The "White line sign": A novel sonographic soft marker of pelvic endometriosis

Conceptualization: DJ, PC. Methodology: DJ, PC, ES, PBS. Investigation: DJ, PBS, ES, PC. Analysis: PC, TT. Writing—original draft: PC, TT, DJ. Writing—review and editing: all authors. Supervision: DJ. All authors have read and approved the final version of the manuscript.

This study was approved by the Joint Research Office (JRO) at University College London Hospital (decision date March 6, 2020). The JRO assessed that the scans and laparoscopies being undertaken for the purpose of this study were part of routine clinical care and further approval from the Health Research Authority, the Research Ethics Committee, or Research and Development was waived. Following that, the study was also approved by the Clinical Governance Committee, The Gynaecology Ultrasound Centre, London. The study was registered on the Research Registry website (unique identifying number: researchregistry8314).

During the study period, 418 eligible women were referred to our ultrasound operator for a TVS. Of these women, 318 did not undergo laparoscopy within 3 months of the scan, had a history of hysterectomy, or were unable to have TVS or transrectal ultrasound scan and were, therefore, not included in the study. This resulted in a final study sample of 100 women (Figure  2 ). Flowchart summarizing the inclusion and exclusion of eligible women to the study. DE, deep endometriosis; OE, ovarian endometriosis; SE, superficial endometriosis; TVS, transvaginal ultrasound; TRS, transrectal ultrasound. The indications for initial attendance to the gynecology clinic are shown in Table  1 . The primary presenting complaint comprised of pelvic pain symptoms in 95/100 (95%) of women. Indications for initial clinical visit. Abbreviation: CI, confidence interval. Pelvic pain includes menstrual and non‐menstrual pelvic pain, dyspareunia, dyschezia, and dysuria. At ultrasound, complete obliteration of the POD was noted in 22/100 (22%), partial obliteration in 5/100 (5%) (2 of these patients had filmy adhesions) and no obliteration in 73/100 (73%) of women. The WLS was visualized in 22/73 (30%) women who were deemed to have a free POD at TVS. In our study, we observed a mean thickness of the WLS of 5.13 mm [SD 1.33], with a range of 2.2–13.7 mm. On review of the whole study group, at histology, any type of endometriosis was diagnosed in 98/100 (98%) of patients, of which 98/98 (100%) had SE, 41/98 (41.8%) had DE, and 21/98 (21.4%) had OE. SE only , without DE/OE, was found in 54/98 (55.1%) women. The location of DE nodules is detailed in Table  2 . Of the 22/100 women seen to have the WLS on TVS, all had confirmed endometriosis at histopathology. All 22 (100%) of these women had SE and 11/22 (50%) had pelvic DE. The most common location for DE in these 22 women was the USL, where nodules were present in seven women (Table  2 ). Four women had DE in the adnexa and four had DE in the retrocervical area. There were 3/22 (13.6%) women with laparoscopic evidence of OE. Of the 11 women with the WLS and no evidence of DE or OE on TVS, SE only was found in 9/11 (81.8%) patients, a small DE nodule in one patient and a small endometrioma (<1 cm) in the other patient. Presence of deep endometriotic nodules, diagnosed at laparoscopy, in various anatomical locations, in our cohort of women. Demographic and clinical characteristics, including presence of other gynecological abnormalities diagnosed at laparoscopy, are compared between women with and without the WLS in Tables  3 and 4 . Women who used hormonal contraception showed a significantly lower frequency of the WLS, which was also the only statistically significant difference observed between the groups presenting with and without the WLS. None [0/22 (0%)] of the women with a positive WLS reported using hormonal contraception, whereas 22/51 (43.1%) of women without the WLS were taking it. Comparison of demographic and clinical characteristics in women with and without the White line sign, in those with a free pouch of Douglas. Note : The statistically significant values were presented in bold. Abbreviations: IQR, interquartile range; WLS, White line sign. Presence of gynecological abnormalities diagnosed at laparoscopy in women with and without the White line sign, in those with a free pouch of Douglas. Abbreviation: WLS, White line sign. For the detection of SE, specificity of the WLS was high [100% (95% CI, 2.5–100%)] (Table  5 ). However, the sensitivity and accuracy were low [30.6% (95% CI, 20.2–42.5%) and 31.5% (95% CI, 21.1–43.4%)] respectively. Sensitivity, specificity, positive and negative predictive values, and accuracy of the White line sign at ultrasound, when compared to histology as the reference standard, for the diagnosis of different subtypes of endometriosis. Abbreviations: CI, confidence interval; DE, deep endometriosis; FN, false negative; FP, false positive; NPV, negative predictive value; OE, ovarian endometriosis; POD, pouch of Douglas; PPV, positive predictive value; SE, superficial endometriosis; TN, true negative; TP, true positive; WLS, White line sign. n  = 73, as women who were diagnosed with an obliterated pouch of Douglas at ultrasound were excluded from this analysis. Women with the WLS were more likely to have DE at laparoscopy compared to women without the WLS (11/22 vs. 3/51; p  < 0.05). There was no difference seen in the occurrence of SE between these groups (22/22 vs. 50/51; p  = 1.0), nor for OE (3/22 vs. 5/51; p  = 0.68). There was a high sensitivity, specificity and accuracy of the WLS for the identification of DE [100% (95% CI, 66.4–100%), 79.7% (95% CI, 67.8–88.7%) and 82.2% (95% CI, 71.5–90.2%)] respectively (Table  2 ). For the diagnosis of OE, the specificity and accuracy were again good [71.9% (59.2–82.4%) and 68.1% (56.0–78.6%)], but the sensitivity was lower [37.5% (8.5–75.5%)]. The inter‐rater agreement for diagnosis of the WLS on TVS was high, with a value for proportions of agreement of 0.86 (95% CI 0.69–0.95). The inter‐rater reliability was also good, with a kappa value of 0.72 (95% CI 0.49–0.94). For the intra‐rater agreement, the value for proportions of agreement was 0.83 (95% CI 0.66–0.93) and the kappa value was 0.64 (95% CI 0.38–0.89). This again suggested good agreement and reliability.

