Full Text Article

Categorical variables are expressed as rates and measurable variables are expressed as mean (standard deviation) or median (interquartile range).

We calculated the ROC curves: The optimal cutoff point was estimated by the point that maximizes the sum of specificity and sensitivity. All statistical analyses were performed with the use of SPSS software (version 17.0), and all reported probability values were 2-sided. We assumed significance at the 5% level (P<0.05).

Results

Patients

A total of 154 women were studied, with a mean (SD) age of 44.3 (14.3), 25.3% of whom had reached menopause status.

GI-RADS

The results of pelvic ultrasononography are described in Table 2 which reveals that half of the cases were classified as GI-RADS 3, the most frequent mass being endometrioma. 24% of the cases presented risk or high risk of malignancy (GI-RADS 4 and GI-RADS 5).

Solid areas (12.3%), thick septations (11.7%) and vascularization as evidenced on

Doppler sonography (13%) were the most frequent malignancy findings.

TABLE 2. Sonographic characteristics

Clinical Outcomes

Of the 154 patients included in our study, one did not undergo follow-up.

The monitoring and therapeutic approach according to the protocol is showed in Figure 1. After the first phase of the follow up study of the adnexal mass, the decision to perform surgery without further testing was taken in 3.9% of the cases. On the first follow-up visit, 14.9% had no pathological images on ultrasound, falling to 9.6% after 6 months.

Twenty-four patients with stable ultrasound images (first and second follow-up visit) underwent surgical treatment. Adnexal masses were removed for different causes: persistent masses changed classification from GI-RADS 2 to 3 or because of associated symptoms such as pain, sensation of weightiness. Surgery was performed in 33.1% of the adnexal masses studied. In 2 patients, the GI-RADS did not correctly discriminate between benignity and malignancy, since the adnexal image was characterized as GI- RADS 2 (being a sarcoma) and GI-RADS 3 (in the case of a serous ovarian carcinoma) and the patients were operated after 6 and 3 months respectively.

Figure 1. Monitoring and therapeutic decision of adnexal masses according to protocol.

The non-operated patients (66.9%) were younger women with a median age of 40, a circumstance that reflects a more conservative therapeutic decision due to unfulfilled reproductive wishes. The majority of these patients had a non-suspicious ultrasound image of malignancy, most frequently classified as GI-RADS 3.

With the disappearance of 21.3% of the adnexal masses during the first two follow-up visit (22 and 11 patients respectively), unnecessary surgeries were avoided.

Clinical Outcomes

Out of 51 patients undergoing surgery, 33.3% presented with ovarian cancer; from whom 88.2% were older women with a score of 4-5 in GI-RADS index. The median time from diagnosis of surgery was 9 months for benign masses and 1.5 months for malignant masses. Five patients with confirmed malignancy (2 cases of ovary, 2 borderline, 1 sarcoma) underwent surgery after 6 months of follow-up.

The case with the greatest delay in the diagnosis (22 months after the first ultrasound scan) was an endometrial sarcoma, a histological finding that was reached after the indicated surgery due to a discrepancy between radiological tests (MRI and CT), which were inconclusive.

Table 3 shows the histologic results of the surgeries performed

Table 3. Histologic results

The tumor marker HE4 was positive in 13 patients (8.4%), Ca125 was positive in 16 (15.5%) and Ca 19.9 in 31 (20.7%). The remaining markers (CEA, BHCG and AFP) were positive in only 3, 1 and 4 cases respectively. From 13 patients with positiveHE4, 4 did not undergo surgical intervention, one due to comorbidity (fragility), despite a high suspicion of malignancy. The other 3 patients did not develop cancer during 6- month follow-up.

All the patients operated on with elevation of the HE4 tumor marker (9 women), had a diagnosis of cancer except for one case of cystadenoma; a false-positive result was obtained in a patient with chronic renal failure and high creatinine levels (Cr: 5.97 mg/dl). Out of 17 malignant neoplasms, 8 patients (47.1%) had positive HE4.

In case of Ca125, only 4 cancer histological had got this tumor marker positive, in 12 cases the result was a false positive

Statistical analysis

When comparing the ultrasonographic characteristics, tumor markers and hormonal status among patients with and without surgical intervention, we observed that the patients who underwent surgery were older, postmenopausal, with larger tumor size, greater HE4 positivity and had a higher GI-RADS [4-5] score, these differences being statistically significant (p <0.05). No differences were found in the Ca125 marker.

Table 4. Comparison of study groups: surgical vs no surgical intervention.

Table 5 contains the results obtained by studying the sensitivity (S) for both the GI- RADS classification and tumor markers according to age.

We obtained a global sensitivity of 88,2% with a specificity (SP) of 83,9%, but in pre- menopausal women the S and SP were 100% and 85,4% respectively. In postmenopausal women we obtained worse results with a S of 83,3% and SP of 77,8%. The sensitivity and specificity of biomarker HE-4 was 70,6% and 92 % respectively. The overall sensitivity obtained for HE4 was 58.8% and for Ca 125 was 26,7%.

Different cut-off points (CP) were calculated for HE4, with the level of 82 pmol/L being the most optimal. When analyzing the HE4 marker according to hormonal status, this tumor marker exhibited better performance in postmenopausal women, although the most appropriate cut-off point was the same for both subgroups. On the other hand, in the study of GI-RADS, we observed a difference of sensitivity according to hormonal state, which raises to 100% in premenopausal women.

Table 5 Calculation of Sensitivity (S), Specificity (E), VPP and VPN of tumor and ultrasound markers.

Discussion

Adnexal masses are found very often in routine gynecology practice. Correct handling of this entity supported by a protocol is highly important.

Our study analyzes the results obtained by applying the adnexal mass protocol in our hospital. This protocol is based on the GI-RADS method, hormonal status and the result in HE4 tumor-marker assay. We have not found any similar protocols in the literature. Based on our results, standardized management of adnexal masses based on the GI- RADS method (Figure 1) is found to be a valid approach.

In most ovarian cancers, suspicion of malignancy came from the findings of the first and/or second ultrasound assessment, with diagnosis taking place later in the case of more rare tumors such as the borderline subtype or uterine sarcoma. In these cases, particularly sarcoma, early detection is very difficult. Two of the cases of carcinoma (serous and endometrioid) in which surgery was performed 6 months after the first ultrasound scan were initially cataloged as GI-RADS 3 and 4, respectively, and the tumor markers were negative. These two are exceptional cases in which standardization fails to detect malignancy.

In 33 patients, the adnexal mass observed in the first ultrasound scan disappeared during follow-up. Thus, the watchful waiting approach indicated in our protocol has an obvious benefit, avoiding the risks derived from overtreatment which involve surgery and its subsequent effects on fertility and risk of menopause.

Most of the cancers that underwent surgery after 6 months were re-classified with a higher GI-RADS score due to the increased complexity evidenced in adnexal image.

Our results reflect an overall GI-RADS sensitivity of 88.2% (100% in premenopausal women) and a specificity of 83.9%. Reviewing the most updated scientific literature there are only two studies that analyze the diagnostic capacity of the GI-RADS method in the adnexal masses, that of Amor, et al. [5] (which reflects a sensitivity of 99.1% and a specificity of 85.7%) and the article by Zhan [8] (sensitivity 96.4% and specificity 84.3%). Both articles conclude that GI-RADS classification performed well in the diagnosis of malignant adnexal masses.

Within the variables included in our protocol, GI-RADS classification was the most decisive for management adnexal mass. The differences found between our results and those appearing in the literature are likely due to the experience of the sonographer performing the characterization. This is considered the main limiting factor of our study.

Although the HE4 marker is a tool that supports diagnosis, in our experience it is less useful for therapy. We obtained a 47.1% sensitivity and a specificity of 96.3%. The most important advantage of this marker is that it is not altered in physiological situations such as the menstrual period. However, special care must be taken when interpreting this value in patients with renal failure. The study by Escudero et al. (9) found that greater levels than 1.3 mg/dL in serum creatinine increased the HE4 concentration. This could explain our false positive of HE4 in a patient with renal insufficiency and creatinine above 5mg/dl who was awaiting kidney transplantation.

Another factor that requires a detailed analysis is the decision of which is the most discriminating point or cut-off value (CP) of the HE4 marker. The CP 150 pmol/L established in 2008 and used in our study showed low figures of sensitivity and specificity. In an attempt to find the most optimal cut-off point and analyzing the s of 150, 130, 100, 82 and 70 pmol/L, the best CP was 82 pmol/L (70.6% sensitivity and 92% specificity). In the study published by Moore et al. (10), they used a CP of 70 pmol/L, a value that caused a loss of specificity which was 88.3%. In our patients, CP 130 was used to improve our diagnostic ability, which has been modified to 100 pmol/L.

On the other hand, the sensitivity of HE4 found in posmenopausal patients was higher (75% sensitivity 81.5% specificity) thus making this a better diagnosis tool in this group.

There are many diagnostic algorithms in the scientific literature (ROMA, CPH-I, RMI, IOTA), though none of them have demonstrated superiority for the management of adnexal masses and have complex scoring systesm. In the study by Adriana Yoshida [11], the author uses the ROMA (Risk of Ovarian Malignancy Algorihm) and the CPH- I (Copenhagen Index) with a sensitivity of 70%, which are lower than ours. The algorithm developed by Sarikapan Wilailak [12] produced the same conclusions as in our study, we observed that HE4 is a better predictor of malignancy than Ca125. In our analysis, the Ca125 marker does not provide additional information as a 26.7% sensitivity was very low, and specificity was 84.1%. The Sarikapan algorithm established a malignancy risk score based on HE4, hormonal status and ultrasound findings (with five features analyzed: multiloculated, solid nodule, bilaterally, ascites and peritoneal metastases). This algorithm had proved superior to the RMI (Risk of Malignancy Index) and ROMA since it obtains better results in sensitivity (77.2%) but not in specificity (86%), similar to our results. In the original study by Timmerman, et al. [13] with the “simple ultrasound-based rules” used in IOTA study, the authors obtained a sensitivity of 93% and a specificity of 90%. However, in order to correctly perform this ultrasound analysis, a specific outpatient visit is needed in order to collect all the items; this method is not reproducible within our “all-in-one” clinical-radiology outpatient system. The GI-RADS reporting system is easier to implement than the IOTA program in a first medical visit by general practitioners.

Moore, et al. [14] compared RMI and ROMA concluding that ROMA is a better diagnostic algorithm because, although both have a specificity of 75%, the sensitivity of ROMA was significantly better than that of the RMI (94.3% to 84.6%), a conclusion we cannot reach in this study because of the use of ultrasound, resonance and TC scan imaging, which are hardly comparable.

According to Frederick R. Ueland [15] their results with OVA1 revealed a sensitivity of 94% but an unacceptable low specificity of 54%.

There were some limitations in our study: namely being a retrospective study with a small cohort from only one hospital, doctors who do ultrasound are experts with years of experience.

As a conclusion, the protocol of action used in our study, based on the GI-RADS method, hormonal status and HE4, facilitates the indication of successful therapeutic measures without overdiagnosing or overtreating patients with benign or functional masses.