Advances in ultrasonography have led to frequent incidental detection of an early ovarian carcinoma at initial scanning.1,2 Other ovarian lesions in a differential diagnosis include ovarian endometriomas, pelvic inflammatory disease (PID), and ovarian dermoids. Several attempts have been made to distinguish ovarian malignancy from questionable ovarian lesions on the basis of gray scale ultrasound and/or color Doppler features.2–5 It has been reported that the presence of intratumoral papillae, solid parts, and thick septa suggests ovarian malignancy. The detectability of these findings varies and their diagnostic value is not established. Thus, the diagnostic ability of gray scale ultrasound is still limited.
Spectral Doppler sonography and color Doppler ultrasound have been used successfully in the evaluation of adnexal tumor vascularity.4,5 Although previous studies reported that both spectral Doppler and color Doppler ultrasound could provide clinically useful information related to adnexal tumor vascularity, these imaging modalities have inherent limitations, such as lack of sensitivity to slow flow, angle dependency, and aliasing, which occurs when the Doppler shift frequency exceeds one half the pulse repetition frequency.6 Furthermore, a nonuniversal selection of Doppler parameters (resistance index [RI] or pulsatility index [PI]); the choice of highest, lowest, or mean impedance values, and the selection of vessels for investigations, together with operator variance and system sensitivity, contribute to the confusion.
Unlike color Doppler ultrasound, which is based on mean frequency shift, power Doppler ultrasound is based on total integrated power of the Doppler spectrum. Concurrently with the development of various sonographic contrast agents, the potential role of power Doppler sonography has increased.7–9 More recently, technological development enabled the physician to generate three-dimensional vascular images that display tumor morphology and branching pattern of intralesional vessels on-screen interactively.10–12
The purposes of our study were to evaluate the vascular pattern in questionable adnexal lesions using three-dimensional power Doppler ultrasound with and without echo-enhancing contrast, to determine whether the imaging patterns of identified vessels could be correlated with histopathologic results, and to evaluate whether the information obtained with echo-enhanced, three-dimensional power Doppler ultrasound could improve the diagnostic accuracy in adnexal lesions compared with that with nonenhanced three-dimensional power Doppler ultrasound.
Materials and Methods
The study population included a consecutive series of patients who had complex ovarian lesions during the preceding 12 months. Our study group was limited to patients with complex adnexal lesions of uncertain malignancy at transvaginal B-mode and/or color Doppler sonography. We tested whether penetrating or mixed penetrating and peripheral patterns of vessels were associated with malignancy. Exclusion criteria were as follows: typical benign cysts with absence of internal echoes, a sharply defined smooth wall, no wall enhancement, and increased sonic-through transmission. A total of 45 patients with suspicious masses (33 benign adnexal lesions and 12 ovarian malignancies) were analyzed with three-dimensional power Doppler sonography before and after injection of sonographic contrast agent (Levovist, SHU 508A; Schering AG, Berlin, Germany). The study enrollment period was 1 year. Twenty-seven of the women studied were premenopausal (mean [range] age 34 [19–49] years). Three women were perimenopausal (mean [range] age 49 [47–54] years). The remaining 15 women were postmenopausal (mean [range] age 62 [51–77] years). Two postmenopausal patients had undergone hysterectomy, 12 and 7 years before entering the study.
A valid questionnaire was completed by participants with the help of trained nurses, who confirmed and checked answers. The questionnaire contained detailed inquiries about occupation, parity, personal and family history, breast-feeding, former or present use of oral contraceptives (OCs), ovulation drugs or hormonal replacement therapy (HRT), use of talc in feminine hygiene, smoking, diet, and alcohol use.
Most of the patients in our study group were active or retired lower-level administrative employees and/or small business entrepreneurs (n = 21). Twelve patients were skilled or specialized workers, five were unskilled workers, and four were managers or other high-level administrative employees, whereas three patients were unemployed. Two patients with ovarian carcinoma had one first-degree relative with this cancer. Furthermore, ovarian malignancy was detected in one patient with a personal history of endometrial carcinoma and in one patient with breast cancer. In the group of patients with benign adnexal lesions, there were no data or personal or family history of ovarian, endometrial, breast, or colorectal malignancy. We provide other demographic data in tabular form (Table 1).