In this study, we investigated if a thickened, hyperechoic peritoneum and subperitoneum in the anterior POD could be used as a diagnostic soft marker for SE. The WLS was present in 22/73 (30%) women with a free POD and all women presenting with the WLS had indeed histologically confirmed SE, resulting in a high specificity (100%) of this soft marker. However, the WLS was only observed in a subgroup of women, resulting in a low sensitivity (30.6%) and consequently low to moderate accuracy (31.5%). This may be because not all patients with endometriosis exhibit peritoneal thickening that results in visible ultrasound findings. Alternatively, the WLS could represent a distinct disease phenotype, potentially reflecting a more active and inflammatory form of the condition, whereas in women without the WLS, the disease may be clinically or biologically less inflammatory. This interpretation is supported by the observation that hormonal contraceptive use was significantly more common amongst women without the WLS, compared to those with the WLS. Indeed, none of the women in the WLS positive group reported hormonal contraceptive use. The protective effect of hormonal contraceptives in the development of endometriosis and their anti‐inflammatory properties have been demonstrated in previous studies. 17 We also found that the WLS was a highly sensitive, specific, and accurate sign for the prediction of DE. Two preliminary conclusions could be drawn from this. Firstly, the WLS is easier to detect than DE and therefore it could be used as a simple test to suspect or rule out DE. This could be useful to inexperienced operators who often find it hard to visualize deep disease on ultrasound. Secondly, the WLS could play a possible role in identifying certain endometriosis phenotypes or disease activity, aiding therapeutic decisions. While the WLS and OE showed some correlation, this can be explained by the frequent association of OE and SE, rather than the utility of the WLS in diagnosing OE. The only other study to have investigated the WLS as a predictor of endometriosis, 11 proposed that this corresponded to thickened tissue in the region of the USL and was indicative of USL SE specifically. Due to advancement in ultrasound equipment and techniques since we feel that the WLS can now be more accurately visualized and defined and in fact represents the subperitoneal layer of tissue overlying the anterior POD. A later study also assessed thickened USL as a predictor of SE and reported a higher sensitivity of 62% and specificity of 73% ( p  < 0.05), compared to Robinson et al. (51% and 55% respectively). 18 The latter group hypothesized that the thickening was secondary to mesothelial proliferation, inflammation, and granuloma formation. Multiple studies have shown elevated levels of activated macrophages, cytokines, chemokines, and prostaglandins in the peritoneal fluid of women diagnosed with endometriosis, compared to women without this disease, 17 resulting in inflammation of the surrounding peritoneum. Sonographic evidence of inflammation can have a similar appearance, regardless of the organ affected. 18 A recognized ultrasound feature of many acute and some chronic inflammatory diseases is the aberrant thickening and enhanced brightness of fat and loose areolar connective tissue. 18 The peritoneum consists of an upper layer of epithelial cells, with a lower layer of connective tissue containing adipose tissue, collagen, fibroblasts, inflammatory cells, blood vessels, and nerves. This could explain why a thick hyperechoic layer of peritoneal tissue could often be seen in the anterior POD (WLS) when endometriosis was present in our study, as well as the findings reported by Robinson et al. 11 Chowdary et al. reported a positive correlation between the occurrence of thickened pericolic fat on TVS and the presence of endometriosis ( p  = 0.03). 18 This ultrasound appearance has also been shown to exist in other inflammatory disorders. 19 The role of an inflammatory process in endometriosis could therefore explain these findings and those from our study. The addendum to the consensus opinion by the IDEA group recommended a standardized protocol for the diagnosis of SE using “direct signs.” 10 They defined the sonographic characteristics of these lesions as hyperechogenic projections, hypoechogenic areas, filamentous adhesions, cystic areas or peritoneal pockets, based on descriptions from recent papers by expert ultrasound operators. 8 , 20 However, given the lack of large diagnostic studies, they have advised using this protocol cautiously to minimize misdiagnosis, especially since similar ultrasound appearances can also be found in other scenarios, such as post‐operative adhesions, pelvic inflammatory disease, menstrual clots and peritoneal carcinomatosis. They advised that soft markers for the sonographic diagnosis of SE may not be very useful. 10 Despite this, the use of ovarian immobility and SST is still recommended as part of their systematic approach to the ultrasound assessment of the pelvis in women with suspected endometriosis. 6 Given that the WLS on TVS provided high sensitivity, specificity, and accuracy for the detection of DE and high specificity for diagnosing SE, it could be argued that this could also be a useful adjunct for the detection of all subtypes of endometriosis in women without an obliterated POD. This is particularly important in cases where there are no signs of endometriosis at TVS, but the patient has symptoms highly suggestive of this disease. Although it needs further investigation, the WLS is a simple sign that should be much easier to visualize on TVS than endometriotic nodules. This could be valuable in a general sonographer setting, where small endometriotic lesions may be difficult to identify, but by eliciting this sign, these women can be referred onwards to a specialist endometriosis service. The strengths of our study include the innovative hypothesis, the prospective design, consecutive sample size, and a high quality of ultrasound and laparoscopic examinations with clearly defined diagnostic criteria. By using a single ultrasound examiner, laparoscopic surgeon and pathologist, inter‐operator variability was minimized. Additionally, our study observed good intra‐ and inter‐operator reliability and agreement for the sonographic diagnosis of the WLS, demonstrating the reproducibility of this sign. This is always important before introducing new diagnostic tests into clinical practice. The main limitation of this study is the highly selected cohort, with 98% of women being diagnosed with endometriosis, all of whom had evidence of SE. Consequently, a low false positive rate for the detection of SE is expected, influencing the observed specificity. However, as we were able to largely phase out diagnostic laparoscopies in our clinical practice due to a high standard of sonographic endometriosis diagnosis, negative laparoscopies are rare, and this was unavoidable. Nevertheless, this finding should be interpreted with caution. Also, given the novelty of the sign, the sample size estimation was only based on clinical observation. Lastly, all scans were performed by an EFSUMB level 3 operator; therefore, the findings may not be transferable to less experienced examiners in less specialized settings.

