Likelihood Ratio of Sonographic Findings in Discriminating Hydrosalpinx from Other Adnexal Masses

Transcription

1 Sonography of Adnexal Masses Women s Imaging Original Research WOMEN S IMAGING Maitray D. Patel 1 Debra L. Acord 1,2 Scott W. Young 1 Patel MD, Acord DL, Young SW Keywords: adnexal masses, hydrosalpinx, ovarian neoplasm, pelvic imaging, sonography, women s imaging DOI: /AJR Received January 18, 2005; accepted after revision February 22, All authors: Department of Radiology, Mayo Clinic, E Shea Blvd., Scottsdale, AZ Address correspondence to M. D. Patel (patel.maitray@mayo.edu). 2 Present address: Department of Radiology, Wesley Medical Center, Wichita, KS AJR 2006; 186: X/06/ American Roentgen Ray Society Likelihood Ratio of Sonographic Findings in Discriminating Hydrosalpinx from Other Adnexal Masses OBJECTIVE. The objective of our study was to describe the waist sign as a feature of hydrosalpinx and to calculate the likelihood ratio of sonographic findings for predicting that a cystic adnexal mass is a hydrosalpinx. MATERIALS AND METHODS. Consecutive institutional records were reviewed to identify patients with surgically excised cystic adnexal masses and preoperative sonograms that had been used to identify the mass. Two radiologists blinded to outcome reviewed the sonograms and recorded the presence of five specific sonographic findings: incomplete septation, short linear projection, small round projection, tubular shape, and presence of a waist. The likelihood ratios of these signs and combinations of signs for the diagnosis of hydrosalpinx were calculated. RESULTS. There were 55 patients with 67 cystic adnexal masses. Of the 67 cystic adnexal masses, there were 26 hydrosalpinges (39%), 36 cystic ovarian neoplasms (54%), and five paraovarian masses (7%). The likelihood ratios for the sonographic findings were as follows: incomplete septation, 2.1; short linear projection, 3.5; small round projection, 2.7; tubular shape, 10.5; and waist sign, The waist sign in combination with tubular shape was found in 12 hydrosalpinges and no other masses (likelihood ratio of between 18.9 and infinity). Small round projection combined with tubular shape was found in 14 hydrosalpinges and one other mass (likelihood ratio of 22.1). CONCLUSION. Hydrosalpinx can be diagnosed with the highest likelihood when a tubular mass with the waist sign or a tubular mass with small round projections is encountered. Incomplete septations and short linear projections are less discriminating findings of hydrosalpinx. rior studies have described various P sonographic features associated with hydrosalpinx [1 9]. These features include a tubular shape, the presence of incomplete septations or short linear projections, and the presence of small hyperechoic wall foci. The presence of a waist in a cystic collection, referring to diametrically opposed indentations in the wall of the collection, has not been previously described as a feature of hydrosalpinx, to our knowledge. The description of the waist sign as a feature of hydrosalpinx is one objective of this study. More importantly, no prior study has specifically evaluated the likelihood ratio of various sonographic findings for predicting that a cystic adnexal mass is a hydrosalpinx, which serves as the primary purpose of this investigation. Materials and Methods Institutional records between January 2000 and July 2003 were reviewed to identify consecutive patients with surgically excised cystic adnexal masses and preoperative sonograms that had been used to identify the mass. For inclusion in the study, the preoperative sonograms must have been obtained at our institution within 3 months of surgery. Patients with dermoid cysts, endometriomas, and hemorrhagic ovarian cysts at pathology as the cause of the sonographically identified mass were excluded. Furthermore, patients with sonographically suspected solid areas on preoperative sonography were excluded from the study. All sonography studies were performed using both transabdominal and endovaginal probes on an Acuson Sequoia unit (Siemens Medical Solutions), ATL-5000 HDI unit (Philips Medical Systems), or Logic 700 unit (GE Healthcare). Two radiologists, each with 1 year of dedicated sonography fellowship training and 10 or 5 years of subsequent clinical sonography experience, were blinded to clinical history and pathologic outcome; they reviewed the sonograms and recorded the consensus determination about the presence of specific sonographic findings. Images were reviewed electronically using an ALI review station (McKesson Medical Imaging Group). An incom- AJR:186, April

