Accuracy of transvaginal and transrectal ultrasounds in the diagnosis of endometriosis: A retrospective cohort study

Abstract Background Early diagnosis and appropriate treatment of endometriosis are vital and may prevent subsequent complications. Objective To investigate the diagnostic accuracy of transvaginal ultrasound sonography (TVUS) and transrectal ultrasound sonography for detecting endometriosis considering the age and body mass index (BMI). Materials and Methods This was a retrospective cohort study of 119 women scheduled for surgery in a tertiary health care center for clinically suspected endometriosis. Married and virgin women underwent TVUS and transrectal ultrasound sonography, respectively, before laparoscopic excision of endometriotic lesions. Results The accuracy of TVUS in the diagnosis of right endometrioma in women with a normal BMI was superior to that in women with a BMI ≥ 30 (95.6% vs. 75.3%; p < 0.001). For the detection of left endometrioma in women with a normal BMI, TVUS demonstrated a sensitivity of 96.9% and a negative predictive value of 92.9%, which was significantly superior to TVUS in women with obesity (sensitivity: 77.4%, negative predictive value: 58.6%). The accuracy of TVUS in the diagnosis of left endometrioma in women under 35 yr was superior to that in women older than 35 yr (93.2% vs. 77.9%; p = 0.04). Similarly, the accuracy of TVUS in the diagnosis of right endometrioma in women under 35 yr was superior to TVUS in women older than 35 yr (86.5% vs. 73.3%; p = 0.04). Conclusion Ultrasound can be a useful technique for detecting endometriosis when used adjunctively with the patient's history and physical findings, especially age and BMI.


Introduction
Endometriosis, which is related to the ectopic endometrial glands and outer stroma of the uterus, is a major gynecological health problem in women of reproductive age, affecting 10-15% of this group (1). Deep endometriosis (where lesions reach a depth of 5 mm), which occurs in 15-30% of all diagnosed cases of endometriosis, can cause symptoms such as cyclical dysmenorrhea, dyschezia, deep dyspareunia, variable digestive complaints, and/or subfertility (2).
Early diagnosis and appropriate treatment are vital and may decrease disease progression and prevent subsequent complications (3). Endometriosis may be suspected by examining the signs and symptoms or by using imaging techniques such as transvaginal ultrasound sonography (TVUS) and magnetic resonance imaging, but the gold standard for diagnosis is laparoscopic identification and histological verification of endometriotic tissue (4). However, due to the invasive nature of laparoscopic identification, non-invasive diagnostic techniques have a higher priority (1). The diagnostic accuracy of TVUS has been assessed in numerous previous studies in various settings and populations with different results. In 2 previous systematic reviews and meta-analyses, although the diagnostic accuracy of TVUS and transrectal ultrasound sonography (TRUS) was estimated as appropriate, high heterogeneity between studies prevented a definitive conclusion (5,6). As a result of the observed heterogeneity, efforts to further investigate the accuracy of TVUS and TRUS in various situations are reasonable.
The present study aimed to investigate the diagnostic accuracy of TVUS and TRUS for detecting endometriosis considering the age and body mass index (BMI) of participants.

Materials and Methods
This retrospective cohort study was carried out from May 2018 and March 2020 in Roointan Arash hospital, a tertiary healthcare center affiliated with Tehran University of Medical Sciences, Tehran, Iran. The hospital is a referral center for endometriosis treatment. Over 2 yr, 119 women who were scheduled for laparoscopic surgery due to signs and symptoms of endometriosis were enrolled. Our inclusion criteria were age > 18 yr and diagnosis of endometriosis based on the symptoms and clinical examination. We excluded those with a history of gynecological surgery or cancer, structural anomalies of the reproductive system, pregnancy, or lack of compliance with TVUS or TRUS. All scans were performed by one of the experienced gynecologists who were blinded to the participants' clinical outcomes.

Transvaginal sonography
The ultrasound technique used was based on the agreed protocol of the International Deep Endometriosis Analysis group. The review protocol included viewing compartments, peritoneum, and structures in the anterior and posterior parts as well as the uterus and ovaries. We performed TVUS with an Accuvix XQ scanner (Accuvix Sonoace, Medison Co., Ltd, Seoul, South Korea) using a 5-9-MHz probe for transvaginal visualization of the urinary bladder, vagina, adnexal regions, uterus, and uterosacral ligaments. The evaluation was conducted on the non-menstrual days of the cycle. The participants were asked to have a semi-filled bladder and were submitted to a simple rectal enema (fleet enema) 1 hr prior to the procedure. The procedure was done using lubricant gel and without administration of sedatives. As per routine practice, interpretations were done in real-time and documented in printed photographs for future reference. We defined the TVUS diagnosis of endometriosis based on the "presence of regular or irregular hypoechogenic nodular structure or hypoechogenic linear thickening with regular or irregular margins" (7).

Transrectal sonography
TRUS was performed with an Accuvix XQ scanner (Accuvix Sonoace, Medison Co., Ltd, Seoul, South Korea) using a 5-9-MHz probe for transrectal visualization of the rectosigmoid wall layers. The evaluation was done in nonmenstrual days of the cycle. All participants were asked to do the following before the sonography: I) have a soft diet on the day before sonography; II) skip breakfast on the day of the procedure; III) have 2 spoonfuls of milk of magnesium syrup orally after lunch; and IV) take 2 suppositories of 10 mg bisacodyl at 6 PM and 12 midnight on the day before the procedure. The participants were asked to have a semi-filled bladder and were submitted to a simple rectal enema (fleet enema) 1 hr prior to the procedure. The procedure was done using lubricant gel and without administration of sedatives. As per routine practice, interpretations were done in realtime and documented in printed photographs for future reference. We determined the diagnosis of endometriosis based on the presence of regular or irregular hypoechoic nodular structure or hypoechoic linear thickening with regular or irregular margins (7).