The WLS could be a useful soft marker of both SE and DE when the POD is not obliterated. This would be particularly helpful when there are no other or only subtle signs of endometriosis visible on scan. Also, the WLS might serve as a predictor of active disease. While the absence of the WLS could not reliably rule out the presence of SE, it could for DE, and it performed well in correctly identifying both subtypes. Larger future studies are required, which place a much stronger emphasis on studying a less selected population to avoid such a high prevalence of endometriosis and high pretest probability of disease.

Endometriosis is a multifaceted disease, comprised of different subtypes and varied clinical symptoms. Anatomically, it is described as peritoneal/superficial endometriosis (SE), ovarian endometriotic cysts/endometriomas (OE), and deep endometriosis (DE). 1 DE and OE can be accurately diagnosed using transvaginal ultrasound (TVS), 2 which is now considered comparable to laparoscopy in this regard. 3 However, the diagnosis of SE, which is defined as endometrium‐like tissue involving the peritoneal surface, still relies predominantly on surgery. 3 Although the exact prevalence of SE is unknown, it may affect up to 80% of women with endometriosis, 4 with one study reporting SE to be the only subtype of endometriosis present in 30% of laparoscopically diagnosed cases. 5 This highlights the need for more accurate, non‐invasive methods of diagnosis, especially in women with pelvic pain but no visible DE on TVS. The IDEA group has reached a consensus on how a systematic pelvic ultrasound scan should be performed in women with suspected endometriosis, including assessment for DE, OE, pouch of Douglas (POD) obliteration (negative sliding sign), and soft markers such as ‘site‐specific tenderness’ (SST) and ovarian mobility. 6 Soft markers represent subtle findings, which could indicate presence or absence of disease. 7 Two studies have reported high specificity (96% and 100%) of TVS for the direct diagnosis of SE, but low sensitivity of 50 and 65%. 8 , 9 In a recent addendum to their consensus opinion, the IDEA group has provided a summary of ultrasound findings associated with SE. However, many of the reported features are non‐specific and can also be caused by post‐operative adhesions, chronic pelvic inflammatory disease, and imaging artifacts. 10 One study described a thickening of the white line seen “adjacent to the vaginal wall, in the region of the uterosacral ligaments (USL),” on TVS. 11 They labeled this the White line sign (WLS), and reported a sensitivity and specificity of 51% and 55% respectively for the diagnosis of USL SE. In our clinical practice, in some women with a positive “sliding sign” at TVS, indicating no obliteration of the POD, we observed a thickened white line in this region. This finding appeared to be more common in women with symptoms suspicious of endometriosis. We propose that it may more specifically correspond to the subperitoneal layer of tissue overlying the anterior POD. We postulated that this could indicate thickening and fibrosis of the peritoneum and subperitoneal adipose tissue in the anterior POD and be a soft marker for endometriosis. This could be particularly important in facilitating the detection of SE or more subtle deep and ovarian lesions. Our ultrasound study aimed to assess the diagnostic accuracy of the WLS for the detection of SE in the pelvis.

TT received personal and indirect speaker fees on lectures on ultrasound from GE Healthcare, Samsung, Olympus, Medtronic, and Gedeon Richter. The other authors report no conflicts of interest.

