Ovarian malignancy is the second leading cause of death from gynecological malignancy and the fourth leading cause of deaths from neoplasms in women 1. Despite the considerable improvement in surgical management and chemotherapeutic agents and regimens over the last few decades, the survival rates of ovarian cancer remain poor 2. The main reason for such a poor prognosis is, undoubtedly, late diagnosis of the disease. It has been estimated that almost 75% of women with ovarian cancer were diagnosed at advanced stages (III/IV) 3. The absence of specific symptoms of the early disease and poor performance of the proposed screening tools are the main underlying causes of such delayed diagnosis 3. It is highly valuable to preoperatively predict whether the lesion, in women presenting with an ovarian mass, is benign or malignant. This preoperative differentiation is of paramount importance for a number of reasons, including the decision to conserve or to operate upon, selecting the center (whether a primary or a tertiary center), the surgeon (whether a general gynecologist or a gyne-oncologist), the route of intervention (whether laparoscopic or open), the extent of the procedure performed (whether conservative or radical), and finally counseling the patient about future prognosis 4. Numerous sonographic and biochemical markers have been proposed for the prediction of the nature of an adnexal mass over the last three decades including serum CA125, transvaginal scan with or without Doppler ultrasound, and even MRI. No single tool was shown to be reliably accurate in differentiating benign from malignant lesions. In 1990, Jacobs et al.5 proposed an ‘index’ that incorporated the serum CA125 level and the sonographic features of the ovarian mass, in addition to the menopausal state of the patient in what is called the risk of malignancy index (RMI). In 1991, Sassone et al.6 proposed another scoring system that focused on the ultrasonographic characteristics of the mass, that is, the wall structure and thickness, presence of septae, and the echogenic pattern of the mass. In 1999, Twickler et al. 7 proposed the so-called ovarian tumor index (OTI), which incorporated the patient’s age, with ovarian volume, sonographic features as well as pulsatility index (PI) of the mass vasculature and the location of vessels on the mass. Recently, Rossi et al.3 proposed the pelvic mass score (PMS), which incorporated the Sassone score (as a representative of the sonographic impression on the mass), serum CA125, menopausal status, as well as the vessel location and the resistance index (RI) of the mass vasculature into one scoring system. The PMS was shown to have the best predictive performance over all previous scoring systems 3. The aim of the current study was to evaluate the value of the PMS in the prediction of the nature of ovarian masses in women presenting to a large tertiary gynecological center, namely, the Ain Shams University Maternity Hospital.
Patients and methods
The current study was carried out at the Ain Shams University Maternity Hospital during the period between January and December 2012. The study included women presenting to the outpatient clinic with a diagnosis of an ovarian mass and scheduled for surgical management. The study had been approved by the Ethical Research Committee, Obstetrics and Gynecology Department, Ain Shams University. An informed consent was signed by every patient before participating in the study and after a thorough explanation of the purpose and procedures of the study was provided. Abdominal and transvaginal ultrasound scans were performed for all the women included. The affected ovary dimensions were measured in two perpendicular planes (coronal and sagittal). The ovarian volume was calculated according to the approximate formula for an ellipsoid (volume=length×width×height×π/6) 8. In addition, detailed characteristics of the ovarian mass were noted, including echogenicity, consistency, locularity, laterality, presence of solid areas, and presence of ascites. Pulsed-wave Doppler ultrasound velocimetry of the ipsilateral ovarian artery was performed. At least three well-formed Doppler waves were retrieved for measurement. The PI was calculated according to the formula PI=S−D/Vmean, where S is the peak systolic velocity, D is the minimum diastolic velocity, and Vmean is the mean flow velocity 9. The RI was calculated according to the formula RI=S−D/S, where S is the peak systolic velocity and D is the minimum diastolic velocity 9. Venous blood samples were withdrawn preoperatively from all the women included for serum CA125 assay [through a solid-phase two-site chemiluminescent enzyme immunometric assay, using IMMULTIE 1000 OM-MA (DPC Biermann, Bad Nauheim, Germany)].
The following scoring systems were used in the current study.