All the examinations in premenopausal patients were performed during the early proliferative phase of the menstrual cycle. Clinical information, family history, and data on CA 125 were not known to the sonographer. In all cases, histopathologic diagnosis was obtained after surgery (Table 2), within 1 to 2 weeks after three-dimensional power Doppler sonography, and the histopathologist was unaware of the scan results.
The sonographic contrast agent used in this study, Levovist (Schering AG), is a suspension of monosaccharide microparticles (galactose) in sterile water. A stabilized microbubble suspension of the agent was administered intravenously at a concentration of 300 mg/mL. An 8.5-mL dose of contrast agent was injected slowly at a rate of 0.2 mL/second by hand to reduce artifacts and to prolong the duration of enhancement. This was followed by an additional 10 mL of physiologic saline solution to flush the cannula, using the same injection rate.
Three-dimensional power Doppler studies were performed by a single examiner (SK) with Voluson 530 (Kretz Medison, Zipf, Austria). Volume data acquisition was performed using a 5-MHz transvaginal volume probe. The amount of the volume influences the duration of the scanning procedure (acquisition time approximately 30–60 seconds). Fixed preinstalled instrument settings for pulse repetition frequency (1.0), signal power (2), wall motion filter (61), persistence (rise 0.1, fall 0.3), center frequency (middle), gray/color balance (greater than 192), quality (4), and density (8) of the volume scan were used throughout the examinations. Only power Doppler gain was adjusted to optimize signal quality. Volumetric data were stored on a hard disk to enable full evaluation without loss of information at a later point. During each examination, we defined a cube enclosing the vessels of the adnexal lesion but excluding the iliac vessels or motion artifacts, which are very common as a result of the very sensitive power Doppler.
After performing an initial scan by three-dimensional power Doppler, we injected the enhancing contrast agent. Thereafter, we had to reduce the power Doppler gain because power Doppler noise, such as color blooming artifact, was produced by the effect of the contrast agent. We manipulated the power Doppler gain at the highest level possible that did not produce considerable artifacts, continuously throughout the entire scanning. The resultant power Doppler gains ranged from 70 to 85%. Volumetric data of power Doppler images were stored on the hard disk at 60, 120, 180, 240, 300, 360, 420, and 480 seconds after the initiation of the contrast agent injection.
At three-dimensional power Doppler ultrasound, the vascular distribution in adnexal lesions was classified as follows: pattern 0, no signal pattern (which indicated no detectable vessels); pattern 1, peripheral pattern, which indicated blood vessels arose outside the lesion and surrounded the lesion; pattern 2, penetrating pattern, indicating that blood vessels arose outside the lesion and coursed towards the center; and pattern 3, mixed penetrating and peripheral pattern. Three-dimensional power Doppler findings after contrast injection were compared with those before beginning the procedure in terms of sensitivity, specificity, and positive and negative predictive values. Diagnostic efficiency was expressed as the sum of true positives and true negatives divided by the sum of true positives, true negatives, false positives, and false negatives. We tested whether penetrating or mixed penetrating and peripheral patterns of vessels were associated with adnexal malignancy. Schematic presentation of vascular distribution is demonstrated in Figure 1.
Vascular features were analyzed separately for each interval following the injection of the contrast medium, and the most representative three-dimensional image of each patient was chosen for further interpretation.
The same surgical team operated on all the cases, and histopathologic diagnosis was considered final. Malignant tumors were classified according to the International Federation of Gynecology and Obstetrics (FIGO) system.13 The study protocol was approved by the hospital's Ethical Committee, and all patients consented to participate in the study.
Forty-five patients were referred for three-dimensional power Doppler ultrasound because of indeterminate findings of the complex adnexal mass on transvaginal B-mode and/or color Doppler ultrasound. At laparotomy, 12 women were found to have ovarian cancer, whereas 33 had benign adnexal lesions. Histopathologic examination of the ovarian malignancies revealed eight cases of serous and two cases of mucinous cystadeno-carcinoma, one endometrioid cystadenocarcinoma, and one clear cell carcinoma. Adequate surgical staging was defined according to FIGO guidelines.13 There were five stage I cases (three stage Ia, one stage Ib, and one stage Ic), four stage II cases, and two stage III cases, and one patient had stage IV ovarian carcinoma. Table 2 lists histopathologic diagnosis of the adnexal lesions analyzed by enhanced and nonenhanced three-dimensional power Doppler sonography.
