sonographic characteristics as variables, were constructed for each histopathological diagnosis. RESULTS: A total of 406 women were included consecutively. The overall diagnostic accuracy of the sonohistological diagnosis was 60% (95% CI ). Only in cases of simple ovarian cysts did the diagnostic accuracy of the respective logistic model exceed that of the sonohistological diagnosis (0.88 versus 0.81, P<0.01). The diagnostic accuracy of the sonohistological diagnosis for endometriotic and dermoid ovarian cysts was significantly better compared with the respective logistic model (0.84 versus 0.71, P < 0.01 and 0.87 versus 0.82, P , respectively). CONCLUSION: In approximately half of the non-malignant ovarian cysts, TVU is capable of distinguishing between the different histopathological diagnoses of non-malignant ovarian masses. Only in the diagnosis of simple ovarian cysts might use of the logistic models be helpful. Key words: adnexal disease/differential diagnosis/logistic regression/sensitivity and specificity/ultrasonography Introduction Transvaginal ultrasound (TVU) has been shown to be highly accurate in the evaluation of ovarian cysts. However, malignancy can only be excluded by histopathological examination (Timmerman et al., 1999a). Differentiation between nonmalignant and malignant ovarian cysts is of main importance to both the patient and the gynaecologist: in most institutions, the type of surgery (laparoscopy versus laparotomy) depends on the estimated probability of malignancy (Kinkel et al., 2000). Therefore, many prognostic models for differentiation between malignant and non-malignant ovarian cysts, some including Doppler criteria, have been published and validated (Rottem et al., 1990; Sassone et al., 1991; Lerner et al., 1994; Tailor et al., 1997; Brown et al., 1998; Osmers et al., 1998; Reimer et al., 1999; Timmerman et al., 1999b). However, in daily practice, the accuracy of these prognostic models may be less compared with the accuracy that was published in the initial studies (Valentin, 1999; Mol et al., 2001). It is also shown that subjective sonographic assessment of an ovarian cyst is at least as accurate as the more complex prognostic models in the differentiation between malignant and non-malignant abnormal ovaries (Timmerman et al., 1999a; Valentin et al., 2001). Differentiation between the many different cystic ovarian abnormalities with non-malignant features is relevant as well since proper treatment depends on the histological origin of the abnormality. Surgical management of endometriosis has been proven favourable in cases of infertility (Marcoux et al., 1997; Olive and Pritts, 2002). On the other hand, a study by Eriksson et al. (1985) demonstrated that two-thirds of the cystic ovaries that required surgery were found to be functional cysts (luteal mass or simple cyst). Surgical management of these cysts may not be beneficial in comparison with either medical treatment in the case of a luteal mass or expectant management in cases of a simple cyst (MacKenna et al., 2000). Moreover, aspiration or laparoscopic fenestration instead of cystectomy of these cysts may be beneficial because of the low recurrence rate, especially in cases of a simple cyst (de Crespigny et al., 1989; De Wilde, 1989). In this context, it is important to remember that the diagnosis simple cyst is based purely on ultrasound findings: a simple cyst is a unilocular, round, anechogenic cyst with a regular 2138 Human Reproduction vol. 19 no. 9 q European Society of Human Reproduction and Embryology 2004; all rights reserved

