Diagnostic value of ultrasound indicators of neoplastic risk in preoperative differentiation of adnexal masses.

AIM: To assess the diagnostic value of the risk of malignancy indices and simple ultrasound- based rules in preoperative differentiation of adnexal masses.
MATERIAL AND METHODS: Retrospective examination of 87 patients admitted to hospital due to adnexal tumors. The lesions were evaluated on the basis of international ultrasound classification of ovarian tumors and four risk of malignancy indices were calculated based on ultrasound examination, concentration of CA 125 and menopausal status.
RESULTS: The patients were aged between 17 and 79, the mean age was 44.5 (standard deviation SD=16.6). Most of the patients (60.91%) were before their menopause. The sensitivity of the simple ultrasound-based rules in the diagnosis of malignancies equaled 64.71% and the specificity constituted 90.00%. A significant statistical difference in the presence of the malignant process was demonstrated in relation to age, menopausal status, CA 125 concentration and analyzed ultrasound score. All indices were characterized by similar sensitivity and specificity. The highest specificity and predictive value of malignant lesions out of the assessed ones was demonstrated by the risk of malignancy index proposed by Yamamoto. The risk of malignancy index according to Jacobs, however, showed the highest predictive value in the case of non-malignant lesions.
CONCLUSIONS: The multiparametric ultrasound examination may facilitate the selection of patients with adnexal tumors to provide them with an appropriate treatment - observation, laparotomy and laparoscopy. These parameters constitute a simple ambulatory method of determining the character of adnexal masses before recommending appropriate treatment.

Introduction

Adnexal masses constitute one of the most common causes of hospitalization on gynecological wards. The preoperative differentiation between malignant and benign tumors constitutes one of the most important stages in lesion evaluation since it decides about further treatment(. Diagnostic difficulties connected with detecting ovarian cancer result from the fact that it is not a uniform disease entity. The latest findings indicate that in the majority of cases, ovarian cancer derives from fallopian tube and endometrial cells. Primary malignant neoplasms of the ovaries, however, which originate from germ cell or gonadal stroma, occur relatively rarely(. The diagnosis and initial evaluation of adnexal tumors are established on the basis of the history, clinical examination, ultrasound scan as well as neoplastic markers(. Ultrasound examination constitutes the primary method which allows for thorough assessment of the lesions. It is, however, somewhat subjective and dependant on the human factor(. In order to facilitate the standardization and simultaneously, the objectivity of ultrasound examinations, an international ultrasound classification of ovarian masses was created (IOTA)(. The IOTA group established simple and transparent rules regarding the interpretation of ultrasound findings irrespective of the examiner's experience. They enable the categorization of adnexal masses as malignant, non-malignant and unclassified. The last one requires an ultrasound verification conducted by physicians with greater experience(.

In search of additional and the most objective ultrasound parameters in evaluating adnexal masses, so called cancer risk indices were created. In 1990s, Jacobs et al. presented a simple diagnostic method based on mathematical estimation of adnexal malignancy risk called the risk of malignancy index (RMI). This index was calculated on the basis of simple ultrasound features, menopausal status and serum CA 125 concentration(. Over the years, RMI was modified and perfected so as to achieve a parameter of the highest possible values of diagnostic specificity and sensitivity. Thus, subsequent modifications of RMI calculations, including adding the tumor size to the mathematical models, which was put forward by Yamamoto et al., significantly improved the results of this diagnostic method(.

The aim of this paper is to analyze the diagnostic value of individual risk of malignancy indices and simple ultrasound-based rules in order to create a simple and objective clinical algorithm for the classification and preoperative differentiation of adnexal masses. The paper constitutes a preliminary report.

Material and methods

The study was retrospective and encompassed 87 subjects, who due to adnexal masses in 2011, were admitted to the Department and Clinic of Obstetrics, Women's Diseases and Gynaecologic Oncology of the Second Faculty of Medicine of the Medical University of Warsaw. The patients, whose level of CA 125 had not been examined as well as those who had undergone an ultrasound scan earlier than 90 days before the surgery, were excluded from the study. The findings of transvaginal ultrasound examinations were subject to analysis. Apart from the assessment of the tumor size, its morphology and echostructure, the degree of the tumor vascularization was determined in Doppler examination.

