- Top of page
The greater omentum, the fatty peritoneal sheet extending inferiorly from the greater curvature of the stomach into the lower abdomen, represents a privileged metastatic site for ovarian as well as uterine and gastrointestinal cancers. In the presence of metastatic omental involvement, the risk of diffuse peritoneal disease is very high1, 2. This is due to the intrinsic characteristics of the process of peritoneal seeding, which, when occurring along the routes of peritoneal fluid circulation, leads in the vast majority of cases to the metastatic colonization of the whole abdominal cavity.
Especially in ovarian cancer patients, preoperative knowledge of the extent of the disease is of crucial importance in planning the most appropriate management, given the established favorable prognostic role of optimal cytoreduction at primary surgery, and this requires the involvement of a specific gynecological oncology team3–5. Of the available radiographic studies, computed tomography (CT) and magnetic resonance imaging (MRI) have been evaluated as means of staging ovarian cancer that provide acceptable accuracy in the preoperative evaluation of omental infiltration6–10.
However, there is scanty data in the literature on the utility and accuracy of transabdominal and transvaginal sonography in the detection of metastatic omental involvement1, 11. In particular, the study by Conte et al.11, performed in a relatively small series of 50 ovarian carcinomas, reported that ultrasound examination was able to detect metastatic omental involvement with an overall accuracy of 83.7%.
Given recent advances in ultrasound technology and equipment, the performance of sonography is likely to have improved dramatically over the years. The aim of this study was to analyze prospectively the ability of ultrasound to detect metastatic omental involvement in a large series of patients with suspicious pelvic masses and to describe the sonographic features of metastatic omental disease.
- Top of page
Between December 2001 and December 2004, 184 patients scheduled for surgery because of suspicious pelvic masses were evaluated preoperatively by means of transabdominal and transvaginal ultrasound examination. One hundred and sixty-two patients were enrolled at the Gynecologic Oncology Unit, University of Sacred Heart, Rome, and 22 at the Department of Obstetrics and Gynecology, University of Bologna. All patients underwent exploratory laparotomy and the type of surgery was chosen according to the intraoperative histopathological findings. In ovarian cancer patients, surgical staging included peritoneal washing, total abdominal hysterectomy, bilateral salpingo-oophorectomy, appendectomy, omentectomy and multiple biopsies. Intestinal resection, pelvic and/or subdiaphragmatic peritonectomy and pelvic/aortic lymphadenectomy were performed12 if required in order to achieve optimal cytoreduction, defined as the absence of any visible residual tumor.
Transabdominal sonographic examination was performed using commercially available equipment (ESAOTE AU5, Genova, Italy until October 2002, and ESAOTE Technos, Genova, Italy, from November 2002 to December 2004) with color and power Doppler capabilities. The abdomen was evaluated first by transabdominal examination with a 3.5–5.0-MHz convex transducer, and then by a transvaginal scan of the pelvic organs with a 6.0–4.0-MHz broadband transducer. All examinations included investigation of the area located above the bowel loops and below the anterior peritoneal surface. This was performed best by reducing the field of view to 3–4 cm in depth, and starting with a transverse section of the supra-umbilical region of the abdomen. From this position, the transducer was moved towards the caudal region while keeping the axial plane, covering the whole abdominal surface. The transducer was then rotated to give a sagittal section of the abdomen, and was subsequently moved from one of the patient's sides to the other. The detection of an intra-abdominal echogenic tissue layer or nodules was defined as an ultrasound-positive examination. Figure 1 shows an ultrasound image of the region of interest, i.e. the upper left part of the abdomen, in a normal patient.
Figure 1. Sonographic image showing the region of interest in a normal case, with no detectable intra-abdominal echogenic tissue layer or nodules.
Download figure to PowerPoint
When a solid mass located beneath the parietal peritoneum was visualized at sonography, we analyzed the shape, maximum thickness (in mm), echogenicity (with respect to the bowel loops) and vascularization (presence/absence). When a solid mass located beneath the parietal peritoneum was visualized at sonography, an assessment of its vascularization was achieved using color or power Doppler, with a pulse-repetition frequency of 750 Hz and a wall filter of 50 Hz to detect very low-velocity blood flow.
Ultrasound digital images were saved and stored on a hard disk for subsequent review and analysis. Histological examination of the primary tumors and omentum was carried out in all cases, and preoperative sonographic findings were compared with pathological results.
Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV) and overall accuracy with respect to omental infiltration were calculated for the sonographic findings.
- Top of page
Satisfactory sonographic visualization of the anatomical region of interest was obtained in 173 of 184 (94.0%) cases; in 11 patients (6.0%) the analysis was hampered by obesity or gas-distended intestine. The flow chart of our patient population is presented in Figure 2. Histopathological examination documented benign conditions in 21/173 (12.1%) cases and borderline ovarian tumors in 9/173 (5.2%) cases. One hundred and forty-three patients (82.6%) had malignant tumors, and these were represented mainly by epithelial ovarian adenocarcinomas (136; 95.1%).
