Review question
How accurate are imaging tests in detecting endometriosis? Can any imaging test be accurate enough to replace or reduce the need for surgery in the diagnosis of endometriosis?
Background
Women with endometriosis have endometrial tissue (the tissue that lines the womb and is shed during menstruation) growing outside the womb within the pelvis, causing chronic abdominal pain and difficulty conceiving. Currently, the only reliable way of diagnosing endometriosis is to perform laparoscopic surgery and visualise the endometrial deposits inside the abdomen. Because surgery is risky and expensive, imaging tests have been assessed for their ability to detect endometriosis non-invasively. An accurate imaging test could lead to the diagnosis of endometriosis without the need for surgery, or it could reduce the need for surgery, so only women who were most likely to have endometriosis would require it. Furthermore, if imaging tests could accurately predict the location of endometriotic lesions, surgeons would have the information they need to plan and improve their surgical approach. Other non-invasive ways of diagnosing endometriosis by using urine, blood and endometrial and combination tests have been evaluated in separate Cochrane reviews from this series.
Study characteristics
Evidence included in this review is current to April 2015. We included 49 studies involving 4807 participants. Thirteen studies evaluated pelvic endometriosis, 10 studies ovarian endometrioma, 15 studies deep endometriosis (endometriosis deeply situated in tissues in the pelvis) and 33 studies endometriosis at specific sites within the pelvic cavity. All studies included women of reproductive age who were undergoing diagnostic surgery because they had symptoms of endometriosis.
Key results
None of the imaging methods was accurate enough to provide this information on overall pelvic endometriosis. Transvaginal ultrasound identified ovarian endometriosis with enough accuracy to help surgeons determine whether surgery was needed, and magnetic resonance imaging (MRI) was sufficiently accurate to replace surgery in diagnosing endometrioma but was evaluated in only a small number of studies. Other imaging tests were assessed in small individual studies and could not be evaluated in a meaningful way. Transvaginal ultrasound could be used to locate more anatomical sites of deep endometriosis when compared with MRI, helping surgeons better plan an operative procedure. Endometriosis in the lower bowel appears to be relatively accurately identified by both transvaginal and transrectal ultrasound, by MRI and by multi-detector computerised tomography enema. New types of ultrasound and MRI show a lot of promise in detecting endometriosis but studies are too few to clearly show their diagnostic value.
Quality of the evidence
Generally the studies were of low methodological quality, and most imaging techniques were assessed by only a small number of studies. Differences between studies involved how they were run, groups of women studied, ways imaging tests were performed and how surgery was undertaken.
Future research
Additional high-quality research is needed to accurately evaluate the diagnostic potential of non-invasive imaging tests for endometriosis.
None of the evaluated imaging modalities were able to detect overall pelvic endometriosis with enough accuracy that they would be suggested to replace surgery. Specifically for endometrioma, TVUS qualified as a SpPin triage test. MRI displayed sufficient accuracy to suggest utility as a replacement test, but the data were too scant to permit meaningful conclusions. TVUS could be used clinically to identify additional anatomical sites of DIE compared with MRI, thus facilitating preoperative planning. Rectosigmoid endometriosis was the only site that could be accurately mapped by using TVUS, TRUS, MRI or MDCT-e. Studies evaluating recent advances in imaging modalities such as TVUS-BP, RWC-TVS, 3.0TMRI and MDCT-e were observed to have high diagnostic accuracies but were too few to allow prudent evaluation of their diagnostic role. In view of the low quality of most of the included studies, the findings of this review should be interpreted with caution. Future well-designed diagnostic studies undertaken to compare imaging tests for diagnostic test accuracy and costs are recommended.
About 10% of women of reproductive age suffer from endometriosis. Endometriosis is a costly chronic disease that causes pelvic pain and subfertility. Laparoscopy, the gold standard diagnostic test for endometriosis, is expensive and carries surgical risks. Currently, no non-invasive tests that can be used to accurately diagnose endometriosis are available in clinical practice. This is the first review of diagnostic test accuracy of imaging tests for endometriosis that uses Cochrane methods to provide an update on the rapidly expanding literature in this field.
• To provide estimates of the diagnostic accuracy of imaging modalities for the diagnosis of pelvic endometriosis, ovarian endometriosis and deeply infiltrating endometriosis (DIE) versus surgical diagnosis as a reference standard.
• To describe performance of imaging tests for mapping of deep endometriotic lesions in the pelvis at specific anatomical sites.
