Assessing The Role and Accuracy of Ultrasonographic Imaging in The Diagnosis of Deep Infiltrating Endometriosis: A Cross-Sectional Study

Endometriosis is described as the presence of endometrial tissue in a space outside of the uterus and endometrial cavity. This disease affects almost 10% of women of reproductive age and is usually diagnosed with clinical history as most of the cases complain of chronic pelvic pain ( 1 , 2 ). The average interval between the start of symptoms and surgical diagnosis is 10.4 years ( 3 ). Beyond the clinical symptoms and physical examination, imaging is the modality for the initial assessment of these patients. Imaging techniques currently used to diagnose endometriosis are magnetic resonance imaging (MRI) and ultrasonography with a preference for sonography in recent years ( 4 ). However, the combination of transvaginal sonography (TVS) and MRI is not recommended for a more accurate diagnosis ( 5 ). But still, other causes such as fibroma, corpus luteum, cystadenoma, tubo-ovarian abscess, teratoma, and carcinoma are needed to be ruled out ( 6 - 8 ). Identification of the endometriotic nodules and their correct localization enables complete lesion mapping before surgery and prevents unexpected plan changes in surgery ( 1 , 6 , 9 , 10 ). Deep infiltrating endometriosis (DIE) is recognized as the most severe form of endometriosis has a complex clinical approach; it is described as a lesion that penetrates >5 mm under the peritoneal surface ( 11 ). DIE accounts for 15 to 30% of all endometriosis cases of which 90% are characterized by chronic pelvic pain and infertility, and 25% are accidentally discovered during laparoscopy or laparotomy ( 12 , 13 ). DIE nodules infiltrate mostly the uterosacral ligaments (USL), rectosigmoid, vaginal fornix, rectovaginal septum, and/or bladder ( 14 ). Intestinal endometriosis comprises a spectrum from simple adhesions between the intestine and cervix to nodular lesions that might involve serous membrane to the mucosa. These kinds of severe involvements require simultaneous cooperation between the colorectal and the gynecology surgeons. Due to various diameters and involvement stages, several surgical approaches have been proposed and used ( 1 , 9 ). While smaller, less invasive, lesions are removed using stapled trans-anal resection, the larger and more invasive ones need segmental resection ( 6 , 15 ). A precise consensus on the definition and severity of endometriosis isn’t reached yet but the most frequently used classification is the American Society of Reproductive Medicine (ASRM) classification; however, it fails to completely represent DIE's characteristics ( 16 , 17 ). It’s suggested that TVS is 79% sensitive and 94% specific in the assessment of the extent of DIE ( 2 ) meanwhile, it is proposed that DIE pelvic ultrasonography, which includes rectal and\or vaginal ultrasonographic imaging, is more accurate regarding the extent and severity ( 1 , 2 , 6 , 11 ). We designed and conducted this cross-sectional study to assess the accuracy of DIE ultrasonography and to do so, we compared the results with pathological and surgical findings, particularly with results of rectal involvement. It’s suggested that TVS is 79% sensitive and 94% specific in the assessment of intestinal DIE. In this study, we assessed the accuracy of DIE ultrasonography (rectal and\or vaginal ultrasonography) which is thought to be more accurate.