The median age of the 145 patients with borderline or malignant ovarian cancer was 58 (range, 19–89) years. Eighty of them had ascites and 65 did not. The median CA 125 level was 256 (range, 5–6000) IU/mL. Histotypes included serous borderline (n = 4), mucinous borderline (n = 5), serous (n = 85), mucinous (n = 7), endometrioid (n = 17), clear cells (n = 7), undifferentiated (n = 11) and there were nine others. Thirty-one patients had FIGO stage I carcinomas13; five were stage II, 95 were stage III and 14 were stage IV. Fifteen patients were graded as G1, 22 were G2, 97 were G3, and in 11 cases the grading was not available.
Metastatic omentum was detected sonographically in 104/173 (60.1%) patients, appearing either as a solid aperistaltic strand of tissue with a free caudal margin, located between the parietal peritoneum and the bowel loops (in 84 (80.8%) cases; Figure 3; median thickness, 19 (range, 9–36) mm), or as solid discrete nodules (in 20 (19.2%) cases; Figure 4; median diameter of the largest nodule, 23 (range, 10–30) mm). There was no difference in the distribution of sonographic pattern (thickened omentum vs. omental nodules) in ovarian vs. non-ovarian malignancies, and according to the different histotypes within the group of ovarian tumors (data not shown).
Figure 4. Sonographic image of nodular omental metastases. The metastatic omentum appears as isoechoic thickened tissue with hyperechoic discrete nodules (arrows).
Download figure to PowerPoint
When considering the echostructure of the surrounding bowel loops, this tissue appeared hypoechoic in 46 (44.2%) cases and isoechoic or slightly hyperechoic in 58 (55.8%) cases. We did not find any difference in the echostructure between ovarian and non-ovarian tumors, nor did we find any among different histotypes in the ovarian malignancies (data not shown). Moreover, the frequency of ultrasound-positive omentum was significantly higher in the presence (90.0%) than it was in the absence (44.6%) of ascites (P = 0.001) (Table 1).
Table 1. Correlations between ascites and FIGO stage vs. ultrasound-positive omentum, vascular-positive omentum and histology-positive omentum
| ||n||Ultrasound-positive (n (%))||P||Vascular-positive (n (%))||P||Histology-positive (n (%))||P|
|All ovarian cancers||136||97 (71.3)|| ||54 (55.6)|| ||95 (69.8)|| |
| Yes||80||72 (90.0)||0.001||38 (52.7)||NS||69 (86.2)||0.0001|
| No||56||25 (44.6)||16 (64)||26 (46.4)|
| I/II||27||3 (11.1)||0.0001||1 (33.3)||NS||0 (0)||0.0001|
| III/IV||109||94 (86.2)||53 (56.3)||95 (87.1)|
Qualitative color and power Doppler evaluation of the thickened omentum was technically successful in only 94 of 104 (90.4%) ultrasound-positive cases, as it was hampered by artifacts in the remaining 10 cases. Intralesional blood vessels were detected in 56 (59.6%) cases.
Table 2 gives the performance of ultrasound examination in detecting metastatic omental involvement. In the overall series, the NPV of ultrasound examination was 92.7%, failing to identify five cases of pathologically documented omental infiltration: false negatives were represented by one case of omental nodules (1 cm in diameter) from ovarian cancer, one case of micronodular (< 5 mm diameter) diffuse peritoneal disease, and three cases with micrometastases from ovarian carcinomas. The PPV of ultrasound examination was 91.3%: false-positives corresponded to two cases of abdominal tuberculosis, two cases with peritoneal but not omental involvement from ovarian cancer, four cases from ovarian (n = 1), endometrial (n = 2) and pancreatic (n = 1) carcinomas, and one case of ovarian carcinoma with neither macroscopic nor microscopic omental metastases at pathological examination. The overall accuracy of ultrasound examination in detecting metastatic omental involvement was 91.9%. When considering only the group of ovarian tumors, the NPV, PPV, and accuracy were 91.9%, 94.6%, and 93.8%, respectively.
Table 2. Performance of ultrasound examination in the detection of metastatic omentum in the 173 patients analyzed
|Ultrasound examination||Total (n)||Metastatic omentum at pathology (n)|
There was no change in ultrasound performance as the study progressed or between the two sonographers over time (data not shown).
We also investigated whether the addition of data resulting from the qualitative color power Doppler examination of ultrasound-positive cases could help to improve the detection of pathologically involved omentum. As shown in Table 3, of 56 ultrasound-positive and color power Doppler-positive cases, 49 had pathologically confirmed metastatic omental infiltration (PPV = 87.5%), while of 38 ultrasound-positive but color power Doppler-negative cases, only two were shown to be negative at pathological examination of the omentum (NPV = 5.3%).
