Deep infiltrating endometriosis (DIE) was defined as an endometriotic lesion that infiltrates the peritoneum and penetrates into the retroperitoneal space or the wall of the pelvic organs to a depth of at least 5 mm (Koninckx and Martin, 1992). It occurs in 15–30% of patients with endometriosis (Yantiss et al., 2001). The most common involved location was uterosacral ligament (USL) (Chapron et al., 2003; Bazot et al., 2007). The involvement of USL may cause many clinical symptoms, such as chronic pelvic pain and deep dyspareunia (Hummelshoj et al., 2014). However, there was always a delay between the onset of the first symptoms and the clinical diagnosis of endometriosis and usually the interval was approximately 7–10 years (Matsuzaki et al., 2006; Hudelist et al., 2012) because of the low sensitivity of TVS for USL endometriosis (Bazot et al., 2004; Vimercati et al., 2012; Holland et al., 2013). The latest meta-analysis (Guerriero et al., 2015) for detection of USL endometriosis demonstrated that the overall pooled sensitivity of transvaginal sonography (TVS) was only 53% (95% confidence interval (CI), 35–70%). There were many reasons for the low diagnostic rate, such as the small space of posterior pelvic compartment and its complex structure, the diversity of ultrasound morphological characteristics and the differences of examiners’ experiences. The aims of this study were to assess the value of TVS for diagnosing USL endometriosis performed by an experienced examiner and summarize the ultrasound morphological features of USL endometriosis, so that the examiners could quickly identify the lesion of the USL and improve the diagnostic rate.

Diagnosing USL endometriosis with TVS has always been a difficult point in clinical research. Diagnostic rates varied greatly due to differences in diagnostic methods and diagnostic experience. A recent meta-analysis (Guerriero et al., 2015) demonstrated that the overall pooled sensitivity of TVS diagnosis USL endometriosis was 53%. A series of two studies by Bazot et al. (2003, 2004) showed that the sensitivity and accuracy of TVS to diagnose USL endometriosis were around 70.6–75% and 77–83.8%, respectively. However, in this study, the sensitivity and accuracy of TVS for the diagnosis of USL endometriosis were 95.3 and 94.1%, respectively, which were higher than those reported in most similar studies.

We obtained such good results due to increased awareness of the disease and improved diagnostic methods. In clinical research we have found many factors could lead to low diagnostic rates. Firstly, the retrocervical area was a small and complex anatomical region, the borders of involved organs become indistinguishable in DIE, especially in patients with severe adhesions. Secondly, many examiners were not familiar with the normal ultrasound imaging of USL, which was crucial to improve the diagnostic rate. Some studies (Bazot et al., 2004; Exacoustos et al., 2017) mentioned that the USL was invisible using TVS. However, as early as Ohba et al. (1996) reported that transrectal ultrasonography can be used to observe the USL in non-endometriosis patients and in patients with endometriosis and that the thickness of the USL was associated with clinical symptoms. In our experience, the normal USL can be easily observed in patients with fossa effusion, and the root portion can be seen in a few patients in the absence of fossa effusion. Our experienced examiners can image normal USL using 2D and 3D ultrasound in patients with fossa effusion (Figure 2). Normal USL is visible as a nearly isoechoic arc from the upper cervix extending to the rectum. Lesions can be identified easily if sonographers can master the anatomy of the USL. Thirdly, the examiner’s experience was also an important factor affecting the accuracy. Researchers generally believed (Exacoustos et al., 2017) that the diagnosis of DIE required far more experience, and TVS was highly accurate for the noninvasive diagnosis of DIE in well-trained staff (Guerriero et al., 2015; Tammaa et al., 2015). The examiner of this study was experienced and professionally trained. At last, supplementary methods such as “tenderness-guided” methods and the use of a stand-off to visualize the near-field area (Guerriero et al., 2007) were also important to improve the detection rate. In our experience, the diagnostic accuracy will be lower if examiners do not pay attention to these factors.

In terms of classification of USL lesions, we summerized four types according to ultrasound characteristics. The purpose of our classification was to improve the examiner’s understanding of the disease and to quickly and accurately identify USL lesions. The results demonstrated that types I and II were very common, but type IV was rare among the four types. This study also concluded the detailed distribution of USL involvement according to surgical findings. The most commonly involved locations were bilateral, the left side and the right side in descending order. Similarly, Charles Chapron discovered that DIE was more likely to affect the left USL than the right side (Chapro et al., 2001). However, several previous studies (Jenkins et al., 1986; Fabio and Ginecologica, 1994) failed to find any significant asymmetry in the location of endometriosis involving the USLs. The reasons for the high incidence of bilateral involvement in this study may be that most of the patients had severe endometriosis, and pelvic involvement was more serious. Of course, our conclusions also need support from a larger sample of clinical statistics.

In addition, we noted that the specificity (90.9%) of TVS in diagnosis of USL endometriosis was relatively lower in our study. Bazot et al. (2004) and Guerriero et al. (2015) reported the specificity of 95.9 and 93%, respectively. There may be two reasons for this difference. Firstly, in our statistical analysis, we classified two patients (who were diagnosed unilateral lesion by TVS, but diagnosed as bilateral involvement during surgery) as false negative cases. However, this situation was not elaborated in other studies. Secondly, the other two cases had multiple lesions and severe adhesion in the posterior of the cervix which lead to difficulty in analysis. In this study, there were also three false positive cases. One patient was confirmed as a tumor by histology and another two patients were confirmed as inflammatory lesions. Therefore, it is also important to note that not all USL lesions are endometriosis, and a differential diagnosis should be made to rule out tumors and inflammatory changes (Nascu et al., 2006).

Some studies (Abrao et al., 2007; Bazot et al., 2009) reported that TVS has a lower sensitivity and accuracy for diagnosing USL involvement than MRI. However, our study showed that TVS has great value in the preoperative diagnosis of USL involvement with a sensitivity and accuracy of 95.3 and 94.1%, respectively. In addition, TVS is cost-effective, well accepted and widely available compared with MRI, so TVS should be used as the first line diagnostic technique (Saccardi et al., 2012).

Four types of USL endometriosis were summarized according to the ultrasound characteristics and the distribution of USL involvement was also concluded according to the results of surgery. TVS has significant value in diagnosis of USL endometriosis and can be used as a first-line tool for diagnosis.

The Medical Science Ethics Committee of Shenzhen People’s Hospital approved this study (No. 2018100). Each patient or an appropriate family member provided informed written consent to obtain clinical materials.

YZ and BD conceived and designed the whole experiments and drafted the manuscript. JX performed the statistical analysis and interpreted the data. QL contributed to the clinical examination. XX contributed to the literature research. FX acquired the data. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.