In total, 150 cases with symptoms were chosen and 109 cases had either DIE or OMA, and 41 were not chosen due to no findings in ultrasonography. As reported by the pathology laboratory, there were 97 cases of pathologically confirmed ovarian endometrioma, 42 cases had intestinal involvement, 56 had uterosacral DIE, 19 cases had uterus adenomyosis and 9 cases were diagnosed with myoma. We also asked the patients to evaluate and score their symptoms from 0 to 10 and on average; The main symptoms that patients complained of were pelvic pain (80.3%), dysmenorrhea (85.3%), dyspareunia (48.6%), dysphasia (43.1%), AUB (29.4%) and infertility (29.4%) respectively ( Fig .1 ). The symptoms were scored as follows; scored as the following dysmenorrhea at 6.74, dyspareunia at 3.36, and dysphasia at 2.72 respectively. Proportion of each symptom felt by patients. In regards to the accuracy of ultrasonography imaging in the diagnosis of intestinal DIE, which was our primary outcome, we found that ultrasonographic imaging performed excellently in overall diagnosis since it was 97.6% sensitive and 73.8% specific. However, laparoscopic evaluation was far more diagnostic (97.6% sensitive and 97.2% specific). As for the levels of involvement in the intestine, we compared the ultrasonographic imaging findings with pathologic results and the results showed lower accuracy; 55.6, 50.0, 66.7% sensitive, and 72.0, 85.6, 91.5% specific for serous membrane, muscular layer and mucus membrane respectively. The average BMI was 24.7 and most of the cases were in the normal range (46.3%) (Tables 1 , 2 , Fig .2 ). Diagnostic accuracy of DIE ultrasonography and laparoscopy in diagnosis of DIE and endometrial lesions DIE US; Ultrasonographic imaging, PPV; Positive predictive value, NPV; Negative predictive value, and OMA; Ovarian endometrioma. Diagnostic accuracy of ultrasonographic imaging in regards to intestinal level of involvement PPV; Positive predictive value, NPV; Negative predictive value, and BMI; Body mass index. ROC curve diagrams showcasing the accuracy of DIE sonography and laparoscopy in regards to intestinal involvement. ROC; Receiver operating characteristic and DIE; Deep infiltrating endometriosis. We also assessed the effect of obesity and weight on US imaging; we compared the results of US imaging of the intestine in 4 BMI brackets as follows: underweight (BMI30). The results showed that both sensitivity and specificity were negatively affected. These results were statistically significant except for the underweight BMI bracket, which we believe was due to the small sample size ( Table 3 ). The effect of body mass composition on accuracy of ultrasonographic imaging PPV; Positive predictive value and NPV; Negative predictive value. We also assessed the accuracy with respect to OMA and the DIE that infiltrates cul de sac and USLs. The results indicated that ultrasonographic imaging was 99.0% sensitive and 84.6% specific. The data of accuracy show that the examination for cul de sac was 100% sensitive, and 50.8% specific while it was 96.4% sensitive and 59.1% specific in assessing USLs. Although imaging is quite sensitive, it can be inaccurate regarding cul de sac and USL assessment since their positive predictive value was 25.0 % and 58% respectively; however, the results for ovarian assessment showed 92.3% PPV ( Table 1 ).

In our study, we identified that overall diagnostic accuracy was 97.6% sensitive and 73.8% specific. However laparoscopic evaluation was found to be far more accurate (97.6% sensitive and 97.2% specific). DIE was also found to lack accuracy in regard to the extent of involvement. It was also not accurate with respect to assessing cul de sac and USL. Imaging has always been an important tool in both the diagnosis and surgical approach to endometriosis. A thorough evaluation can help diagnosis and the entire approach and planning. Thus, it’s of utmost importance that the data pertaining to the lesion is both accurate and reproducible, therefore we aimed to assess DIE ultrasonographic imaging as a complementary and multi-perspective imaging approach. DIE pelvic ultrasonography consists of vaginal and\or rectal US imaging ( 18 ). In a study conducted by S. Alborzi et al. ( 19 ), it was stated that ultrasonographic imaging (transvaginal or transrectal) is as accurate as MRI in the detection of lesions. In our study, the diagnostic accuracy of ultrasonographic imaging in the identification of intestinal lesions, which was our primary outcome, was almost as high as laparoscopic evaluation. Therefore, we suggest that, in the overall diagnosis of DIE in the intestine, this procedure could be useful. In a multicenter prospective and retrospective cohort study conducted in the royal college of obstetrics and gynecology the accuracy of the preoperative ultrasound‐based endometriosis staging system (UBESS) regarding the complexity of surgery was assessed; this study showed that US-based imaging can be utilized to plan the surgery ( 16 ). We also assessed the accuracy of DIE US imaging with respect to other pelvic cavities and sites; in regards to ovarian endometrioma, we concluded that DIE ultrasonography can be a very efficient and accurate tool (99.0% sensitive and 84.6% specific) and as manifested by several other studies such as the study conducted by Holland et al. ( 16 ) can distinguish between different pathologies. Their study showed that TVS is an accurate assessment tool for the severity of pelvic endometriosis and the results are mostly in accord with laparoscopic findings. Meanwhile, we also studied the accuracy of DIE ultrasonographic imaging in the diagnosis of lesions located at USLs and cul de sac and concluded that even though sensitivity for these lesions was high (100% and 96.4% sensitive for cul de sac and USLs respectively) the tests can be inaccurate as their PPV and specificity were low. There were some limitations in our study that reduced the diagnostic accuracy of ultrasonography in DIE patients. We believe that ultrasound imaging accuracy could be hampered as poor bowel preparation can limit ultrasound wave penetration. On the other hand, the procedure itself (TRUS) is painful. These two can both limit the time required for investigation. Another reason that has led to lower accuracy could be the fact that linear nodules could be missed during laparoscopic surgery, particularly in cul de sac. we lacked sufficient samples for specific groups such as the patients with obese body composition. In regards to the body composition of the subjects, we concluded that with higher BMI values the efficacy of US imaging plummets. As described by Bushberg et al. ( 20 ), due to fat impedance, 94% of the original sound wave is attenuated particularly in patients with more than 8 cm of subcutaneous fat before it even reaches the peritoneal cavity; hence, this phenomenon affects the acuity of ultrasonographic imaging.