2 Sonographic assessment of benign ovarian masses and thin wall with a maximum diameter of 5 cm (Conway et al., 1998). For these types of cysts, even in post-menopausal women, it is extremely unlikely to be malignant (Conway et al., 1998). Finally, spill of the cystic contents of an ovarian abnormality may result in severe morbidity, especially in cases of a dermoid cyst (Canis et al., 1994; Parker et al., 1996). Since the different non-malignant ovarian masses do have distinct sonographic characteristics, TVU may be able to differentiate these cysts pre-operatively and allow for proper treatment (Mais et al., 1993; Atri et al., 1994; Kinkel et al., 2000; Jermy et al., 2001). The objective of this prospective study was to determine the accuracy of subjective ovarian mass assessment in the prediction of the histopathological diagnosis of non-malignant cystic ovarian masses. Moreover, the accuracy of the subjective assessment was compared with the accuracy of the logistic models that we developed for each histopathological diagnosis. Materials and methods All women who had surgery in Leiden University Medical Center (LUMC) between January 1, 1992 and July 31, 2002 because of a TVU-evaluated cystic ovarian mass that was expected to be non-malignant were eligible for this study and were included prospectively. In this department, TVU is performed according to international guidelines (ACOG Technical Bulletin, 1995) by skilled, experienced (at least 500 supervised pelvic scans) and specially trained sonographers. TVU was performed with a multifrequency vaginal probe (5 7.5 MHz) and up to date ultrasound machines (Toshiba Medical Systems, Zoetermeer, The Netherlands). The protocol for sonographic assessment of ovarian cysts includes evaluation of the size (in mm; mean of three orthogonal diameters) and the most indicative sonographic characteristics of ovarian cysts as obtained from the literature and displayed in Table I (Kinkel et al., 2000). Colour Doppler flow imaging and measurement of Doppler indices were not performed on a routine basis. All cystic masses with a Sassone score.9 were excluded because these cystic masses are more likely to be malignant (Sassone et al., 1991). Apart from the ovarian cyst assessment according to the protocol, the sonographer stated, for use in this study only, the expected histopathological diagnosis from a list of possible diagnoses. The list contained the following histopathological diagnoses: endometriotic cyst, dermoid cyst, simple cyst, mucinous adenoma, serous adenoma, fibroma, pseudo (or para-ovarian) cyst and luteal cyst. This Table I. Sonographic characteristics assessed for each ovarian cyst Sonographic characteristic Possible outcome Solid components None (1), hyperechogenic (2) or non-hyperechogenic (3) Locularity Unilocular (1) or multilocular (2) Echogeniety Anechogenic (1), echogenic (2) or mixed echogenic (3) Regularity of shape Regular (1) or irregular (2) Wall,3 mm (1), $3 mm and smooth (2), $3 mm and papillary formations (3) Septa None (1),,3 mm (2), $3 mm (3) Location Ovarian (1) or para-ovarian (2) Intraperitoneal fluid Absent (1) or present (2) The values in parentheses were used to evaluate the characteristics in the logistic model. was called the sonohistological diagnosis. Together with pictures of the ovarian cyst and a copy of the results of the standardized sonography, the sonohistological diagnosis was recorded on a separate sheet and not disclosed to the referring physician. The referring physician decided about further treatment according to the clinical situation and results of the TVU, which were explained and discussed in cooperation with the sonographer. The department of pathology examined the specimen; the pathologist was unaware of the ultrasound findings. The final histopathological diagnosis was compared with the sonohistological diagnosis and diagnostic parameters [diagnostic accuracy, sensitivity, specificity, negative predictive value, positive predictive value, pre- and posttest probabilities, likelihood ratios (LRs) and accuracy], and 95% confidence intervals (95% CIs) were calculated. For each histopathological diagnosis, we constructed a logistic model according to the method suggested by Bland (2000). The models provide the estimated probability of a specific ovarian cyst to be of the respective histopathological origin. This probability was equal to 1/(1 þ e 2z ) in which the respective value for z for each model is calculated resulting in the highest accuracy, and e is the base value of natural logarithms. The