Table 3. Performance of color Doppler examination in the detection of metastatic omentum in 94 ultrasound-positive cases
|Color power Doppler examination||Total (n)||Metastatic omentum at pathology (n)|
Data on the surgical outcome of ovarian cancer patients were available in 114 cases (Table 4). The NPV (i.e. the ability of ultrasound-negative omentum to predict optimal cytoreduction) was 93.3%, while the PPV (i.e. the ability of ultrasound-positive omentum to predict suboptimal cytoreduction) was 81.1%. Overall, the accuracy rate of sonographically defined omental status in predicting surgical outcome was 86%.
Table 4. Surgical outcome estimation according to ultrasound-defined status of the omentum in 114 cases of ovarian cancer
|Ultrasound examination of the omentum||Total (n)||Surgical outcome (n)|
|Optimal cytoreduction||Suboptimal cytoreduction|
- Top of page
This study, conducted in a large prospective series of women with suspicious pelvic masses, has demonstrated that ultrasound examination is highly accurate in detecting metastatic omental involvement.
Specifically, we obtained an NPV of 92.7% and a sensitivity of 95.0%. These values are quite high in comparison with previously published data11, especially taking into account the fact that we did not exclude from analysis microscopic and micronodular metastatic involvement, which represented the vast majority (4/5) of false negatives in our series. Considering that microscopic/micronodular disease is obviously below the resolution limits of the ultrasound procedure, our observations indicate that ultrasound performance in terms of sensitivity and NPV can theoretically approach 100%. These encouraging results can be explained partly by advances in the development of more sophisticated equipment, which likely improves not only the definition of specific sonographic features, but also the identification and characterization of novel sonographic findings. For instance, in the past, metastatic omentum has been described as hypoechoic thickened tissue (‘omental cake’)11, while it is now acknowledged that extensive infiltration by the tumor results in marked thickening that is usually echogenic1, as documented in our series.
On the other hand, the values of PPV and specificity in our study seem to be less impressive than those reported by Conte et al.11; however, our study was designed to evaluate prospectively suspicious pelvic masses, which led ultimately to the enrollment of women with benign conditions, borderline ovarian tumors and non-gynecological tumors metastasizing to the omentum. Indeed, when analyzing ultrasound performance in the group of ovarian tumors, we observed a slightly higher PPV (91.3% in the overall series and 94.6% in ovarian tumors), which suggests that non-ovarian tumors might more frequently contribute to the false-positive than they do to the false-negative rate, although the results must be taken with caution given the small size of the non-ovarian malignancy series.
While the role of color power Doppler in improving the diagnostic discrimination of pelvic masses is well recognized14, 15, our results seem to suggest that it does not provide any significant improvement in the accuracy of ultrasound detection of metastatic omentum, probably because of the high prevalence of negative color power Doppler results, as described for the first time in our series. It is difficult to reconcile our color power Doppler findings with the fact that the omentum is usually recognized as a very vascularized structure. However, it must be taken into account that pathological findings might not necessarily correspond to color power Doppler results, considering the substantial difference between a functional assessment of blood flow and a morphological description of blood vessels; indeed, it cannot be excluded that the presence of diffuse miliary omental carcinomatosis could lead to collapse/obstruction of smaller blood/lymphatic vessels (which is considered the mechanical basis of ascitic fluid formation), thus making them ‘undetectable’ at color power Doppler examination. Finally, the potential artifacts related to bowel peristalsis, breathing movements, presence of ascites, and patient weight must be considered.
We have shown that ultrasound examination is highly accurate in detecting metastatic omental involvement in cases of suspicious pelvic masses. This observation should lead to reconsideration of the role of ultrasound in the preoperative work-up of ovarian cancer, given the relatively low cost of ultrasound, its widespread availability, and the rapidity of the procedure. Moreover, ultrasound examination allows ‘dynamic’ exploration of the abdomen by exerting pressure with the probe over the abdomen; this can help in recognizing the thickened, rigid, aperistaltic omental tissue layer sliding over the surrounding bowel loops. In this context, the recognition of the existence of two different ultrasound patterns of sonographic appearance of metastatic omentum might provide additional information for detecting omental involvement, in order to correctly triage the patient to a referral center.
Although this was not the primary objective of the study, we also investigated the association between the ultrasound definition of omental status and the chance of achieving complete removal of the tumor in cases of ovarian cancer. We are the first to report that ultrasound examination of the omentum can play a role in the prediction of surgical outcome in patients with ovarian carcinoma. In particular, the NPV and PPV of ultrasound definition of omentum were 93.3% and 81.1%, respectively, which are good compared with results obtained with more sophisticated techniques6, 16–21.
Clearly, the association between the ultrasound-defined status of the omentum and the probability of achieving optimal cytoreduction is likely to be just the epiphenomenon of the strict association between omental infiltration and other signs of diffused abdominal disease such as mesenteric infiltration and peritoneal carcinomatosis, which are recognized as the most important causes of tumor unresectability5, thus indicating the need to analyze by multivariate analysis the influence of several sonographic features of ovarian carcinoma extension on surgical outcome.
In this context, a prospective study, as well as a comparison with CT scanning, has been planned at our institutions, aimed at analyzing the role of several sonographically assessed parameters of tumor extension in ovarian cancer cytoreduction.