ndometrial cancer is one of the most common gynecological malignancies. It develops in about 142,000 women worldwide, and lead to approximately 42,000 of mortality [1]. Transvaginal ultrasound followed by endometrial biopsy is the most cost-effective diagnostic approach in the population with post-menopausal bleeding [2]. We therefore consider TVU as the first step in any woman presenting with postmenopausal bleeding [3]. Ultrasonography is a non-invasive method that could assess the morphologic structures of endometrium [4,5]. Sonographically determined endometrial thickness measurement shows high diagnostic performance for detection of endometrial cancer in symptomatic postmenopausal women [6].In addition, there is no universally accepted sonomorphologic criteria to define benign or malignant structure on the endometrium. In order to make the prediction accuracy better, some studies created a scoring system involving clinical and ultrasound parameters [7,8].This article was aimed to appraise studies that assess the accuracy of endometrial malignancy prediction system or index which involving ultrasound as one of the predictors.

The search was conducted on the Cochrane Library®, PubMed® and EMBASE® with the keywords of "endometrial" AND "malignancy" AND "scoring" OR "prediction" OR "index" on each databases with certain techniques (figure 1). Search focused on articles in diagnostic type showing diagnostic values of the studies. Reference lists of relevant articles were searched for other possibly relevant studies. After obtaining a result, a first selection was done by screening the study titles and abstracts. Eight articles were available as full text, and all of them included in our analysis.

Appraisal of 8 diagnostic studies involving 5543 patients underwent clinical and ultrasound for predicting endometrial malignancy confirmed with the histopathology result was conducted finding of the diagnostic values (Se, Sp, PPV, NPV). Review study or study without diagnostic values reported were excluded. We used diagnostic appraisal questions developed by Centre of Evidence-Based Medicine (CEBM), University of Oxford (available at: http://www.cebm.net/criticalappraisal/).

Eight eligible studies were appraised to assess the accuracy of ultrasound and/or ultrasound index to predict malignancy. The incidence of endometrial malignancy confirmed by histopathology examination was ranging from 10.5 to 58% from 8 studies. The accuracy of ultrasound-based index to predict endometrial malignancy was ranging from 75% -98% from eight studies. Opolskiene, et al conducted a consecutive study of 729 post-menopausal bleeding, to evaluate the diagnostic performance of models predicting endometrial cancer. They stated that the accuracy was increased significantly when endometrial thickness and power Doppler assessment are added to clinical variables. Clinical model including the variables age, use of warfarin and use of hormone replacement therapy had the largest area under the receiveroperating characteristics curve (AUC), with a value of 0.74 (95% confidence interval (CI), 0.67-0.81). A model including age, use of warfarin and endometrial thickness had an AUC of 0.82 (95% CI, 0.76-0.87), and one including age, use of hormone replacement therapy, endometrial thickness and vascularity index had an AUC of 0.91 (95% CI, 0.87 -0.95) [9].

Dueholm, et al concluded that simple Doppler score (which considered only presence of vascularity and not presence of single/double dominant vessel, multiple vessels, large vessels, color splash or densely packed vessels) had an AUC of 0.83 in the prediction of endometrial cancer. Prediction index including endometrial thickness, Doppler score and interrupted endomyometrial junction on unenhanced TVS predicted endometrial cancer with an AUC of 0.95 (95% CI, 0.92 -0.99) and, with addition of irregular surface on GIS, the AUC was 0.97 (95% CI, 0.94 -0.99) [10].

Burbos, et al created a model to predict endometrial carcinoma in postmenopausal women called DEFAB (Diabetes, Endometrial thickness, Frequency of bleeding, Age, and BMI). In the DEFAB criteria, presence of diabetes in a patient scores 2; endometrial thickness ? 14mm scores 1, recurrent episodes of bleeding scores 4; age ?64 years scores 1; and BMI ?31 kg ?? 2 scores 1. The value ?3 has a positive predictive value (PPV) of 7.78% and negative predictive value (NPV) of 98.2%, whereas a score equal to or greater than 5 has a PPV of 11.9% and NPV of 97.8% [12].

Seek in, et al investigated the accuracy of endometrial thickness in predicting endometrial pathologies in both of symptomatic (group 1) and asymptomatic (group 2) postmenopausal women. The best cut-off point for endometrial thickness in predicting endometrial carcinoma in group 1 was 8.2 mm, which provided 75% sensitivity and 74% specificity; area under the AUC of 0.88; 95% CI, 0.76-1.00%. In group 2, the AUC was 0.76 (95% CI, 0.46-1.00; p 5 0.114). 6 . In other study, Patel, et al stated that threshold of 4 mm, the sensitivity is 90.6% and increases to 96.9% when decreasing the threshold to 3 mm [13].