different sonographic criteria as obtained from the meta-analysis by Kinkel et al. (2000) were valued as displayed in Table I and used as factors in the models to calculate z. The histopathological diagnosis was used as the dependent variable. The significance of each sonographic characteristic was assessed by means of the multivariate odds ratio (OR; with 95% CI). Finally, we compared the accuracy of each logistic model with that of the respective sonohistological diagnosis. The Statistical Package for Social Sciences (SPSS, version 10, SPSS Inc., Chicago, IL) was used for all statistical calculations. The x 2 test and Fisher s exact test were used, whichever was most appropriate, to compare categorical data, while Student s t-test was used to compare continuous parametric data. Significance was achieved if P, The diagnostic parameters and 95% CI were calculated with the use of senspc.asp. Results In the designated study period, 463 women could be included consecutively. In all these women, TVU showed an ovarian cyst, which was expected to be non-malignant, that required surgery. Of these, 57 women (12.3%) had to be excluded since the sonographer did not explicitly state the sonohistological diagnosis. Finally, 406 women (87.7%) could be included in the analyses. The mean age of these women was 39.3 years (SD 11.7; 7 88 years). TVU was performed because of abdominal pain (n ¼ 219, 55.2%), irregular bleeding (n ¼ 59, 14.3%) or because of abnormal findings at Table II. Histological diagnoses of included non-malignant ovarian cysts (n ¼ 406) compared with the excluded ovarian cysts (n ¼ 57) Histological diagnosis Included n (%) Excluded n (%) P-value Simple cyst 57 (14.1) 8 (14.1) 0.96 Endometriotic cyst 124 (30.5) 11 (19.3) 0.09 Dermoid 99 (24.4) 11 (19.3) 1.00 Serous 44 (10.8) 7 (12.3) 0.71 Mucinous 46 (11.3) 12 (21.1) 0.01 Fibroid 7 (1.7) 2 (3.5) 0.35 Pseudocyst 7 (1.7) 2 (3.5) 0.35 Luteal cyst 22 (5.4) 2 (3.5)

3 C.D.de Kroon et al. Table III. Comparison of the sonohistological and histopathological diagnosis Histopathological diagnosis Total Simple cyst a Endometriotic cyst Dermoid cyst Serous Mucinous Fibroma Pseudocyst Luteal cyst Sonohistological diagnosis Simple cyst Endometriotic cyst Dermoid cyst Serous Mucinous Fibroma Pseudocyst Luteal cyst Total a All cystoma simplex except for three follicular cysts. abdominal sonography at the Department of Radiology of our hospital (n ¼ 48, 11.8%). In 80 women (19.7%), TVU was performed for various other reasons. In Table II, the histological diagnoses of the excluded women are compared with those of the included women. In the excluded women, significantly more women had a mucinous (P ¼ 0.01). The size of the cystic masses in both groups did not differ significantly (P ¼ 0.45). In Table III, the sonohistological diagnoses are compared with the histopathological diagnoses. Since no malignancies were observed at histopathology, all women suspected of a malignant ovarian cyst were excluded successfully. All 16 women who had surgery because of a luteal mass suffered from abdominal pain. The mean size of these cysts was 5.5 cm (range cm). The overall agreement between the sonohistological and histopathological diagnosis was 60.3% (95% CI , n ¼ 245). Table IV shows the diagnostic parameters of the sonohistological diagnosis. Although the diagnostic accuracies are high for the most prevalent cysts (e.g. dermoid, simple and endometriotic cysts: 0.87, 0.81 and 0.84, respectively), the highest accuracy is observed in ovarian cysts with the lowest prevalence (fibroma and pseudocyst: prevalence 0.02 and 0.02, diagnostic accuracy 0.99 and 0.98). For each histopathological diagnosis, a logistic regression model was developed (Table V). The diagnostic accuracy of the logistic model was compared with that of the respective sonohistological diagnosis. In the diagnosis of an endometriotic cyst (0.84 versus 0.71; P, 0.01) and a dermoid cyst (0.87 versus 0.82; P ¼ 0.03), the sonographer performs significantly better compared with the logistic model. However, the logistic model performs