This study was conducted in line with the Standards for Reporting of Diagnostic Accuracy (STARD) 2015 guidelines. 12 This was a prospective diagnostic accuracy study conducted between January 2021 and January 2023. Consecutive patients aged 18–50 years presenting with a history of chronic pelvic pain attended for TVS. They were all examined by a single EFSUMB Level 3 operator (DJ). Those who underwent subsequent laparoscopic surgery were all operated on by the same expert minimally invasive surgeon with interest in pelvic endometriosis (PBS). Exclusion criteria were previous hysterectomy and inability to tolerate TVS. The primary outcome of this study was the diagnostic accuracy of the WLS for predicting the presence of SE. The secondary outcomes were the predictive value of the WLS for the presence of DE and the reproducibility of the WLS at TVS. Histopathology was used as the reference standard for the presence of SE and DE. As per routine clinical practice, a demographic and clinical history was taken from all participants, with data stored in a secure hospital database (ViewPoint, GE Healthcare Technologies, Chicago, Illinois, USA). In particular, age, gravidity, parity, indication for attendance to clinic, and previous history of endometriosis (diagnosed at ultrasound or laparoscopy) were recorded. Two‐ and three‐dimensional TVS assessments were systematically carried out using a 4–9‐MHz probe (Voluson E8; GE Healthcare Technologies, Chicago, Illinois, United States of America), as described below. A transabdominal scan to assess the kidneys was performed using a 2–5‐MHz probe. Endometriosis was diagnosed using the detailed approach specified by the IDEA group [6]. Our previous study describes the features used to diagnose and measure OE, DE, and POD obliteration, as well as other gynecological abnormalities. 13 Obliteration of the POD with adhesions was defined as “none,” “partial,” and “complete.” Filmy adhesions were included in the “partial” category. In addition to this, the POD was also assessed for the WLS, which was described as a thick layer of hyperechoic tissue in the peritoneal and subperitoneal tissue in the anterior aspect of the POD (Figure  1 , Video  S1 ). The WLS was only apparent when the POD was not obliterated with adhesions. When present, the thickness of the white line at the level of the external cervical os was measured and recorded in millimeters (mm). We postulated that the finding of the WLS indicated the presence of SE. Transvaginal ultrasound, B‐mode, illustrating the ‘White line sign.’ (A) Normal pelvic anatomy, without the ‘White line sign.’ (B) A thick layer of hyperechoic tissue (W) is visible in the peritoneal and subperitoneal layers in the anterior aspect of the Pouch of Douglas, indicating presence of the ‘White line sign.’ (C) cervix; P; pouch of Douglas; Ant, anterior; Post, posterior. To assess the intra‐ and inter‐observer reproducibility of the WLS at ultrasound, 36 consecutive ultrasound scans were undertaken and recorded by the EFSUMB Level 3 ultrasound operator (DJ). The recorded videos were reviewed 6 months later by DJ and an EFSUMB Level 2 operator (PC). Both examiners were blinded from the prospectively documented findings by DJ, and from each other's results. Laparoscopy was performed within 3 months after the scan. Assessment of the pelvis was conducted in a systematic fashion for each patient, in line with the recommendations by the World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project. 14 This involved a detailed examination of the anterior and posterior pelvic compartments, all surfaces of the pelvic organs, pelvic side walls, the mid and upper abdomen, using a close‐tip technique, keeping a 2–5 cm distance between the laparoscope and peritoneal surface. As per routine clinical practice and consistent with current recommendations, written, photographic and video documentation of all surgical procedures was undertaken. 