If the tumors were larger than 5 cm and extended beyond the pelvis minor, the transvaginal ultrasound was supplemented with transabdominal examination. In the case of bilateral lesions, only the mass of more complex echostructure was subject to analysis. The ultrasound examinations were performed in ambulatory conditions by the attending physician prior to the referral to the hospital. The patients were examined by physicians of various levels of experience by means of various ultrasound apparatuses. The classification into given categories was based on “simple ultrasound-based rules” according to IOTA. Thus, five ultrasound characteristics were determined which indicated the malignant character of the tumor (M): M1 – irregular solid tumor; M2 – ascites; M3 – the presence of at least four papillary projections; M4 – irregular multilocular tumor with the presence of solid components and the size exceeding 100 mm; M5 – increased vascularization detected in color coded Doppler examination. The ultrasound features of benign lesions (B) were as follows: B1 – unilocular cyst; B2 – the presence of solid components where the largest is not greater than 7 mm; B3 – acoustic shadow; B4 – multilocular cyst with smooth walls; B5 – no signs of blood flow in color Doppler examination.

On the basis of these features, the tumor was classified as malignant if it demonstrated at least one “M” feature and no “B” features (rule 1) (figs. 1, 2). The mass was considered benign if at least one “B” feature was present and no “M” features were confirmed (rule 2) (fig. 3). Finally, the ultrasound findings were considered unclassifiable if both “M” and “B” features were detected simultaneously or none of them occurred (rule 3) (fig. 4)(. Next, four risk of malignancy indices (RMI) were calculated as provided by various authors: Jacobs in 1990, Tingulstad in 1996, Tingulstad in 1999 and Yamamoto in 2009(. These indices are based on three parameters: ultrasound parameter (U), menopausal status (M) and CA 125 concentration expressed in units per milliliter. The ultrasound parameter (U) encompasses five characteristic features (each detected feature is given one score). These parameters constitute: multilocular cyst, presence of solid components within the lesion, evidence for metastases in ultrasound scan, ascites and bilateral character of lesions. The menopausal status was defined as amenorrhea lasting at least a year and the age of 50 and more in patients with the history of hysterectomy. The serum CA 125 concentration is used directly in all models of RMI calculation. RMI value constitutes the product of the following equation: RMI=U×M×CA 125(. According to the RMI I system created by Jacobs et al., the U parameter may have the following values: U=0 (score: 0) when no features mentioned above are present; U=1 (score: 1) when one feature is found; U=3 (score: 2–5) when the number of detected features is equal of greater than 2. The M parameter may have the following values: M=1 in the case of pre-menopausal patients or M=3 in the case of post-menopausal patients(. According to the RMI II system proposed by Tingulstad et al.: U=1 (score: 0–1) when only one or none of the features are present; U=4 (score: 2 and more) when there are two or more features. The M parameter may have the following values: M=1 in the case of pre-menopausal patients or M=4 in the case of post-menopausal patients(. According to the RMI III system proposed by Tingulstad et al.: U=1 (score: 0–1) when only one or none of the features are present; U=3 (score: 2 and more) when there are two or more features. The M parameter has the following values: M=1 in the case of pre-menopausal patients or M=3 in the case of post-menopausal patients(. The RMI IV system put forward by Yamamoto et al. includes one additional ultrasound element. Namely, the greatest diameter of the adnexal tumor (S). Thus, the RMI IV value is the product of the following equation: RMI=U×M×CA 125×S. According to this system: U=1 (score: 0–1) when only one or none of the features are present; U=4 (score: 2 and more) when there are two or more features. The M parameter has the following values: M=1 in the case of pre-menopausal patients or M=4 in the case of post-menopausal patients. Finally, the parameter S=1 when the greatest size of the tumor is lower that 7 cm or S=2 when the greatest size of the tumor constitutes 7 cm or more(.

Fig. 1

Solid-cystic ovarian mass with irregular margins and the size of 43×28 mm. Color Doppler examination shows increased blood flow. According to IOTA, the ultrasound image presents a malignant adnexal tumor

Fig. 2

Low-resistance flow (RI=0.33) in the wall of the ruptured cystic adnexal tumor

Fig. 3

A bilocular cyst of the ovary with smooth walls and the dimensions of 58×68 mm. According to IOTA, the ultrasound image presents a non-malignant adnexal tumor

Fig. 4

Polycystic tumor with solid components of irregular margins and the diameter of about 200 mm visualized in a transabdominal ultrasound examination. Color Doppler scan does not present any features of vascularization. According to IOTA, ultrasound findings are unclassifiable

Due to the large number of analysed parameters, there are no exact threshold values for these four indices. According to the references quoted herein, it was assumed that the values between 0–200 in RMI I-III indices attested to a non-malignant lesion and the values above 200 – to a malignant one. In RMI IV, however, the values between 0–500 and more than 500 are attested to non-malignant and malignant lesions respectively(. The final diagnosis of adnexal masses was determined on the basis of post-operative histopathological examinations. The stage of malignant neoplasms was determined on the basis of the FIGO classification. In the statistical analysis, the tumors of borderline malignancy were considered malignant. The conducted analyses concerned the specificity, sensitivity as well as positive and negative predictive values of the risk of malignancy indices and the method of simple ultrasound-based rules. The Mann-Whitney U test was used in order to conduct statistical analysis of the correlation of CA 125 concentration with the types of adnexal tumors. The chi-squared test of independence was used for the statistical analysis of the patients with various tumors in relation to age, menopausal status, ultrasound score and tumor size.