significantly better in the diagnosis of a simple ovarian cyst (0.88 versus 0.81; P, 0.01). In all other diagnoses, the subjective assessment, by means of the sonohistological diagnosis, and the logistic model performed equally accurately (Table V). The multivariant ORs, as calculated from the multiregression analysis, are displayed in Table VI. Anechogenicity (OR 2.40, 95% CI ) and ovarian location (OR 0.24, 95% CI ) are the significant variables in the logistic model for the simple ovarian cyst. Size (OR 1.01, 95% 2140 CI ), the absence of solid components (OR 0.59, 95% CI ) and the absence of septa (OR 0.57, 95% CI ) are the significant variables in the logistic model for endometriotic cysts. Moreover, dermoid cysts are characterised by hyperechogenic components (OR 0.33, 95% CI ) and thick walls (OR 0.77, 95% CI ). Discussion The results of our prospective study show that sonographers are only moderately able to predict the histopathological diagnosis of non-malignant ovarian cysts accurately according to their findings at TVU. In only 60% of the assessed non-malignant ovarian cysts was the sonohistological diagnosis in accordance with the histopathological diagnosis. The use of logistic models does not improve the performance of TVU in the diagnosis of these ovarian cysts. Only in the case of simple cysts did the developed logistic model performed better compared with the subjective assessment. The value of a diagnostic test can be assessed by its LRs. A system for scoring the positive and negative LRs (LR þ and LR 2, respectively) has been developed by Jaeschke et al. (1994). A test can be considered useless if the LR þ was,2 or the LR 2 was.0.5 while the value and usefulness of a test increases with increasing LR þ and decreasing LR 2. As displayed in Table IV, the LR þ are between 4.0 (simple cyst) and 85.5 (pseudocyst) and the LR 2 between 0.84 (serous cysteadenoma) and 0.21 (dermoid cyst). According to Jeaschke s scoring system, the performance of the subjective assessment of non-malignant ovarian cysts by TVU is small to considerable. However, our study may be biased by the exclusion of possible malignant ovarian cysts. Moreover, 57 (12%) of the initially included ovarian cysts had to be excluded since the sonographer did not state a specific histopathological diagnosis. Although we are unaware of the reason why the sonographer failed to state a diagnosis, it may be because the diagnosis was dificult to determine. The exclusion of these cysts may also have resulted in the rather surprising finding that none of the more seldomly observed histological findings

4 Table IV. Diagnostic parameters of the sonohistological diagnoses Sonohistological diagnosis Diagnostic parameters (95% CI) Pre-test probability Diagnostic accuracy Sensitivity Specificity Positive predictive value Negative predictive value LR þ test LR test Simple cyst 0.14 ( ) 0.81 ( ) 0.67 ( ) 0.83 ( ) 0.40 ( ) 0.94 ( ) 4.00 ( ) 0.39 ( ) Endometriotic cyst 0.30 ( ) 0.84 ( ) 0.75 ( ) 0.87 ( ) 0.72 ( ) 0.89 ( ) 5.94 ( ) 0.28 ( ) Dermoid cyst 0.24 ( ) 0.87 ( ) 0.80 ( ) 0.89 ( ) 0.71 ( ) 0.93 ( ) 7.51 ( ) 0.21 ( ) Serous 0.11 ( ) 0.88 ( ) 0.18 ( ) 0.96 ( ) 0.42 ( ) 0.91 ( ) 5.98 ( ) 0.84 ( ) Mucinous 0.11 ( ) 0.90 ( ) 0.30 ( ) 0.97 ( ) 0.61 ( ) 0.92 ( ) 12.2 ( ( ) Fibroma 0.02 ( ) 0.99 ( ) 0.28 ( ) 1.00 ( ) 1.00 ( ) Pseudocyst 0.02 ( ) 0.98 ( ) 0.43 ( ) 0.99 ( ) 0.60 ( ) 0.99 ( ) 85.5 ( ) 0.57 ( ) Luteal cyst 0.05 ( ) 0.93 ( ) 0.23 ( ) 0.97 ( ) 0.31 ( ) 0.96 ( ) 7.93 ( ) 0.79 ( ) Table V. Comparison of the diagnostic accuracy of the sonohistological diagnosis and the logistic models 2141 Diagnostic accuracy sonohistology Diagnostic accuracy logistic model P-value Logistic model a : Simple cyst 0.81 ( ) 0.88 ( ), þ (0.02*A) þ (0.23*B) 2 (0.17*C) þ (0.87*D) þ (1.24*E) þ (0.32*F) þ (1.16*G) 2 (1.42*H) þ (0.4 1*I)I Endometriotic cyst 0.84 ( ) 0.71 ( ), þ (0.02*A) 2 (0.52*B) þ (0.12*C) 2 (0.17*D) þ (0.17*E) þ (0.02*F) 2 (0.57*G) þ (1,50*H) þ (0.48*I) Dermoid cyst 0.87 ( ) 0.82 ( ) (0.01*A) þ (0.31*B) þ (0.64*C) 2 (1,11*D) 2 (0.00*E) 2 (0.26*F) þ (0.25*G) þ (5,74*H) þ (0.17*I) Serous 0.88 ( ) 0.89 ( ) (0.01*A) þ (0.11*B) 2 (0.28*C) þ (1.18*D) þ (0.92*E) þ (0.20*F) 2 (0.15*G) þ (0.58*H) þ (0.57*I) Mucinous 0.90 ( ) 0.89 ( ) ,50 2 (0.03*A) 2 (0.21*B) 2 (0.74*C) þ (0.34*D) 2 (0.81*E) þ (0.07*F) þ (0.10*G) þ (0.65*H) þ (0.08*I) Fibroma 0.99 ( ) 0.98 ( ) ,0 þ (0.02*A) 2 (0.86*B) 2 (0.45*C) þ (0.30*D) þ (7.31*E) 2 (0.29*F) 2 (0.35*G) þ (6.34*H) þ (7.26*I) Pseudocyst 0.98 ( ) 0.99 ( ) ,8 2 (0.01*A) þ (2.14*B) 2 (0.35*C) þ (0.86*D) 2 (2.65*E) 2 (0.70*F) þ (0.87*G) 2 (3.25*H) 2 (2.70*I) Luteal cyst 0.93 ( ) 0.95 ( ) þ (0.02*A) þ (0.13*B) 2 (2.28*C) 2 (0.07*D) 2 (1.10*E) þ (0.15*F) þ (1.76*G) þ (7.23*H) 2 (0.77*I) a A ¼ size in mm; B ¼ presence of solid components; C ¼ locularity, D ¼ echogenicity; E ¼ regularity of shape; F ¼ wall; G ¼ septa; H ¼ location; and I ¼ presence of peritoneal fluid. (B 2 I as valued in Table I). Sonographic assessment of benign ovarian masses

5 C.D.de Kroon et al. Table VI. Multivariate odds ratios (95% CI) of the sonographic characteristics in each logistic model with significant variables in bold Sonographic characteristics inserted as variable in the logistic model Size Solid components Locularity Echogenicity Regularity of shape Wall Septa Location Peritoneal fluid Logistic model Simple cyst 1.02 ( ) 1.26 ( ) 0.84 ( ) 2.40 ( ) 3.48 ( ) 1.37 ( ) 3.20 ( ) 0.24 ( ) 1.51 ( ) Endometriotic 1.02 ( ) 0.59 ( ) 1.13 ( ) 0.85 ( ) 1.18 ( ) 1.03 ( ) 0.57 ( ) 4.47 ( ) 1.61 ( ) cyst Dermoid cyst 0.99 ( ) 1.36 ( ) 1.89 ( ) 0.33 ( ) 1.00 ( ) 0.77 ( ) 1.29 ( ) 313 ( ) 1.18 ( ) Serous 0.99 ( ) 1.11 ( ) 1.33 ( ) 3.27 ( ) 2.50 ( ) 1.22 ( ) 0.86 ( ) 1.79 ( ) 1.76 ( ) Mucinous 0.97 ( ) 0.80 ( ) 0.48 ( ) 1.40 ( ) 0.44 ( ) 1.07 ( ) 1.10 ( ) 1.91 ( ) 0.92 ( ) Fibroma 1.02 ( ) 0.42 ( ) 0.64 ( ) 1.35 ( ) 1500 (0.00 1) 0.74 ( ) 0.70 ( ) 564 (0.00 1) 1431 (0.00 1) Pseudo cyst 0.98 ( ) 8.51(0.06 1) 0.70 ( ) 2.36 ( ) 0.07 ( ) 0.49 ( ) 2.40 ( ) 0.04 ( ) 0.07 ( ) Luteal cyst 1.02 ( ) 1.14 ( ) 0.10 ( ) 0.93 ( ) 0.33 ( ) 1.16 ( ) 5.84 ( ) 1385 (0.00 1) 0.45 ( ) among ovarian cysts, such as Brenners tumour, hydrosalpinx, tubo-ovarian abscess and fibromas, were present in our sample of ovarian cysts. However, except for mucinous s (Table I), the histological diagnoses between excluded and included ovarian cysts did not differ in both groups. Therefore, we do not think exclusion of these cysts will have biased our results considerably. It is well established that non-malignant ovarian abnormalities do have distinct sonographic characteristics (Table V) (Mais et al., 1993; Atri et al., 1994; Kinkel et al., 2000; Jermy et al., 2001). These well-established objective ultrasound characteristics were used in our logistic models. In accordance with the literature, our results indicate that simple cysts are characterized by anechogenicity (OR 2.40) and dermoid cysts are significantly characterized by hyperechogenic components (OR 0.33). The absence of solid components and the presence of septa characterize endometriotic cysts (ORs 0.57 and 0.59, respectively). Although these characteristics are significant, the overall heterogeneity of the sonographic characteristics of non-malignant ovarian cysts prevents complete differentiation of these cysts by TVU. However, common ultrasound characteristics, such as the ground glass appearance of endometriotic cysts (Patel et al., 1999) or the typical bright white line representing hair in dermoid cysts, have not been included in our logistic models. These ultrasound characteristics are typically used in the subjective assessment of ovarian cysts in contrast to the objective characteristics that we included in the logistic regression analysis. Our results show that TVU is able to diagnose 77% of the functional cysts accurately. In particular, the differentiation between functional and non-functional cysts is of great clinical importance in order to refrain from surgery whenever possible (MacKenna et al., 2000). However, TVU as a diagnostic tool is only of small value in the diagnosis of a functional cyst since the LR þ and LR 2 are 3.23 