2 plete septation was defined as a linear protrusion arising from one wall of the mass extending more than one third of the distance to the opposite wall (Figs. 1 3). If the linear protrusion did not extend more than one third of the distance to the opposite wall, it was classified as a short linear projection (Figs. 1, 3, and 4). A short round projection was defined as a tiny 2- to 3-mm hyperechoic mural nodule (Figs. 4 and 5). The mass was subjectively assessed for tubular shape (Figs. 1 and 4 6). The waist sign referred to diametrically opposed indentations along the wall of the cystic mass (Figs. 1, 4, 5, and 7). Each finding was assessed independent from other findings without consideration of the sonologists overall subjective impression or expectation of the pathologic outcome. A Fig year-old woman with left hydrosalpinx causing cystic adnexal mass. A, Coronal sonogram of left adnexa shows mass, which was classified as having tubular shape, with incomplete septation (arrow) and waist sign (asterisks). B, Coronal sonogram of tubular cystic mass with slightly different orientation more clearly shows incomplete septation (arrow). Waist sign (asterisks) is less well shown on this image but is still evident. There is suggestion of second waist sign at level of incomplete septation, with subtle indentation of contralateral wall. Fig year-old woman with left ovarian cystadenofibroma. Oblique coronal sonogram shows incomplete septation (arrow). There is no indentation along contralateral wall to form waist sign. This mass did not have tubular shape. The final diagnosis was determined by review of the final written pathology report with correlation to the sonographically identified mass using surgical notes and sonographic findings. The likelihood ratios of each sonographic finding and combinations of findings for the diagnosis of hydrosalpinx were calculated. Results Fifty-five patients with 67 cystic adnexal masses met the inclusion criteria and constitute the study population. Patient age ranged from 26 to 87 years (mean, 62 years), but only six patients were less than 45 years old. Most patients were peri- or postmenopausal. None of the patients had clinical evidence of acute pelvic inflammatory disease. Fig year-old woman with right ovarian cystadenoma. Oblique coronal sonogram shows incomplete septation (arrowhead), short linear projection (thick arrow), and small round projection (thin arrow). Short linear projection, by arbitrary definition in this study, differs from incomplete septation in that it does not extend more than one third of the distance to the opposite wall. This mass did not have tubular shape. The sonographic findings identified in each mass category and the resulting likelihood ratio for the prediction of hydrosalpinx based on these findings and selected finding combinations are presented in Table 1. Of the 67 cystic adnexal masses, there were 26 hydrosalpinges (39%), 36 cystic ovarian neoplasms (54%), and five paraovarian masses (7%). Among the 36 ovarian neoplasms, there were 20 cystadenomas, 15 cystadenofibromas, and one cystadenocarcinoma. Among the five paraovarian masses, there were four paratubal cysts and one lymphangioma. At least one incomplete septation was found in 17 (65%) of the 26 hydrosalpinges and in 13 other masses, yielding a likelihood ratio of 2.1 (Figs. 1 3). At least B 1034 AJR:186, April 2006

3 Sonography of Adnexal Masses Fig year-old woman with right hydrosalpinx. Oblique coronal sonogram shows tubular cystic mass with short linear projection (arrow) and waist sign (asterisks). In this case, short linear projection forms half of waist. Fig year-old woman with left hydrosalpinx. A, Oblique coronal sonogram shows tubular cystic mass with waist sign (asterisks) and small round projection (arrow). B, Small round projection (arrow) is better shown on this sagittal image through mass. A B Fig year-old woman with right ovarian cystadenofibroma. Observers classified mass shape as tubular as shown on this sagittal sonogram. Fig year-old woman with right ovarian cystadenofibroma. Observers noted waist sign (asterisks) in this sagittal sonogram. Note that one half of waist is caused by obtuse indentation along wall, which differs from more common acute indentation seen with other hydrosalpinges. AJR:186, April

