Visual endometriosis diagnosis is reliable but outpatient tests needed

Schliep and colleagues (BJOG 2016; 124:220–9) have undertaken a pragmatic study to examine the reliability of endometriosis diagnoses by comparing 46 general gynaecologists with four gynaecologists with specific expertise in endometriosis. Information to make a diagnosis was obtained from laparoscopy, operative records and magnetic resonance imaging (MRI) and/or histological samples where available. The four expert assessors had the opportunity to refine their diagnoses according to the information provided to them in sequence. Reassuringly for the general gynaecologist, the authors find that visual diagnosis at laparoscopy is reproducible and that agreement is even higher if operative records and results from peritoneal biopsies are scrutinised. Data for this study represent a random subset taken from the larger ENDO study; a matched exposure cohort to assess the impact of environmental chemicals, lifestyle and medical history on the development of endometriosis (www.nichd.nih.gov/about/org/diphr/od/research/Pages/endometriosis.aspx). Inevitably, as the current study was not primarily designed to evaluate endometriosis diagnosis, the sample evaluated does not provide a complete picture because only a minority of women had histological testing or MRI to augment the diagnostic process. Histological results, where available, led to greater uniformity in diagnosis between gynaecologists and experts. Hence, it appears that clinicians as well as researchers (Wykes et al. BJOG 2004;111:1204–12) place greater weight on pathological tissue samples. However, it is not uncommon in clinical practice to receive conflicting histological results from biopsied areas of visually characteristic endometriotic lesions. As a result, biopsies should not to be considered a panacea; the size—including excision margin—and number of representative biopsies are important as well as the quality of the tissue samples, which will be affected by the method of removal. It is not surprising that MRI did not add to the level of agreement between the gynaecologist and experts given that the majority of the sample had nonsevere endometriotic disease according to the revised American Association of Reproductive Medicine (rASRM) criteria. The utility of MRI has been restricted to widespread and deep infiltrating disease (Medeiros et al. Arch Gynecol Obstet 2015;291:611–21) and its wider role in the diagnostic work up of chronic pelvic pain is currently being assessed in the MEDAL trial (ISRCTN13028601). Staging of endometriosis was also shown to be reproducible overall between assessors but was more variable than diagnosis. This finding seems intuitive given that consensus is lacking and many endometriosis classification systems have been proposed. The authors chose to use the rASRM classification, which is somewhat complex as recognised by the authors of the current study who allowed the expert reviewers to make an initial ‘empiric’ assessment according to the rASRM categories ‘minimal’, ‘mild’, ‘moderate’ and ‘severe’ and then again with recourse to the official rASRM algorithm. The authors conclude that whereas the risk of erroneous laparoscopic diagnosis of endometriosis and estimates of disease severity is small, we should be cognisant that additional review of operative reports and histology may aid ‘sound judgement’. In practical terms, most gynaecologists who are proficient in taking laparoscopic biopsies for minimal to moderate endometriotic disease would do this not only to confirm diagnosis, but also to effect treatment. The role of diagnostic laparoscopy in severe disease is to assess severity and planning of surgery rather than to obtain confirmatory biopsies as presence of endometriosis is self-evident. Hence, it appears that a clinician's judgement would be mostly enhanced if an accurate, noninvasive diagnostic test could be found. The ICMJE disclosure form is available as online supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.