and 0.45, respectively. Additional cytological aspiration of the cystic fluid, advocated by some authors because of the combination of diagnosis and therapy (Balat et al., 1996), does not seem meaningful because of the low sensitivity (Jansen et al., 1997; Ganjei et al., 1999) and the risk of spilling in the case of malignancy (Vergote et al., 2001). Moreover, the resolution of functional cysts did not differ between women who had their cysts aspirated and those who had their cysts observed by frequent sonography (Zanetta et al., 1996). Especially in women in whom assisted reproduction is being considered, the accurate diagnosis of endometriotic cysts is important since removal of these cysts improves treatment outcome and pregnancy rate in these women (Marcoux et al., 1997; Olive and Pritts, 2002). However, the diagnostic power of TVU in the diagnosis of endometriotic cysts is small (LR þ and LR 2, 4.0 and 0.39, respectively). In combination with clinical parameters such as dysmenorrhoea, the predictive value of the diagnostic pathway for endometriotic cysts may increase. Patel et al. (1999) showed better performance of ultrasound in the prediction of endometriotic cysts, but in their retrospective study the endometriotic cysts had to be distinguished from the other types of 2142

6 Sonographic assessment of benign ovarian masses ovarian cysts. Therefore, comparing their data with ours is difficult. In conclusion, our results show that TVU lacks sufficient accuracy in the differentiation of the various non-malignant ovarian cysts. This low diagnostic accuracy is likely to be due to the heterogeneity of the sonographic characteristics of these cysts. Whenever further treatment depends on the histopathological diagnosis, one has to bear in mind that TVU will misdiagnose one-third of ovarian masses. Especially in the case of failed medical treatment or expectant management according to the sonohistological diagnosis and in women in whom assisted reproduction is considered, diagnostic laparoscopy with histopathological examination by cystectomy is still warranted. References ACOG Technical Bulletin (1995) Gynecological ultrasonography. Number 215. Int J Gynecol Obstet 53, Atri M, Nazarnia S, Bret PM, Aldis AE, Klintzen G and Reinhold C (1994) Endovaginal sonographic appearance of benign ovarian masses. Radiograph 14, Balat O, Sarac K and Sonmez S (1996) Ultrasound guided aspiration of benign ovarian cysts: an alternative to surgery? Eur J Radiol 22, Bland M (2000) Multifactorial methods. In Bland (ed.) An Introduction to Medical Statistics, 3rd edn. Oxford University Press, Oxford, pp Brown DL, Doubilet PM, Miller FH, Frates MC, Laing FC, DiSalvo DN, Benson CB and Lerner MH (1998) Benign and malignant ovarian masses: selection of the most discriminating gray-scale and Doppler sonographic features. Radiology 208, Canis M, Mage G, Pouly JL, Wattiez A, Manhes H and Bruhat MA (1994) Laparoscopic diagnosis of adnexal cystic masses: a 12-year experience with long-term follow-up. Obstet Gynecol 83, Conway C, Zalud I, Dilena M, Maulik D, Schulman H, Haley J and Simonelli K (1998) Simple cyst in the postmenopausal patient: detection and management. J Ultrasound Med 17, de Crespigny LCh, Robinson HP, Davoren RAM and Fortune DW (1989) The simple ovarian cyst: aspirate or operate. Br J Obstet Gynaecol 96, De Wilde RL (1989) Recurrence of functional ovarian cysts after laparoscopic fenestration. Am J Obstet Gynecol 161,839. Eriksson L, Kjellgren O and von Schoultz B (1985) Functional cyst or ovarian cancer: histopathological findings during 1 year of surgery. Gynecol Obstet Invest 19, Ganjei P, Dickinson B, Harrison T, Nassiri M and Lu Y (1996) Aspiration cytology of neoplastic and non-neoplastic ovarian cysts: is it accurate? Int J Gynecol Pathol 15, Jaeschke R, Guyatt GH and Sackett DL (1994) User s guide to medical literature III. How to use an article about a diagnostic test. A. Are the results of the study valid? J Am Med Assoc 271, Jansen FW, Tanahatoe S, Veselic M and Trimbos JB (1997) Laparoscopic aspiration of ovarian