4 TABLE 1: Sonographic Findings Identified in Cystic Adnexal Masses and Their Calculated Likelihood Ratio for Prediction of Hydrosalpinx Sonographic Finding Hydrosalpinx (n = 26) Cystic Neoplasm (n = 36) a Paraovarian Mass (n =5) b Likelihood Ratio for Hydrosalpinx Incomplete septation Short linear projection Small round projection Tubular shape Waist sign Tubular shape and Incomplete septation Short linear projection Small round projection Waist sign a Twenty cystadenomas, 15 cystadenofibromas, and one cystadenocarcinoma. b Four paratubal cysts and one lymphangioma. one short linear projection was found in 20 hydrosalpinges (77%) and in nine other masses, yielding a likelihood ratio of 3.5 (Figs. 1, 3, and 4). At least one small round projection was found in 17 hydrosalpinges (65%) and in 10 other masses, yielding a likelihood ratio of 2.7 (Figs. 3 and 5). Tubular shape was found in 20 hydrosalpinges (77%) and in three other masses, yielding a likelihood ratio of 10.5 (Figs. 1 and 4 6). The waist sign was found in 13 hydrosalpinges (50%) and in one other mass, yielding a likelihood ratio of 20.5 (Figs. 1, 4, 5, and 7). Incomplete septation combined with tubular shape was found in 14 (54%) of the 26 hydrosalpinges and in three other masses, yielding a likelihood ratio of 7.4 (Fig. 1). Short linear projection combined with tubular shape was found in eight hydrosalpinges (31%) and in two other masses, yielding a likelihood ratio of 6.3 (Fig. 4). Small round projection combined with tubular shape was found in 14 (54%) of the 26 hydrosalpinges and in one other mass, yielding a likelihood ratio of 22.1 (Fig. 5). The waist sign in combination with tubular shape was found in 12 hydrosalpinges (46%) and no other masses, yielding a likelihood ratio between 18.9 and infinity (Figs. 1, 4, and 5). Discussion Previous studies have described sonographic findings associated with hydrosalpinx. Tessler and colleagues [1] found a tubular structure with folded configuration (forming an incomplete septation) as the most consistent feature in the 12 cases of hydrosalpinx in their study. Short linear projections were also seen in approximately half of their patients. Timor-Tritsch and colleagues [6] expanded the analysis of the sonographic features of hydrosalpinx by analyzing the shape of the mass, wall structure, wall thickness, and extent of ovarian involvement. Their analysis suggested that many hydrosalpinges were ovoid or pearshaped fluid collections containing incomplete septa, short linear projections (cogwheel sign), or small hyperechoic mural nodules ( beads-on-a-string ). Several other studies have used some of these findings as characteristic features of hydrosalpinx, with most emphasizing tubular shape with incomplete septations [3, 7, 8]. Of interest, in the study by Timor-Tritsch and colleagues, three false-positive cases using the findings associated with hydrosalpinx were a paraovarian cyst, ovarian cystadenoma, and appendiceal mucocele. Guerriero and colleagues [8] attempted to evaluate the accuracy of sonography in identifying hydrosalpinx in two groups of premenopausal patients. In their study, hydrosalpinx was diagnosed when an elongated mass with incomplete septa or hyperechoic mural nodules were identified. In one group of 378 premenopausal women (756 adnexa) undergoing surgical evaluation for a variety of conditions, including infertility, pelvic pain, uterine fibroids, endometrial hyperplasia, or adnexal masses, the authors reported presurgical sonographic identification of 22 of 26 hydrosalpinges; there were two false-positive diagnoses of hydrosalpinx that proved to be cystadenomas at surgical evaluation using their criteria. Because the authors did not indicate how many of the 730 adnexa not con- taining a hydrosalpinx contained an alternative sonographically identified mass, their data do not allow calculation of the positive likelihood ratio of their diagnostic criteria for the discrimination of hydrosalpinx from other cystic adnexal masses. In the second part of their study, Guerriero and colleagues [8] tested the diagnostic power of transvaginal sonography in the differential diagnosis of hydrosalpinx in 239 premenopausal patients with 256 adnexal masses. While reporting a sensitivity of 93.3% and specificity of 99.6%, the authors do not specify the number and types of other masses encountered, thus limiting the analysis of the effectiveness of their diagnostic criteria for distinguishing hydrosalpinx from other adnexal masses. Specifically, it is unclear from the authors data how well their diagnostic criteria distinguished hydrosalpinx from cystic neoplasms and paraovarian cystic masses. Nevertheless, despite this limitation, they concluded that the presence of incomplete septa almost uniformly indicates the diagnosis of a Fallopian tube, since the true septa of ovarian tumors are very seldom, if ever, incomplete. The clinical value of specific sonographic findings associated with hydrosalpinx is best evaluated when characteristic alternative possibilities are excluded from consideration. Previous data have shown that hemorrhagic ovarian cysts, cystic teratomas, and endometriomas often exhibit characteristic sonographic features that allow accurate sonographic diagnosis [10 12]. Thus, if one tests sonographic findings for the diagnosis of hydrosalpinx in a group of patients in whom the main competing diagnoses are hemorrhagic ovarian cyst, cystic teratoma, or endometrioma, it is reasonable to expect