Our study showed that while DIE pelvic ultrasonographic imaging can be a helpful paraclinical tool in the assessment and diagnosis of DIE and endometriosis in general and particularly with adnexal and bowel lesions, it can have some shortcomings with respect to cul de sac and USLs. We also suggest that in overweight patients these procedures should be performed more meticulously and probably in conjunction with other imaging methods such as MRI.

We designed and conducted this cross-sectional study on patients with severe endometriotic symptoms who were a candidate for laparoscopic surgery and their disease was later confirmed histologically from December 2019 to December 2020. Our patients who were suspected of DIE were assessed in regards to the following characteristics and variables: age, body mass index (BMI) category, confirmed DIE or ovarian endometrioma (OMA), and the respective location and the level of involvement. Our patients were 35.41 years old on average with a standard deviation of 5.94. The symptoms included pelvic pain, dysmenorrhea, dyspareunia, infertility, abnormal uterine bleeding (AUB), and dysphasia. The patients were enrolled from the laparoscopic office of Arash hospital at Tehran university of medical sciences. Our exclusion criteria included the patients who were pregnant, menopausal, or had a non-endometrial mass in adnexa, or other malignancies. We also excluded any patients who had any contraindications from the surgery. The patients who were of reproductive age and had a typical medical history compatible with endometriosis were also assessed, and if their imaging and pathological findings were consistent, they were included in the study. All patients included in the study provided informed consent. In this study we considered pathologically approved surgical results as our gold standard; thus, all our data was compared and tested with surgical findings confirmed by pathology. All patients were assessed by both the attending professor and the fellowship trainees, and all data relating to endometriosis such as pelvic pain, dysmenorrhea, dyspareunia, infertility, and AUB dysphasia were collected. All the features and data gathered from ultrasonographic imaging along with surgical and pathological findings were collected, recorded, and analyzed. The patient’s intestinal involvement was scored from 0 to 3 (0 being no involvement and 3 being full mucosal involvement). Other anatomical sites and areas such as adnexa, cul de sac, USLs, and the salpinx were also assessed and compared. We also gathered general body statistics of the patients and assessed the accuracy using the aforementioned data. Based on the assumption from previous studies that DIE ultrasonography is up to 96% sensitive we calculated that our minimum cases should include 70 patients (Cochrane’s sample size formula). In total, 109 cases were chosen for the study, and the data were analyzed using IBM’s SPSS v26 software (IBM, USA). Our primary goal was to assess the sensitivity, specificity, and positive predictive value of DIE ultrasonographic examination particularly in the extent of intestinal involvement. We also used cross tabulation and chi-square tests to assess the significance of the tests. This study was ethically approved by the Ethical Committee of the Tehran University of Medical Sciences (IR.TUMS.MEDICINE.REC.1399.936) and all patients had signed informed consent forms.