Laparoscopy, radical resection of endometriosis, and histology
All histological confirmations of endometriosis were performed by a pathologist who was blinded to clinical examination and TVUS findings. 2 gynecologists with more than 20 yr experience in radical laparoscopic surgery performed the laparoscopy. We defined deep infiltrating endometriosis as follows: subperitoneal endometriotic infiltration of tissues > 5 mm ( Figure 1). All the biopsies were transferred onto a glass slide and appropriately stained by hematoxylin and eosin for microscopic evaluation. An experienced pathologist performed the diagnosis of endometriosis for all resected tissue samples after evaluating both glands and stroma.

Ethical considerations
Ethics approval was obtained from the Ethical Committee of the Tehran University of Medical Sciences, Tehran, Iran (Code: IR.TUMS.MEDICINE.REC.1399.065). All participants read and signed an informed consent form prior to enrollment in the study. Participants' data were kept confidential and anonymous.

Statistical analysis
The analyses were carried out using Stata software version 16 (Stata Corp, College Station, Texas, United States). BMI was categorized as underweight (< 18.5), normal (18.5-24.9), overweight (25-29.9) or obese (≥ 30). Continuous variables were described by mean ± standard deviation (SD). Categorical variables were shown as numbers and percentages. We defined accuracy as the results of a diagnosis test (positive or negative) against the true disease using a gold standard (presence or absence). The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), diagnostic odds ratio (DOR), and area under the curve of TVUS and TRUS were evaluated for each involvement site considering BMI and age categories. All accuracy indices were presented with 95% confidence intervals to determine the precision of the results. The accuracy was compared between ultrasound and laparoscopy using McNemar's test. All calculated p-values were 2-tailed. P < 0.05 indicated statistical significance.

Results
Out of the 168 eligible women, 119 participants were enrolled in this study and 49 were excluded because of having a history of previous surgery for deep infiltrating endometriosis (n = 38), a history of gynecological cancer (n = 5), or because they were not willing to participate in the study (n = 6). The participants' mean ± SD of age was 33.76 ± 7.10 (median: 34 yr; interquartile range: 38, 29), and 26 (21.85%) of them were virgin. Dysmenorrhea was the most common symptom among the participants (91.59%), followed by dyspareunia (52.10%), dyschezia (28.57%), and chronic pelvic pain (10.92%) ( Table I) (Table II).

Accuracy of TRUS considering age categories
The performance of TVUS in the diagnosis of endometriosis considering age categories is summarized in table IV. The accuracy of TVUS in the diagnosis of left endometrioma in women under 35 yr was superior to that of TVUS in women older than 35 yr (93.2% vs. 77.9%; p = 0.04). Similarly, the accuracy of TVUS in the diagnosis of right endometrioma in women under 35 yr was superior to that of TVUS in women older than 35 yr (86.5% vs. 73.3%; p = 0.04). The accuracy of TVUS for detecting endometriotic lesions or nodules in other sites did not differ between the age categories. TVUS in women under 35 yr seemed to be more specific than TVUS in women above 35 yr in terms of right endometrioma (92.1% vs. 52.6%; p < 0.001) and left uterosacral ligaments (82.9% vs. 65.5%; p = 0.02). Also, the LR+ of TVUS for predicting right endometrioma (10.1 vs. 1.98; p < 0.001) and left endometrioma (10.5 vs. 4.35; p < 0.001) among the women younger than 35 yr were significantly higher than in women older than 35 yr (Table IV).

Accuracy of TRUS considering age categories
Although the accuracy of TRUS in the diagnosis of rectosigmoid in women older than 35 yr (76.9%, 95% CI: 65.1, 88.7) was superior to in women younger than 35 yr (68.6%, 95% CI: 54.8, 82.4), there was no statistically significant difference between these (p = 0.32).

Accuracy of TVUS considering BMI categories
The performance of TVUS in the diagnosis of endometriosis considering BMI categories is summarized in table V. The accuracy of TVUS in the diagnosis of right endometrioma (95.6% vs. 75.3%; p < 0.001) in women with a normal BMI was superior to that of TVUS in women with a BMI higher than 30. Also, the LR+ and specificity of TVUS for predicting right endometrioma among women with a normal BMI were superior to those of TVUS in women with a BMI higher than 30. For the detection of left endometrioma in women with a normal BMI, TVUS demonstrated a sensitivity of 96.9%, an NPV of 92.9%, an LR-of 0.03, and a DOR of 215.13, which were significantly superior to TVUS in women with a BMI over 30 (sensitivity: 77.4%, NPV: 58.6%, LR-: 0.25, DOR: 29.1). The accuracy of TVUS for detecting endometriotic lesions or nodules in other sites did not differ between the BMI categories.

Limitations
There were several limitations in this study that need to be discussed. First, the study was performed in a referral center for the treatment of gynecological diseases; therefore, the probability of endometrial lesions in the study population was high, representing a selection bias. So, we cannot extrapolate the findings to the general population of women with clinical suspicion of endometriosis.
Second, the sonographer was aware of the findings of the preoperative clinical examination.
Third, TVUS examinations were performed by an experienced sonographer; therefore, these results may not be repeated by an inexperienced sonographer. Fourth, there was a low frequency of bladder, ureter, and vagina endometriosis, which increased the random error, and it was not possible to calculate diagnostic indicators.

Conclusion
In conclusion, ultrasound can be a useful technique for detecting endometriosis when used adjunctively with the patient's history and physical findings, especially age and BMI.