Risk of malignancy index
RMI was calculated according to the formula: RMI=U×M×CA125, where U is the ultrasound score, M is the menstrual status, and CA125 is the serum level of CA125 in IU/ml. One point was assigned to any of the following ultrasound features: multilocular mass, bilateral masses, presence of solid areas, presence of metastases, or presence of ascites. The ultrasound score (U) was 1 if the summation of the points was 0–1 and was 3 if the summation of the points was 2 or more. Menstrual status (M) was assigned a score of 1 if the patient was premenopausal and 3 if the patient was postmenopausal or undergone a hysterectomy previously 5.
The Sassone score was calculated according to the characteristics presented in Table 1.
Ovarian tumor index
The OTI included the Sassone score, PI of mass vasculature, and the patient’s age 3.
Pelvic mass score
The PMS was calculated according to the following formula 3:
where log (CA125) is the base 10 logarithm of the serum CA125 concentration (IU/ml); MS is the menopausal state (premenopausal=1; postmenopausal=3); RI is the numeric value of RI of the pelvic mass; SASS is the numeric value of the Sassone score; and VAS is the type of vascularization (peripheral=1; central/septal=2).
All the women included underwent a surgical intervention in the form of laparotomy for cytoreduction, oophorectomy/adnexectomy, ovarian cystectomy, or just deroofing and biopsy. Postoperatively, the removed specimen was sent for histopathological analysis at the Histopathology Lab at the early cancer detection unit at Ain Shams University Maternity Hospital.
Sample size justification
The risk of malignancy in women presenting with an adnexal mass ranges between 8 and 32%. Data from a previous similar study showed that the sensitivity of PMS in the prediction of ovarian malignancy was 93% [95% confidence interval (CI) 82–98%] 3. Calculation according to these values to determine the least statistically acceptable sample size produced a minimal sample size of 175 women.
Statistical analysis was carried out using Microsoft Excel version 2010 and SPSS for Windows version 15.0 (SPSS Inc., Chicago, Illinois, USA). Data were described as range, mean and SD (for numeric parametric variables), range, median and interquartile range (for numeric nonparametric variables), or number and percentage (for categorical variables). The difference between two independent groups was estimated using an independent Student’s t-test (for numeric parametric variables), the Mann–Whitney U-test (for numeric nonparametric variables), or χ2 (for categorical variables). Significance level was set at 0.05. Receiver operator characteristic (ROC) curves were constructed for different scoring systems as predictors of malignancy, advanced stages, and high grades. The accuracy of the different scoring systems was determined according to both previously defined cutoff values (200 for RMI 5, 9 for Sassone score 6, 100 for OTI 7, and 29 for PMS 3), and other cutoff values proposed by the current study that showed slightly better predictive performance. Accuracy was expressed in terms of sensitivity, specificity, and positive predictive value (PPV) and negative predictive value (NPV).
A total of 176 women with an ovarian mass were included in the study. The mean age of the women included was 42.94±13.53 years (range 18–76 years). The median parity was 3 (range 0–11, interquartile range 1–4). Of the 176 women included, 116 (65.9%) were premenopausal, whereas 60 (34.1%) were postmenopausal. The ovarian mass was unilateral in 152 (86.4%) women and bilateral in 24 (13.6%) women. Of the 176 women included, 88 (50%) underwent staging laparotomy, 26 (14.8%) underwent adnexectomy or oophorectomy, 59 (33.5%) underwent ovarian cystectomy, and three (1.7%) underwent just biopsy with or without deroofing. The ovarian mass was of a benign nature in 98 (55.7%) women, of a borderline nature in eight (4.5%), and of a malignant nature in 70 (39.8%).
The proportion of postmenopausal women was significantly higher among those who had malignant tumors compared with those who had benign or borderline tumors. The median serum CA125 was significantly higher among those who had malignant tumors compared with those who had benign or borderline tumors. The median ovarian volume was significantly higher in women with borderline tumors and malignant tumors compared with those with benign tumors. The difference between those with malignant and those with borderline tumors in the ovarian volume was, however, not significant (Table 2).