The best image quality was obtained in 39 patients 180 seconds after initiation of the contrast injection; after 120 seconds in four patients; and after 240 seconds in two patients.
Of the 12 ovarian cancers, four (33.3%) showed penetrating vessels, three (25.0%) had mixed penetrating and peripheral pattern, and three (25.0%) showed peripheral vessels; in two cases (16.7%), no flow was detectable by nonenhanced three-dimensional power Doppler imaging. The mean (range) diameter of malignant lesions in which penetrating vessels were found at three-dimensional power Doppler ultrasound was 5.2 (2.2–9.5) cm, of lesions with mixed penetrating and peripheral pattern was 5.6 (2.5–9.2) cm, of lesions with peripheral vessels was 5.4 (3–8) cm, and of lesions with no detectable vessels was 9 and 12 cm in diameter. In the group of benign lesions, no detectable flow was found in 12 (36.4%) patients, peripheral vessels were seen in 20 (60.6%), and penetrating vessels were seen in one (3.0%) case. The benign lesion with penetrating vessels was subsequently found to be cystadenofibroma.
By using the presence of penetrating vessels as the diagnostic criterion for malignancy, three-dimensional power Doppler ultrasound demonstrated a diagnostic sensitivity of 58.3% and specificity of 97.0%. The positive and negative predictive values were 87.5% and 86.5%, respectively (Table 3).
Contrast-enhanced, three-dimensional power Doppler sonography showed penetrating pattern in ten (83.3%) and mixed penetrating and peripheral pattern in two (16.7%) patients with ovarian malignancy. Two malignant tumors with a lack of any signal on noncontrast three-dimensional power Doppler sonography showed markedly increased power Doppler signals after the injection of a contrast agent. Peripheral and penetrating vessels with irregular course were clearly obtained in these patients. In the group of benign lesions, contrast-enhanced, three-dimensional power Doppler sonography demonstrated peripheral distribution of the vessels in 29 (87.9%) cases, whereas two (6.1%) lesions (ovarian dermoid and chronic pelvic inflammatory disease [PID]) remained avascularized. In two patients, one with a fibroma and one with a cystadenofibroma, contrast-enhanced, three-dimensional power Doppler sonography led to a misdiagnosis of malignant lesions because of the frank enhancement of the vessels of the solid components, which were interpreted as penetrating vessels.
With respect to differential diagnosis between malignant and benign ovarian lesions, contrast-enhanced, three-dimensional power Doppler sonography reached diagnostic sensitivity and specificity of 100% and 93.9%, respectively. The positive and negative predictive values of this method were 85.7% and 100%, respectively. Therefore, the diagnostic efficiency was improved with the use of sonographic contrast agent from 86.7% to 95.6% (Table 3).
In complex adnexal lesions, differentiating benign from malignant lesions can be problematic even though modern ultrasound studies have improved the diagnostic accuracy of these lesions.2,4,10 In the present study, contrast-enhanced, three-dimensional power Doppler imaging was used for evaluation of adnexal tumor vascularity without considering morphologic findings that suggest malignancy. We found that contrast-enhanced, three-dimensional power Doppler imaging improves visualization of intratumoral vascularity, which may aid in detection of malignant adnexal tumors.