Results

The examined group comprised 87 women aged between 17 and 79, the mean age was 44.5 (SD=16.6). In 70 subjects (80.5%), non-malignant adnexal lesions were detected and 17 patients (19.5%) were diagnosed with a malignant tumor. The malignancies included: 7 cases (41.2%) of serous cystadenocarcinoma, 3 cases (17.6%) of metastatic tumors from the gastrointestinal tract, 3 cases (17.6%) of borderline tumors as well as one case of mucinous cystadenocarcinoma, fallopian tube carcinoma, mixed carcinoma and undifferentiated carcinoma (5.9%). The stages of the malignancies according to FIGO classification were as follows: stage I – 7 cases (41.2%), stage III – 6 cases (35.3%), stage IV – 1 case (5.9%). Three cases (17.6%) were related to malignant neoplasms which originated in the gastrointestinal tract.

The statistical analysis showed that in the preoperative assessment of adnexal masses, the sensitivity of simple ultrasound parameters according to IOTA constituted 64.71% and the specificity was on the level of 90.00%. The predictive values of malignant and non-malignant lesions were 91.67% and 98.44% respectively. Tab. 1 presents the classification of adnexal tumors according to IOTA which was the basis for the estimation of the aforementioned statistical data.

Tab. 1

Types of neoplastic lesions according to IOTA

US assessment	Histopathological findings
	Non-malignant (n=70)		Malignant (n=17)
	N	%	N	%
Non-malignant lesion	63	90,00	1	5,88
Inconclusive finding	6	8,57	5	29,41
Malignant lesion	1	1,43	11	64,71

The subsequent stage of the study aimed at determining whether the patients with various types of lesions differed in age, menopausal status, ultrasound score and tumor size. The obtained results are presented in tab. 2.

Tab. 2

Age, menopausal status, ultrasound score and tumor size in the examined group

Parameter in question	Type of tumor (histopathology)				Chi2, p level
	Non-malignant		Malignant
	N	% of the group	N	% of the group
Age (years)
Below 20	4	5,71	0	0,00	chi2(3)=18,16, p<0,001
21–40 lat	38	54,29	2	11,76
41–50 lat	8	11,43	0	0,00
Above 50	20	28,57	15	88,24
Menopausal status
Before menopause	50	71,43	3	17,65	chi2(1)=16,82, p<0,001
After menopause	20	28,57	14	82,35
Ultrasound score
0 pkt	27	38,57	0	0,00	chi2(2)=15,29, p<0,001
1 pkt	19	27,14	3	17,65
2–5 pkt	24	34,29	14	82,35
Greatest tumor size
<7 cm	53	75,71	4	23,53	Chi2(1)=15,93, p<0,001
>7 cm	17	24,29	13	76,47

The analysis of the aforementioned parameters demonstrated that there is a statistically significant difference concerning age, menopausal status, ultrasound score and tumor size between the groups of patients with non-malignant and malignant adnexal tumors. The mean age of the patients with non-malignant and malignant tumors constituted 41.2 and 57.6 respectively. The group of subjects with non-malignant masses encompassed 50 premenopausal patients (71.43%) and 20 postmenopausal patients (28.57%). In comparison, the group with malignant lesions, encompassed 3 (17.65%) and 14 (82.35%) pre- and postmenopausal patients respectively. The analysis of Jacobs rule demonstrated that among the patients with benign lesions 27 (38.57%) scored 0, 19 (27.14%) scored 1 and 24 (34.29%) scored 2–5. In the group with malignant tumors, on the other hand, all patients’ scores exceeded 0. Three subjects (17.65%) scored 1 and 14 (82.35%) scored 2–5. The analysis of CA 125 concentration level, showed that its level was higher in the women with malignant tumors than in those with benign lesions. Here, the Mann-Whitney U test was applied. Tab. 3 presents the statistical data obtained in this analysis.

Tab. 3

Statistical data for CA 125 concentration in relation to the type of lesion

Lesion	Average	Minimal concentration	Maximal concentration	SD	U Test and p level
Non-malignant	36,95	2,5	160,0	39,56	Z=4,51, p<0,001
Malignant	418,32	13,8	1767,0	560,07

Tabs. 4–7 present the results of predicting the type of neoplastic lesion on the basis of the parameters in RMI I, II, III and IV.