Table 3 shows the differences between the women included with various types of ovarian masses in terms of the Sassone score, OTI, RMI, and PMS. The median Sassone score was significantly higher in women with malignant and borderline ovarian tumors compared with women with benign ovarian lesions. There was no significant difference between women with malignant and women with borderline tumors in the median Sassone score. The median OTI was significantly higher in women with malignant and borderline ovarian tumors compared with women with benign ovarian lesions. There was no significant difference between malignant and borderline ovarian tumors in OTI and its components. The median RMI was significantly higher in women with malignant ovarian tumors compared with both those with benign and borderline ovarian lesions. The median RMI was significantly higher in women with borderline ovarian tumors compared with women with benign ovarian lesions. The median PMS was significantly higher in women with malignant ovarian tumors and those with borderline ovarian tumors compared with those with benign ovarian lesions. There were no significant differences between women with malignant and those with borderline ovarian tumors in the PMS score and its components (Table 4).
ROC curves were constructed to determine the association between an ovarian malignant tumor and each factor of ovarian volume, mass vasculature PI and RI, Sassone score, OTI, RMI, and PMS (Fig. 1). All of these variables were significantly associated with ovarian malignant lesions as indicated by the significantly large area under the curves (AUCs). PMS was the most significant predictor among all these variables and score by having the largest AUC (0.919, 95% CI 0.878–0.960, P<0.001), followed by RMI (AUC 0.900, 95% CI 0.850–0.951, P<0.001) and OTI (AUC 0.802, 95% CI 0.735–0.869, P<0.001). Table 5 shows the accuracy of ovarian volume, mass vasculature PI and RI, Sassone score, OTI, RMI, and PMS in the prediction of ovarian malignancy. Tables 6 and 7 show false-positive and false-negative cases of these parameters. PMS showed the highest sensitivity (92.9%) and NPV (94.4%) and the lowest false-negative rate (FNR) among all the variables and scoring systems. PMS with a cutoff value of 29 missed only five (7.1%) cases of ovarian malignancy (two cases of endometrioid adenocarcinoma, two cases of sex-cord stromal tumors, and a case of a Krukenberg tumor). PMS according to the same cutoff value (≥29) had the second best specificity (80.2%) and PPV (75.6%), and the second lowest false-positive rate (FPR) (19.8%). It was only superseded by RMI (the corresponding values were 81.1, 75, and 18.9%, respectively, for a cutoff value of 134; and 84, 77.3, and 16%, respectively, for the universally accepted cutoff value of 200). Both PMS and RMI showed a false-positive result in endometrioma, mature cystic teratoma, serous and mucinous cystadenoma, and borderline tumors. PMS showed a false-positive result in inflammatory masses and in pseudomyxoma peritoneii as well, whereas RMI according to a cutoff value of 200 did not.
ROC curves were constructed for measured variables and scoring systems as predictors of advanced stages of ovarian malignancy (stages III/IV). RMI and PMS were the only scoring systems that showed significant predictability (AUC 0.779, 95% CI 0.668–0.891, P<0.001 and AUC 0.698, 95% CI 0.569–0.827, P=0.006, respectively) (Fig. 2). Table 7 shows the validity of the best cutoff values for RMI (≥1005) and PMS (≥63) as predictors of advanced stages of ovarian malignancy (stages III/IV).
ROC curves were constructed for measured variables and scoring systems as predictors of high grades of ovarian malignancy (grades 2/3). Sassone score, PMS, and RMI were the only significant predictors of advanced grades of ovarian malignancy (AUC 0.709, 95% CI 0.551–0.867, P=0.012; AUC 0.692, 95% CI 0.521–0.863, P=0.022; AUC 0.678, 95% CI 0.520–0.835, P=0.034, respectively) (Fig. 3). Table 8 shows the diagnostic validity of the best cutoff values for Sassone score (≥10), PMS (≥945), and RMI (≥45) as predictors of high grading of ovarian malignancy (grades 2/3).