Imaging tests were evaluated as replacement tests for diagnostic surgery and as triage tests that would assist decision making regarding diagnostic surgery for endometriosis.
We searched the following databases to 20 April 2015: MEDLINE, CENTRAL, EMBASE, CINAHL, PsycINFO, Web of Science, LILACS, OAIster, TRIP, ClinicalTrials.gov, MEDION, DARE, and PubMed. Searches were not restricted to a particular study design or language nor to specific publication dates. The search strategy incorporated words in the title, abstracts, text words across the record and medical subject headings (MeSH).
We considered published peer-reviewed cross-sectional studies and randomised controlled trials of any size that included prospectively recruited women of reproductive age suspected of having one or more of the following target conditions: endometrioma, pelvic endometriosis, DIE or endometriotic lesions at specific intrapelvic anatomical locations. We included studies that compared the diagnostic test accuracy of one or more imaging modalities versus findings of surgical visualisation of endometriotic lesions.
Two review authors independently collected and performed a quality assessment of data from each study. For each imaging test, data were classified as positive or negative for surgical detection of endometriosis, and sensitivity and specificity estimates were calculated. If two or more tests were evaluated in the same cohort, each was considered as a separate data set. We used the bivariate model to obtain pooled estimates of sensitivity and specificity when sufficient data sets were available. Predetermined criteria for a clinically useful imaging test to replace diagnostic surgery included sensitivity ≥ 94% and specificity ≥ 79%. Criteria for triage tests were set at sensitivity ≥ 95% and specificity ≥ 50%, ruling out the diagnosis with a negative result (SnNout test - if sensitivity is high, a negative test rules out pathology) or at sensitivity ≥ 50% with specificity ≥ 95%, ruling in the diagnosis with a positive result (SpPin test - if specificity is high, a positive test rules in pathology).
We included 49 studies involving 4807 women: 13 studies evaluated pelvic endometriosis, 10 endometriomas and 15 DIE, and 33 studies addressed endometriosis at specific anatomical sites. Most studies were of poor methodological quality. The most studied modalities were transvaginal ultrasound (TVUS) and magnetic resonance imaging (MRI), with outcome measures commonly demonstrating diversity in diagnostic estimates; however, sources of heterogeneity could not be reliably determined. No imaging test met the criteria for a replacement or triage test for detecting pelvic endometriosis, albeit TVUS approached the criteria for a SpPin triage test. For endometrioma, TVUS (eight studies, 765 participants; sensitivity 0.93 (95% confidence interval (CI) 0.87, 0.99), specificity 0.96 (95% CI 0.92, 0.99)) qualified as a SpPin triage test and approached the criteria for a replacement and SnNout triage test, whereas MRI (three studies, 179 participants; sensitivity 0.95 (95% CI 0.90, 1.00), specificity 0.91 (95% CI 0.86, 0.97)) met the criteria for a replacement and SnNout triage test and approached the criteria for a SpPin test. For DIE, TVUS (nine studies, 12 data sets, 934 participants; sensitivity 0.79 (95% CI 0.69, 0.89) and specificity 0.94 (95% CI 0.88, 1.00)) approached the criteria for a SpPin triage test, and MRI (six studies, seven data sets, 266 participants; sensitivity 0.94 (95% CI 0.90, 0.97), specificity 0.77 (95% CI 0.44, 1.00)) approached the criteria for a replacement and SnNout triage test. Other imaging tests assessed in small individual studies could not be statistically evaluated.
TVUS met the criteria for a SpPin triage test in mapping DIE to uterosacral ligaments, rectovaginal septum, vaginal wall, pouch of Douglas (POD) and rectosigmoid. MRI met the criteria for a SpPin triage test for POD and vaginal and rectosigmoid endometriosis. Transrectal ultrasonography (TRUS) might qualify as a SpPin triage test for rectosigmoid involvement but could not be adequately assessed for other anatomical sites because heterogeneous data were scant. Multi-detector computerised tomography enema (MDCT-e) displayed the highest diagnostic performance for rectosigmoid and other bowel endometriosis and met the criteria for both SpPin and SnNout triage tests, but studies were too few to provide meaningful results.
Diagnostic accuracies were higher for TVUS with bowel preparation (TVUS-BP) and rectal water contrast (RWC-TVS) and for 3.0TMRI than for conventional methods, although the paucity of studies precluded statistical evaluation.