Tab. 4

Evaluation of neoplastic lesions according to RMI I

Type of tumor according to RMI I	Type of tumor (histopathology)
	Non-malignant		Malignant
	N	%	N	%
Non-malignant	62	88,57	2	11,76
Malignant	8	11,43	15	88,24

Tab. 7

Evaluation of neoplastic lesions according to RMI IV

Test result according to RMI IV	Type of tumor (histopathology)
	Non-malignant		Malignant
	N	%	N	%
Non-malignant	66	94,29	3	17,65
Malignant	4	5,71	14	82,35

The chi-squared test of independence has shown statistically significant differences of the RMI I test: chi2(1)=38.43, p<0.001; RMI II: chi2(1)=27.31, p<0.001; RMI III: chi2(1)=35.22, p<0.001; RMI IV: chi2(1)=44.40, p<0.001. All the indices demonstrated a similar diagnostic sensitivity. RMI IV was characterized by the highest specificity and predictive value in the case of malignant lesions. RMI I, on the other hand, demonstrated the highest predictive value in the case of benign lesions. Tab. 8 presents a detailed summary of the diagnostic value for individual RMI.

Tab. 8

Comparison of RMI in the patients with adnexal tumors

RMI	Sensitivity	Specificity	PPV	NPV	Estimation accuracy
RMI I	88,24	88,57	65,22	96,88	88,51
RMI II	88,24	81,43	53,57	96,61	82,76
RMI III	88,24	87,14	62,50	96,83	87,36
RMI IV	82,35	94,29	77,78	95,65	91,95

PPV – positive predictive value; NPV – negative predictive value.

Discussion

The classification of tumors according to IOTA and by means of risk of malignancy indices (RMI) is a simple diagnostic method of preoperative discrimination between malignant and benign adnexal masses. It may be performed in ambulatory conditions(. The differentiation of such lesions on the basis of one ultrasound model frequently presents insufficient levels of specificity and negative prognostic value. This results from the subjectivity of the examination and is related to the examiner's experience. Consequently, such results prevent the classification of the patients to appropriate type of treatment(. The serum concentration of CA 125 is in turn burdened by a large number of false positive errors which also decrease the specificity of the test(. The possibility to exclude malignant adnexal lesions is of vital importance especially for young patients who wish to conserve their reproductive capacity. Conservative treatment or minimally invasive procedures are possible in such patients(. On the other hand, the patients with the suspicion of malignancies should be treated radically after careful treatment planning. Such radical actions influence the basic prognostic factor determining 5-year survival(. A proper surgical plan, among others laparotomy and midline vertical incision, which facilitates complete staging, may decrease the iatrogenic risk of increasing FIGO stage by, for instance, intraoperative rupture of the adnexal tumor(. Therefore, it is crucial to establish a diagnostic method enabling precise risk estimation of the presence of a malignant tumor in a preoperative period. It is believed that only multiparametric assessment allows for precise diagnosis in these difficult cases. It seems that the incorporation of the risk of malignancy index (RMI) into the ultrasound algorithm fully satisfies these needs. The presented research shows that RMI indices demonstrated high sensitivity in detecting malignant lesions. Out of the assessed indices, the highest specificity and predictive value of malignant lesions was provided by RMI IV. RMI I, in turn, was characterized by the highest value in diagnosing benign lesions. The false positive results were detected in 8 (11.43%), 13 (18.57%), 9 (12.86%) and 4 (5.71%) patients with benign lesions according to RMI I-IV respectively.

The sensitivity of the simple ultrasound-based rules was slightly higher when compared with RMI. In our research ultrasound findings were unclassifiable in 5 cases (3 cases of serous cystadenocarcinoma, one mucous cystadenocarcinoma and one borderline tumor). In spite of more and more advanced ultrasound equipment, adnexal tumors still constitute a significant diagnostic problem. Such problems, first and foremost, concern malignant lesions at an early stage, which constitute the cases requiring precise preoperative diagnosis in order to establish appropriate treatment. This, in turn, affects prognosis in a significant way and improves the overall results of ovarian cancer treatment. This study shows that the multiparametric ultrasound assessment may constitute the basis for preoperative differentiation of adnexal tumors. On the other hand, it needs to be remembered that ultrasound findings must be carefully interpreted together with a complete clinical picture supported with the history , especially the oncological one. The interpretation should also be based on the concentration levels of other neoplastic markers and, if need be, on the findings of other imaging examinations such as computed tomography and magnetic resonance imaging(.

Tab. 5

Evaluation of neoplastic lesions according to RMI II

Type of tumor according to RMI II	Type of tumor (histopathology)
	Non-malignant		Malignant
	N	%	N	%
Non-malignant	57	81,43	2	11,76
Malignant	13	18,57	15	88,24

Tab. 6

Evaluation of neoplastic lesions according to RMI III

Type of tumor according to RMI III	Type of tumor (histopathology)
	Non-malignant		Malignant
	N	%	N	%
Non-malignant	61	87,14	2	11,76
Malignant	9	12,86	15	88,24