Nonsurgical approaches to the diagnosis and evaluation of endometriosis

Ultrasound and MRI have demonstrated good sensitivity and high specificity in diagnosing advanced-stage endometriosis but often fail to detect early-stage disease. The final two pieces in this Views and Reviews are an in-depth exploration of the literature in this discipline. Despite a 2016 Cochrane review ( 18 ) concluding that there is insufficient evidence that ultrasound can replace laparoscopy for endometriosis diagnosis, except in the case of ovarian endometriomas, transvaginal ultrasound, using specific techniques, can also detect deep infiltrating endometriosis, with sensitivity enhanced by a bowel prep ( 19 – 23 ). For rectosigmoid endometriosis ultrasound has a sensitivity of 80–91% and a specificity of 94–97%. A similarly high sensitivity and specificity is seen for ovarian endometriomas. Sensitivity for detection of other deep endometriosis lesions (e.g. bladder, uterosacral ligaments, or vagina is a bit lower, but specificity remains high. Thus, ultrasound is a useful tool for advanced stage endometriosis, but its ability to detect earlier stage endometriosis remains poor. Ultrasound remains useful in many cases for surgical planning and as part of the diagnostic evaluation. Furthermore, the “sliding sign” and other maneuvers during transvaginal ultrasound can be useful to detect the presence of significant adhesive disease even in the absence of DIE ( 24 ). Ultrasound innovations, such as targeted nanoparticles have been used to diagnose and treat endometriosis in a mouse model ( 25 ), but the physics of ultrasound and the modest specificity do not currently allow this approach in larger animals, including humans. MRI appears to have a similar sensitivity and specificity to ultrasound (especially 3-D ultrasound) in the detection of deep infiltrating endometriosis ( 26 ). Our group has been studying the use of 18-fluorine labeled Positron Emission Tomography (PET) probes simultaneously with MRI or CT in both women and rhesus macaques. The rhesus studies do not show sufficient sensitivity for early stage disease ( 27 ) using 16α-[18F]fluoroestradiol (FES) and the human studies using this probe have not yet been published, but build on earlier case reports ( 28 ) that did not include MRI. Our preliminary conclusion is that FES is insufficiently sensitive, and we are working on the use of a progestin-based probe, 21-[18F]fluoro-furanyl-nor-progesterone (FFNP), with higher affinity and less background in humans. It is our hope that these or other imaging innovations will allow detection and assessment of peritoneal endometriosis lesion.

In 2016, three systematic reviews on diagnostic, non-surgical biomarkers for endometriosis were published by Cochrane Reviews, focused on urinary, blood, and endometrial tissue analytes ( 29 – 31 ) as well as a separate review of combined non-surgical modalities ( 32 ). The review of 8 studies of 6 urinary biomarkers ( 30 ) concluded that no urinary biomarker had been well enough tested to recommend as a triage test or to replace laparoscopy. The review authors they noted that one study ( 33 ) found five peptide peaks using Liquid Chromotography plus Tandem Mass Spectrometry that came close (in this single study) for meeting criteria as a triage test and as a replacement for laparoscopy. The lack of additional papers from that group or using these methods suggests that they were not as robust as hoped. The review of 141 studies of 122 blood markers ( 31 ) also did not find that any biomarker could be shown to be clinically reliable enough to act as a triage test or as a replacement for laparoscopy. The next piece in this Views and Reviews series focuses on blood biomarkers. Briefly, only four of the biomarkers (anti-endometrial autoantibodies, interleukin-6 (IL-6), CA-19.9 and CA-125) had enough studies to provide a meaningful assessment of test accuracy. Thus, many of remaining biomarkers could be sufficient for a triage or replacement test if further studies are consistent. As of 2020, eighteen studies had found 63 different miRNAs that were differentially expressed in the blood of women with endometriosis versus controls. Of these 63, however, differential expression of only 14 miRNA species were shared in one or more studies. Some studies using multiple miRNA species for a diagnostic test shows significant promise ( 34 ) and others a concerning lack of specificity ( 35 ). However, considerable commercial enthusiasm has developed for the potential for one or more miRNA species to be used in a blood test, with multiple companies testing the efficacy of such markers and posting websites promising future clinical tests. One of these companies, Dot Labs, is currently conducting a multicenter trial ( ClinicalTrials.gov NCT04598698 ), presumably based on the promising initial studies ( 34 ) using 6 miRNA species, showing an ROC area under the curve of 0.939. The use of salivary miRNA-based (and piRNA) testing is also being explored, but it is too early to determine whether this could be of possible clinical use ( 36 , 37 ). Endometrial tissue biomarkers would seem to be the most likely to be useful for endometriosis diagnosis, since the tissue is similar to and likely the source of most endometriosis lesions and since they are likely exposed to endocrine and paracrine factors, such as cytokines, produced by nearby lesions in higher amounts than the rest of the body. The Cochrane review of 54 studies of endometrial biomarkers ( 29 ) concluded that only two of the reported endometrial biomarkers (PGP 9.5 and CYP19) had sufficient data to be evaluated. Of these, PGP 9.5, a nerve fiber marker appeared to meet criteria to replace laparoscopy for diagnosis after discarding an outlier study with a sensitivity of 0.96 and a specificity of 0.86, while CYP19 did not meet the triage or replacement test criteria with a sensitivity of 0.77 and a specificity of 0.74. The remaining data was insufficient to subject to meta-analysis, but biomarkers identified 17EHSD2, IL-1R2, caldesmon, and neural markers VIP, CGRP, SP, NPY were deemed of interest for further study. Interestingly, further work on PGP 9.5 demonstrated a sensitivity of 55.4% and a specificity of 92.7%, and the sensitivity of the test improved to approximately 68% if deep endometrial biopsy was performed using a uterine curette. These estimated test characteristics are not sufficient for a replacement or triage test, but more study is needed. The first clinically used endometrial biomarker for endometriosis was absent mid-secretory endometrial immunostaining for the beta3 integrin subunit. The test, formerly marketed in the U.S. as “etegrity” was based on the fundamental discovery of marked and rapid alterations of integrin subunit expression over the menstrual cycle ( 38 ). Expression of the beta3 integrin subunit was correlated with IVF success ( 39 ) and studies in animal models showed likely functional relevance of beta3 integrin during embryo implantation ( 40 , 41 ). The finding that lack of expression was also tightly correlated with a diagnosis of endometriosis Lack of expression was also tightly correlated with a diagnosis of endometriosis ( 42 ), encouraged further clinical use. This finding was not replicated in a much smaller study using similar methods ( 43 ). Furthermore, the patients in original study would have been classified as having unexplained infertility had not endometriosis been identified, and it remains unclear whether the immunostaining reflects the infertility or the endometriosis. Notably, beta3 integrin immunostaning is highly correlated with epithelial histological changes and a lag in those changes (sometimes referred to as out of phase) will result in negative staining. Since an epithelial lag is common in normal women ( 44 ), the clinical utility of beta3 integrin testing is minimized.The relationships between endometriosis, endometrial receptivity, and beta3 integrin continue to be studied ( 45 ), but clinical application of beta3 integrin as an individual biomarker remains of very limited use. A more recent endometrial test is immunostaining for BCL6 protein (please note my conflict of interest in this area). Early evidence suggested a very high sensitivity and specificity for using BCL6 immunostaining as an endometriosis biomarker ( 46 ). Independent clinical evidence ( 47 , 48 ) confirmed increased staining in women with endometriosis. Furthermore, a similar increase in BCL6 was seen after endometriosis induction in a baboon model ( 49 ), bolstering the hypothesis of a tight association between the disease and the test positivity. Further interest was garnered by observational studies in women correlating pregnancy rate in IVF-ET with BCL6 immunostaining ( 50 ) and demonstrating an increased IVF success when BCL-6 positive patients were treated for endometriosis prior to embryo transfer ( 51 ). Another study, using a different antibody, a different detection method, a different patient population, and a cutoff value with unclear rationale did not confirm BCL6 as a predictor of embryo implantation ( 52 ). However, despite methodological differences, these data clearly indicate the need for more research to validate the characteristics of the commercially-available test in different patient populations ( 53 ). The identification of a partner protein for BCL6, SIRT1, whose overexpression in women with endometriosis largely mirrors that of BCL6 and which has been shown to interact directly with BCL6 has added a potential for SIRT1 immunostaining to improve the test characteristics of BCL6 immunostaining, especially given the independent verification of SIRT1 overexpression in women with endometriosis ( 49 , 54 , 55 ). The clinical usefulness of BCL6 and SIRT1 immunostaining remain to be validated, but present an exciting possibility for both endometriosis diagnosis and evaluation of uterine receptivity.