that the diagnostic criteria for hydrosalpinx will show excellent results because these alternative pathologic entities have characteristic sonographic findings that are not mistaken for a hydrosalpinx. Furthermore, in current practice, a cystic adnexal mass with sonographically identified solid areas is generally managed as a probable neoplasm; even if there were other coexisting features suggestive of hydrosalpinx, such as tubular shape, incomplete septations, or tiny hyperechoic mural nodules, the suspicious solid areas would mandate surgical evaluation. For these reasons, we designed our study mainly to understand the predictive value of sonographic findings for distinguishing hydrosalpinx from cystadenomas and cystadenofibromas by focusing on cases with surgical proof (thus eliminating self-limiting hemorrhagic ovarian cysts and other nonneoplastic 1036 AJR:186, April 2006

5 Sonography of Adnexal Masses cysts) and excluding cases in which the sonographic or pathologic findings indicated a diagnosis of cystic teratoma, endometrioma, or cystic mass with solid areas. Distinguishing between a hydrosalpinx and a benign cystic ovarian neoplasm when a cystic adnexal mass is encountered sonographically is of obvious benefit. In the asymptomatic patient, a cystic adnexal mass with features confidently predicting hydrosalpinx could be followed and subsequently possibly ignored, whereas masses suspected of having a reasonable chance of being a neoplasm even those with benign features are generally surgically removed. Sonologists who are able to convey a high degree of confidence in the assessment of an adnexal mass as a hydrosalpinx empower the patient and her surgeon to choose the most appropriate management strategy. When trying to distinguish a hydrosalpinx from a benign cystic ovarian neoplasm, it should be recognized that the sonographic identification of a normal-appearing ovary ipsilateral to a cystic adnexal mass should have an extremely positive effect on the ability of the sonologist to accurately predict that the mass is a hydrosalpinx and not a cystadenoma or cystadenofibroma. The power of this sonographic observation in the diagnosis of hydrosalpinx was not studied in this investigation. Of course, paraovarian location is not exclusive to hydrosalpinges and can be found in rare cystadenomas and cystadenofibromas [13]. Our data suggest that the sonologist who encounters a cystic mass without suspected solid areas and without features predictive of a dermoid cyst, endometrioma, or hemorrhagic ovarian cyst can be most confident of the diagnosis of hydrosalpinx when the mass can be characterized as being tubular and either shows the waist sign or contains small round projections. Of interest, we found that observing a short linear projection or incomplete septation in a cystic adnexal mass had no additional value in predicting that a mass was a hydrosalpinx when combined with the observation of tubular shape. Among all five sonographic features tested, the waist sign had the highest likelihood ratio for the prediction of hydrosalpinx. One limitation of our study is that we used a consensus interpretation of two observers to characterize the presence or absence of a given finding. We chose to use a consensus interpretation because of the retrospective nature of our study. The inability to discuss images with the examining sonographer, review recorded cine loops, or clarify confusing images by directly examining the patient with sonography served to handicap the reviewing sonologists in the detection and characterization of subtle findings. This limitation could influence the observed performance characteristics for any of the sonographic findings evaluated. Thus, it is possible that any or all of the tested sonographic features could perform better in clinical practice with the use of realtime imaging, sonographer sonologist discussion, or both. The use of consensus interpretation also results in an inability to test and characterize the interobserver variability of the sonographic features we analyzed. Several factors are important to consider when comparing our study with others. By virtue of our study design, the likelihood ratios of the particular sonographic findings and combinations of findings studied in our investigation are most relevant when applied to distinguish hydrosalpinx from benign cystic ovarian neoplasms (cystadenomas and cystadenofibromas). Statistically, these findings would all show higher likelihood ratios if there were many more true-negatives, as might be expected if the study population consisted mostly of premenopausal women with hemorrhagic cysts, endometriomas, and dermoid cysts. Previous studies have shown these masses to exhibit alternate characteristic sonographic findings in most cases. Differences in