Mansour, et al evaluated the role of endometrial/uterine corporeal volume ratio (EV/UCV) assessment in the prediction of endometrial cancer. EV/UCV of a cutoff value 0.017 was predictive of malignancy. Endometrial/uterine volume ratio was more sensitive than endometrial volume and endometrial thickness for prediction of endometrial cancer [7].

Mihajovic, created the transvaginal ultrasound score for endometrial malignancy prediction consisted of: thickness of endometrium (up to five mm = 0, from five to eight mm = 1,> eight mm = 2), echogenicity of the endometrium compared to the myometrium: normal echogenicity = 0, hyperechogenous = 1, hypoechogenous = 2 , the border of the endometrium towards the myometrium -subendometrial hypoechogenous zone (whole = 0. intermittent = 1), homogeneity of the texture of the endometrium (homogenous = 1. inhomogeneous = 2), presence of the colored signals in the endometrium (present = 2. absent = 1), index of resistance in newly-formed blood vessels in the endometrium (> 0.4 = 1. < 0.40 = 2), volume of the endometrium by an ultrasound check-up (< 13 ml = 1. > 13 ml = 2). Score system showed that the value 8 had the best validity for the detection of endometrial malignity, with the sensitivity of 0.857 and specificity of 0.785 [4].

In our study, the incidence of endometrial malignancy was varied among studies. It could possibly explain by the variation of the population. In some studies, they included women with a complaint of postmenopausal bleeding who has endometrial thickness ? 4.5 mm 9 , while other studies included subjects without considering the ET. 12,14 We found the incidence of endometrial malignancy from 5 to 58%. It was similar with the finding from The Gynecologic Oncology Group (GOG) that found 42.6% of endometrial malignancy, 123 of 289 specimens [14].

Sorosky in their review stated that the positive predictive value and negative predictive value of an

In further study [11] they compare the offline and real time evaluation during scanning to assess efficiency of two-dimensional (2D) and threedimensional (3D) TVU, power Doppler angiography (PDA) and gel infusion sonography (GIS) to detect endometrial malignancy. Diagnostic efficiency of 3D analysis may be improved by use of risk of endometrial cancer (REC)-scoring systems, without the need for calculation of vascular or endometrial volume. The REC consisted of: (1) body mass index ? 30 (+1 point), (2) total endometrial thickness ? 10 mm (+1 point), (3) total endometrial thickness ? 15 mm (+1 point), (4) interrupted endomyometrial junction (+1 point) and ( 5) irregular surface at gel instillation sonography (GIS) (+1 point). The first model included BMI, endometrial thickness, presence of an interrupted endomyometrial junction and Doppler score, had AUC of 0.879. Evaluation of 3D-GIS with BMI, an interrupted endomyometrial junction, Doppler score and irregular endometrial surface at 3D-GIS, had the highest diagnostic efficiency on multivariate regression, with an AUC of 0.908. Application of the REC-score system at 3D-PDA or 3D-GIS had comparable efficiency compared with their respective 2D models [11].

office biopsy are greater than 90% [14].TVS screening for endometrial cancer has good sensitivity in postmenopausal women [15]. In addition, in certain conditions in which the cervical canal could not be accessed by curettage, the role of ultrasound will be useful to predict malignancy.

Monsour had the highest appraisal score, because they show all the diagnostic parameters of their result. Transvaginal 3D render mode ultrasound was used to assess the volume of the uterus in the coronal plane using manual lining technique. Volumes were manually calculated in the coronal plane with 30° rotation steps. They found that EV/UCV had the best in prediction of malignancy compared to endometrial thickness and endometrial volume; AUC (area under the curve) for endometrial thickness, volume and EV/UCV was respectively 75, 92 and 100%. However, further studies should be conducted with a larger number of subjects to support these findings. 7 The interobserver and intraobserver reproducibility of 3D ultrasound for assessment of endometrial volume measurements in patients with postmenopausal bleeding was well proved, showing better reproducibility than 2D measurement of endometrial thickness [7].

Using ultrasound parameter, the accuracy of prediction index was higher compared to the nonultrasound based index. In our study, the accuracy of prediction index involving ultrasound parameters was ranging from 0,75 to 0.98. It was higher compared to the clinical-based prediction index. Burbos, et al created a clinical predictive model called FAD 31 (F for the frequency of bleeding episodes, A for the age of the patient, D for diabetes, and number 31 represents the BMI cut-off value). The AUC was 0.73. Among 14 recognized indexes in our appraisal study, only 3 indexes had the AUC below 0.8 [8].

Ultrasound-based index to predict endometrial malignancy had good accuracy. Addition of endometrial thickness and power Doppler to clinical parameters had increased the prediction accuracy. EV/UCV had the best in prediction of malignancy compared to endometrial thickness and endometrial volume. Further larger study should be conducted to assess the effectivity and eligibility of several ultrasound parameters.