The PMS incorporates five items: menopausal status, the Sassone score, serum level of CA125, type of vascularization of the mass, and RI of the mass vasculature. Among the 70 women included who had ovarian malignancy, 43 (61.4%) were postmenopausal, emphasizing the high risk of malignancy in postmenopausal women presenting with an ovarian mass. Menopausal status is a significant ‘sharer’ in the risk of malignancy, and should, therefore, be included in any scoring system for the prediction of ovarian malignancy 3. The two scoring systems that incorporate the menopausal status (namely the RMI and the PMS) were those that had the highest predictive accuracy. Both of the sonographic systems described (Sassone score and OTI) were shown to have comparable predictive accuracy. The OTI had an advantage as it includes the patient’s age and a Doppler index (PI of the ovarian vasculature). The OTI, however, had the limitation that it included only the ovarian volume. The ovarian volume was shown by the current and previous studies to be a poor predictor of ovarian malignancy 10. Ovarian volume is also a misleading factor; both practice and the literature report a large number of cases with huge ovarian masses that were benign in nature. However, the Sassone score focuses on detailed characteristics of the wall structure and thickness, as well as echogenicity and presence or absence of septae, characteristics that are strongly associated with malignancy. The Doppler index and the location of vascularization, which were missing in the Sassone score, were incorporated into the proposed PMS to compensate for such a limitation. The PMS, therefore, ‘merged’ the advantages of OTI and the Sassone score, and excluded the parameter with poor predictability (ovarian volume). Finally, serum CA125 has long been known to be a sensitive marker in ovarian malignancy, particularly the epithelial types. Again, the two scoring systems that incorporate serum CA125 level (the RMI and the PMS) had the best predictive performance among other scoring systems. The accuracy indices of PMS in the current study were comparable with those reported by Rossi et al.3 at the same cutoff value (≥29) (sensitivity: 92.9, 93%; specificity: 80.2, 88%; PPV: 75.6, 76%; and NPV: 94.4%, 97%, respectively). Both studies rated PMS with the best overall diagnostic accuracy among all other scoring systems. Ovarian volume and Doppler indices (PI and RI of mass vasculature) individually showed markedly lower accuracy indices, particularly in specificity and PPV. The predictive performance of the Sassone score, OTI, and RMI was comparable with the results published in previous studies 3,5–7,11–14. The current study, however, proposed quite different cutoff values for these scoring systems (10 rather than 9 for the Sassone score; 54 rather than 100 for OTI; and 134 rather than 200 for RMI). These different cutoff values showed slightly better overall accuracy in the women included compared with the previously defined ones.
In the current study, PMS 29 or more was associated with the lowest FNR (7.1%) [two cases of endometrioid adenocarcinoma, two cases of nonepithelial neoplasms (sex-cord tumors), and one case of secondary malignancy (Krukenberg tumor)]. RMI 200 or more missed three further cases (serous and clear cystadenocarcinoma as well as immature teratoma); thus, it had an FNR of 17.1%. OTI 54 or more was associated with an FNR of 22.9% (eight cases of endometrioid adenocarcinoma, four cases of serous cystadenocarcinoma, one case of mixed epithelial tumor, and twp cases of undifferentiated epithelial carcinoma). Sassone score 9 or more was associated with an FNR of 17.1% (five cases of serous cystadenocarcinoma, five cases of endometrioid adenocarcinoma, one case of clear cell carcinoma, and one case of Leydig cell tumor). Therefore, PMS was the most sensitive among all the scoring systems tested. All scoring systems and proposed preoperative predictors share a common drawback, namely, the rather high FPR. All markers (whether sonographic or biochemical) show positive results with various benign lesions, including inflammatory masses, endometriomas, and mature teratomas. In the current study, OTI 100 or more was associated with the lowest FPR (3.8%); yet, such a cutoff value was associated with unacceptably high FNR (71.4%). PMS 29 or more had the second acceptable lowest FPR (19.8%) (only superseded by the RMI≥200, which had an FPR of 16%).
All scoring systems could differentiate between benign and borderline ovarian tumors. However, Only the RMI, could differentiate malignant from borderline tumors, moreover, a second point that features the RMI over the PMS is the prediction of advanced stages (III/IV) and high tumor grading (grades 2–3), only the RMI and the PMS showed significant predictive performance, with the RMI slightly, but not significantly, better than the PMS in association with advanced stages. The sample size, although statistically justifiable, was too small to validate a new scoring system. Larger and even multicenter studies pooling data from different gynecological and oncological centers with a much larger sample size need to validate the results of the current study practically and clinically.