Although our study population included only 45 cases, the use of a contrast agent in three-dimensional power Doppler ultrasound appears to improve sensitivity for differentiating benign from malignant adnexal lesions by allowing better detection of malignant tumor perfusion than is obtained by imaging without contrast. However, our results have a fundamental bias because the differential diagnosis between benign and malignant ovarian lesions was based only on the presence or absence of enhancement within the lesion. Although an optimized protocol for contrast-enhanced, three-dimensional power Doppler ultrasound has not been established yet, the selection of an optimal infusion rate and concentration of sonographic contrast agent must be very important in obtaining reliable results. Slower injection rates (approximately 0.2 mL/second) and use of sonographic contrast agent with a 300 mg/dL concentration can reduce the emergence of “blooming artifacts” in early stages of the investigation.7 It is expected that these artifacts and high sensitivity to motion may be reduced by optimization of the instrument's settings.14 In the clinical setting, the enhancement provided by the contrast agents and resultant color blooming could be compensated for by reducing the scanner's color flow sensitivity soon after injection and gradually increasing the sensitivity as the effect of contrast diminishes over time.
We found that contrast-enhanced, three-dimensional power Doppler sonography provided better visualization of tumor vascularity in suspicious adnexal lesions than that obtained with non–contrast-enhanced, three-dimensional power Doppler sonography, and this led to a more exact differential diagnosis. In the present study, contrast-enhanced, three-dimensional power Doppler ultrasound showed 100% negative predictive value for malignant ovarian lesions and 85.7% positive predictive value, which was similar to non–contrast-enhanced power Doppler ultrasound. Furthermore, our results show that the pattern of irregularly branching penetrating vessels in suspicious adnexal lesions demonstrated on three-dimensional power Doppler ultrasound with or without contrast enhancement is an important feature that should be considered with other sonographic criteria to predict the likelihood of malignancy.
In deep-lying and necrotic adnexal lesions confirmed by histopathology, non–contrast-enhanced, three-dimensional power Doppler sonography did not demonstrate intratumoral vessels as a result of low velocity flow. In two cases of ovarian malignancy measuring 9 and 12 cm, initial scan by three-dimensional power Doppler did not reveal intratumoral vascularity. Contrast medium administration increased the strength of the returning signal, generating a clear image of central stellate feeding vessels from which a diagnosis can be made (Figure 2). Therefore, the combination of echo-enhancing contrast with three-dimensional power Doppler technique brought us a step closer to angiographic images.
Detectable vessels are seen commonly in benign lesions as well, and these can also be prominent. However, the morphology and distribution pattern of vessels in benign lesions are usually distinguishable from those in malignant tumors.
Contrast-enhanced, three-dimensional power Doppler ultrasound was especially useful in patients with ovarian dermoids (n = 9), chronic PID (n = 9), and organizing hematoma with hemorrhagic cyst (n = 1). In these cases, contrast-enhanced, three-dimensional power Doppler ultrasound examination did not reveal signs of vascularity within the solid parts, but visualized peripheral vessels, which was mandatory to avoid false-positive results for malignancy (Figure 3). In seven patients with cystadenoma and five endometriomas, only discrete peripheral vascularization was detected, whereas two lesions (chronic PID and hemorrhagic cyst) remained avascularized. However, in two patients with benign lesions (one fibroma and one cystadenofibroma), contrast-enhanced, three-dimensional power Doppler sonography demonstrated penetrating vessels within the solid component, which led to misdiagnosis of ovarian malignancy. Because the treatment of benign lesions with suspicious morphology consists of surgical resection, these false-positive results did not affect the patients' management protocols.
The accuracy of contrast-enhanced, three-dimensional power Doppler findings for the differentiation of adnexal tumors may be increased if the findings are analyzed together with morphologic parameters obtained by three-dimensional ultrasound. It is possible with three-dimensional power Doppler sonography to visualize vessel continuity more completely (in three orthogonal projections) and to demonstrate vessel branching (three-dimensional vascular reconstruction) more clearly. Furthermore, sonographic findings that suggest malignancy such as irregular and thick cystic walls, solid component, papillary protrusions, thick septa, and nonhomogeneous echogenicity can be analyzed precisely with three-dimensional ultrasound.15 Therefore, we expect combined use of three-dimensional ultrasound and enhanced three-dimensional power Doppler modality to be very useful in differential diagnosis of questionable adnexal masses, particularly in discriminating malignant from benign lesions.
To determine the practical value of this modality, comparative studies of the diagnostic performance of three-dimensional power Doppler ultrasound before and after injection of contrast agents, including evaluation of cost-effectiveness, reproducibility, examination time, and invasiveness, must be performed.