14 This included the presence and location of endometriotic lesions, the presence and extent of adhesions in the POD, classified in the same way as it was at ultrasound. The length of the procedure, any complications, and whether multidisciplinary team involvement was required was also noted. SE was differentiated from DE at laparoscopy, according to the revised terminology. 1 Endometriotic lesions involving only the peritoneal surface represented SE, and DE was defined as lesions extending on or under the peritoneal surface that are usually nodular, able to invade adjacent structures and associated with fibrosis and distortion of the normal anatomy. 1 OE was diagnosed at laparoscopy when either invagination or true cysts were seen in the ovary, with the cyst and cyst wall containing dark blood‐stained fluid. 1 Pelvic adhesions were diagnosed when thick or thin/filmy bands of fibrous scar tissue were seen, often adhering the abdominal and pelvic organs to each other. 1 Peritoneal biopsies were obtained from all women. All tissue samples were collected, stored, processed, and examined in the same systematic fashion, in line with the World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project guidelines. 15 The same pathologist, with extensive experience in the examination of endometriotic lesions, was used for all patients. Robinson et al. reported a sensitivity of 51% and specificity of 55% of the WLS in predicting SE of the USL. 11 As there are no other existing studies evaluating the diagnostic performance of the WLS for diagnosing SE in the pelvis, we used these results, as well as our clinical observations, to base our sample size calculation on. We set the sensitivity to 50%, but the specificity at a higher level of 85%, given that our study did not limit the location of SE to the USL. The confidence interval (CI) was set to 95%, expected dropout rate to 0%, and disease prevalence to 50%, given the symptomatic cohort. 3 This gave us a minimum effective sample size for specificity of 98. Assuming an expected κ value of 0.8, power of 80% and alpha of 5%, a minimum sample size of 29 was required to assess reproducibility of the WLS at TVS. The κ value in the null hypothesis was set at 0.3, as agreement between the observers was likely to be better than expected by chance and using a value of zero would therefore be inappropriate. 16 Data distribution was determined using The Kolmogorov–Smirnov test of normality. Data were presented as mean ± standard deviation (SD) if normally distributed or median + interquartile range (IQR) if non‐normally distributed. Categorical variables were compared between groups using the Chi‐squared test, except in the case of rarely occurring characteristics, when Fisher's exact test was applied. Continuous variables were compared between groups using the unpaired t ‐test when data were normally distributed, or the Mann–Whitney U ‐test when they were non‐normally distributed. P ‐values of <0.05 were considered statistically significant. Sensitivity, specificity, and accuracy were used to assess the diagnostic accuracy of TVS in the assessment of the WLS in determining the presence of different subtypes of endometriosis (SE, OE, DE). Histological findings were used as the reference standard. Only the subgroup of women without an obliterated POD were included in the analysis concerning the WLS, as this sign is only relevant when the POD is not obliterated. The intra‐ and inter‐observer reproducibility of the WLS at TVS was evaluated by assessing the agreement and reliability, using proportions of agreement and Cohen's kappa coefficient ( κ ), respectively. We considered a κ value of ≤0.2 to be very poor, 0.21–0.4 poor, 0.41–0.6 moderate, 0.61–0.8 good, and 0.81–1.0 very good. Statistical analysis was performed using Stata version 15.1 (StataCorp., College Station, TX, USA).

Video S1. Video demonstrating a thick layer of hyperechoic tissue in the peritoneal and subperitoneal layers of the anterior aspect of the pouch of Douglas, which is not obliterated with adhesions, as evidenced by a positive sliding sign. This thickening is thought to indicate inflammation caused by endometriosis.