PMS seems to be a significant preoperative predictor of ovarian malignancy in women with a diagnosis of an ovarian mass. It incorporates sonographic, biochemical, and Doppler parameters as well as the patient’s menopausal status into one scoring system. It, therefore, showed a slightly higher sensitivity, and a lower, yet comparable, specificity, compared with the widely acceptable RMI.
Conflicts of interest
There are no conflicts of interest.
1. D’Alò D, Stracci F, Cassetti T, Scheibel M, Pascucci C, La Rosa F. Recent trends in incidence, mortality and survival after cancer of the female breast and reproductive organs. Umbria, Italy: 1978–2005.Eur J Gynaecol Oncol 2010; 31:174–180.
2. Valentin L, Hagen B, Tingulstad S, Eik-Nes S. Comparison of pattern recognition’ and logistic regression models for discrimination between benign and malignant pelvic masses: a prospective cross validation. Ultrasound Obstet Gynecol 2001; 18:357–365.
3. Rossi A, Braghin C, Soldano F, Isola M, Capodicasa V, Londero AP, et al.. A proposal for a new scoring system to evaluate pelvic masses: Pelvic Masses Score (PMS). Eur J Obstet Gynecol Reprod Biol 2011; 157:84–88.
4. Van Trappen PO, Rufford BD, Mills TD, Sohaib SA, Webb JA, Sahdev A, et al.. Differential diagnosis of adnexal masses: risk of malignancy index
, ultrasonography, magnetic resonance imaging, and radioimmunoscintigraphy. Int J Gynecol Cancer 2007; 17:61–67.
5. Jacobs I, Oram D, Fairbanks J, Turner J, Frost C, Grudzinskas JG. A risk of malignancy index
incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol 1990; 97:922–929.
6. Sassone AM, Timor-Tritsch IE, Artner A, Westhoff C, Warren WB. Transvaginal sonographic characterization of ovarian disease: evaluation of a new scoring system to predict ovarian malignancy. Obstet Gynecol 1991; 78:70–76.
7. Twickler DM, Forte TB, Santos-Ramos R, McIntire D, Harris P, Miller DS. The ovarian tumor index predicts risk for malignancy. Cancer 1999; 86:2280–2290.
8. Pavlik EJ, Depriest PD, Gallion HH, Ueland FR, Reedy MB, Kryscio RJ, Van Nagell JR Jr. Ovarian volume related to age. Gynecol Oncol 2000; 77:410–412.
9. Tongsong T, Wanapirak C, Neeyalavira V, Khunamornpong S, Sukpan K. E-flow Doppler indices for prediction of benign and malignant ovarian tumors. Asian Pac J Cancer Prev 2009; 10:139–142.
10. DePriest PD, Van Nagell JR Jr, Gallion HH, Shenson D, Hunter JE, Andrews SJ, et al.. Ovarian cancer screening in asymptomatic postmenopausal women. Gynecol Oncol 1993; 51:205–209.
11. Wanapirak C, Srisupundit K, Tongsong T. Sonographic morphology scores (SMS) for differentiation between benign and malignant adnexal masses. Asian Pac J Cancer Prev 2006; 7:407–410.
12. Hossain F, Karim MN, Rahman S, Khan N, Siddiqui M, Hussain R. Preoperative detection of ovarian cancer by color Doppler ultrasonography and CA 125. Bangladesh Med Res Counc Bull 2010; 36:68–73.
13. Clarke SE, Grimshaw R, Rittenberg P, Kieser K, Bentley J. Risk of malignancy index
in the evaluation of patients with adnexal masses. J Obst Gynaecol Can 2009; 31:440–445.
14. Bouzari Z, Yazdani S, Kelagar ZS, Abbaszadeh N. Risk of malignancy index
as an evaluation of preoperative pelvic mass. Caspian J Intern Med 2011; 2:331–335.