patient population between our study and those of preceding investigations are important to consider. Most of our patients were postmenopausal. It is important to recognize that the only prior study attempting to evaluate the diagnostic performance of sonography in discriminating hydrosalpinx from other adnexal masses evaluated only premenopausal women, with presumably relatively few cystic ovarian neoplasms. Although we did not intentionally exclude patients with pelvic inflammatory disease and those with tuboovarian abscess with our study design, we had no patients with either of these conditions who met our inclusion criteria. All of our patients underwent surgery, introducing bias into the study. Undoubtedly, there are patients with a hydrosalpinx that is so characteristic on imaging (perhaps because of identification of an ipsilateral ovary on sonography or MRI or for other reasons) that they do not need surgical evaluation; these patients were not captured by our study design. In addition, patients with hydrosalpinx but without a surgically excised adnexal mass were also not captured by our study design; infertility patients undergoing laparoscopy for suspected endometriosis or other reasons would fall into this gap. These factors might contribute to some disparity in our analysis of sonographic findings previously associated with hydrosalpinx (incomplete septation and short linear projections) as compared with prior studies. To the extent that these previously described findings are identified in nonsurgical cases of hydrosalpinx, such as in cases of mild or moderate pelvic inflammatory disease, they could have more predictive utility than we have noted. To be clear, our analysis does not indicate that incomplete septations and short linear projections are not associated with hydrosalpinx; indeed, in concordance with prior studies, our data show that these sonographic findings have positive predictive value for the diagnosis of hydrosalpinx. Nevertheless, the conclusion of prior investigators that the presence of incomplete septations uniformly indicates hydrosalpinx is erroneous. Incomplete septations do indeed exist in cystic ovarian neoplasms. Our analysis suggests that the most accurate sonographic diagnosis of hydrosalpinx as the cause of a cystic adnexal mass is achieved by determining whether the mass appears tubular and then focusing on whether a waist sign or small round projections can be identified. When these combinations are identified in a cystic adnexal mass that does not have solid-appearing areas or features characteristic of hemorrhagic cysts, endometriomas, or dermoid cysts, sonologists can convey a high level of confidence in the diagnosis of hydrosalpinx. References 1. Tessler FN, Perrella RR, Fleischer AC, Grant EG. Endovaginal sonographic diagnosis of dilated fallopian tubes. AJR 1989; 153: Cacciatore B, Leminen A, Ingman-Friberg S, Ylostalo P, Paavonen J. Transvaginal sonographic findings in ambulatory patients with suspected pelvic inflammatory disease. Obstet Gynecol 1992; 80: Bulas DI, Ahlstrom PA, Sivit CJ, Blask AR, O Donnell RM. Pelvic inflammatory disease in the adolescent: comparison of transabdominal and transvaginal sonographic evaluation. Radiology 1992; 183: Atri M, Tran CN, Bret PM, Aldis AE, Kintzen GM. Accuracy of endovaginal sonography for the detection of fallopian tube blockage. J Ultrasound Med 1994; 13: Taipale P, Tarjanne H, Ylostalo P. Transvaginal sonography in suspected pelvic inflammatory disease. Ultrasound Obstet Gynecol 1995; 6: Timor-Tritsch IE, Lerner JP, Monteagudo A, Mur- AJR:186, April

6 phy KE, Heller DS. Transvaginal sonographic markers of tubal inflammatory disease. Ultrasound Obstet Gynecol 1998; 12: Patten RM, Vincent LM, Wolner-Hanssen P, Thorpe E Jr. Pelvic inflammatory disease: endovaginal sonography with laparoscopic correlation. J Ultrasound Med 1990; 9: Guerriero S, Ajossa S, Lai MP, Mais V, Paoletti AM, Melis GB. Transvaginal ultrasonography associated with colour Doppler energy in the diagnosis of hydrosalpinx. Hum Reprod 2000; 15: Molander P, Sjoberg J, Paavonen J, Cacciatore B. Transvaginal power Doppler findings in laparoscopically proven acute pelvic inflammatory disease. Ultrasound Obstet Gynecol 2001; 17: Patel MD, Feldstein VA, Lipson SD, Chen DC, Filly RA. Cystic teratomas of the ovary: diagnostic value of sonography. AJR 1998; 171: Patel MD, Feldstein VA, Chen DC, Lipson SD, Filly RA. Endometriomas: diagnostic performance of US. Radiology 1999; 210: Patel MD, Feldstein VA, Filly RA. The likelihood ratio of sonographic findings for the diagnosis of hemorrhagic ovarian cysts. J Ultrasound Med 2005; 24: Korbin CD, Brown DL, Welch WR. Paraovarian cystadenomas and cystadenofibromas: sonographic characteristics in 14 cases. Radiology 1998; 208: AJR:186, April 2006