cysts: an unreliable technique in primary diagnosis of (sonographically) benign ovarian lesions. Gynaecol Endosc 6, Jermy K, Luise C and Bourne TH (2001) The characterization of common ovarian cysts in premenopausal women. Ultrasound Obstet Gynecol 17, Kinkel K, Hricak H, Lu Y, Tsuda K and Filly RA (2000) US characteristics of ovarian massess: a meta-analysis. Radiology 217, Lerner JP, Timor-Tritsch IE, Federman A and Abramovich G (1994) Transvaginal ultrasonographic characterization of ovarian masses with an improved, weighted scoring system. Am J Obstet Gynecol 170, MacKenna A, Fabres C, Alam V and Morales V (2000) Clinical management of functional ovarian cysts: a prospective and randomized study. Hum Reprod 15, Mais V, Guerriero S, Ajossa S, Angiolucci M, Paoletti AM and Melis GB (1993) The efficiency of transvaginal ultrasonography in the diagnosis of endometrioma. Fertil Steril 60, Marcoux S, Maheux R and Berube S (1997) Laparoscopic surgery in infertile women with minimal or mild endometriosis: Canadian Collaborative Group in Endometriosis. N Engl J Med 337, Mol BWJ, Boll D, De Kanter M, Heintz APM, Sijmons EA, Oei SG, Bal H and Brölmann HAM (2001) Distinguishing the benign and malignant adnexal mass: an external validation of prognostic models. Gynecol Oncol 80, Olive DL and Pritts EA (2002) The treatment of endometriosis: a review of evidence. Ann NY Acad Sci 955, Osmers RG, Osmers M, von Maydell B, Wagner B and Kuhn W (1998) Evaluation of ovarian tumors in postmenopausal women by transvaginal sonography. Eur J Obstet Gynecol Reprod Biol 77, Parker WH, Childers JM, Canis M, Phillips DR and Topel H (1996) Laparoscopic management of benign cystic teratomas during pregnancy. Am J Obstet Gynecol 174, Patel MD, Feldstein VA, Chen DC, Lipson SD and Filly RA (1999) Endometriomas: diagnostic performance of US. Radiology 210, Reimer T, Gerber B, Muller H, Jeschke U, Krause A and Friese K (1999) Differential diagnosis of peri- and postmenopausal ovarian cysts. Maturitas 4, Rottem S, Levit N, Thaler I, Yoffe N, Bronshtein M, Manor D and Brandes JM (1990) Classification of ovarian lesions by high-frequency transvaginal sonography. J Clin Ultrasound 18, Sassone AM, Timor-Tritsch IE, Artner A, Westhoff C and Warren WB (1991) Transvaginal sonographic characterization of ovarian disease: evaluation of a new scoring system to predict ovarian malignancy. Obstet Gynecol 78, Tailor A, Jurkovic D, Bourne TH, Collins WP and Campbell S (1997) Sonographic prediction of malignancy in adnexal masses using multivariate logistic regression analysis. Ultrasound Obstet Gynecol 10, Timmerman D, Bourne TH, Tailor A, Collins WP, Verrelst H, Vandenberghe K and Vergote I (1999a) A comparison of methods for preoperative discrimination between malignant and benign adnexal masses: the development of a new logistic regression model. Am J Obstet Gynecol 181, Timmerman D, Schwärzler P, Collins WP, Claerhout F, Coenen M, Amant F, Vergote I and Bourne TH (1999b) Subjective assessment of adnexal masses with the use of ultrasonography: an analysis of interobserver variability and experience. Ultrasound Obstet Gynecol 13, Valentin L (1999) Prospective cross-validation of Doppler ultrasound examination and gray-scale ultrasound imaging for discrimination of benign and malignant pelvic masses. Ultrasound Obstet Gynecol 14, Valentin L, Hagen B, Tingulstad S and Eik-Nes S (2001) Comparison of pattern recognition and logistic regression models for discrimination between benign and malignant pelvic masses: a prospective cross validation. Ultrasound Obstet Gynecol 18, Vergote I, De Brabanter J, Fyles A, Bertelsen K, Einhorn N, Sevelde P, Gore ME, Kaern J, Verrelst H, Sjövall K et al. (2001) Prognostic importance of degree of differentiation and cyst rupture in stage I epithelial ovarian carcinoma. Lancet 357, Zanetta G, Lissoni A, Torri V, Dalla VC, Trio D, Rangoni G and Mangioni C (1996) Role of puncture and aspiration in expectant management of simple ovarian cysts: a randomised study. Br Med J 313, Submitted on October 27, 2003; accepted on May 11,