Despite much work in the area, a non-surgical diagnostic technique of early-stage endometriosis remains to be developed or at least fully validated. Questionnaires are unlikely to be helpful for the REI physician. Although imaging-based diagnosis is robust for deep infiltrating disease and ovarian endometriomas, novel imaging techniques such as those in ongoing studies of PET probes and MRI techniques will need to be tested. In the next five years, developing information about blood miRNA and endometrial tissue BCL6 and/or SIRT1 immunostaining should help to better understand the usefulness of these tests. Other tissue markers, including 17EHSD2, IL-1R2, caldesmon, and neural markers VIP, CGRP, SP, NPY, are likely deserving of further study. However, the lack of publications about these biomarkers over the last 8 or more years and the difficulty of publishing “negative data” suggests the possibility that these biomarkers failed further validation. It is also possible that novel combinations of biomarkers, potentially aided by machine-learning algorithms ( 56 ), will improve their diagnostic ability. We must also contemplate the possibility that failed validation of so many prior candidate biomarkers and diagnostic techniques ( 31 ) reflects substantial disease heterogeneity which prevents the use of a universal set of biomarkers. It is hoped that ongoing international efforts to better describe and standardize endometriosis subtypes ( 57 ) will allow a better understanding of this heterogeneity and allow development of subtype specific testing.

For clinicians who seldom see endometriosis and for patients searching for a possible explanation of their symptoms, questionnaires can be excellent tools to increase the saliency of endometriosis in a differential diagnosis. It is also possible that such a survey could also increase the consideration of endometriosis among REI physicians. Thirty years ago, Forman et al. ( 14 ) examined the usefulness of a simple 7-point questionnaire in patients with 2 years of infertility who were already scheduled for a diagnostic laparoscopy, which was done solely for infertility indications. Forty percent of patients had endometriosis and the most predictive symptom was severe dysmenorrhea, though the symptom survey was not predictive of minimal disease (ASRM stage 1). In the ensuing thirty years, many questionnaires have been developed with the intent to diagnose or screen based on patient-reported symptoms, and these have been extensively described in a scoping review ( 15 ). The reviewed studies further reinforced the association of known symptoms, especially dysmenorrhea, with presence of lesions at laparoscopy. However, as stated in that review, there are no “…fully validated, symptom-based, patient-reported questionnaires for endometriosis screening in adult women.” Importantly, none of these surveys would serve to identify a higher risk of the disease in patients whose only symptom is infertility. Subsequent retrospective or population-based studies have used newer techniques to model the impact of symptoms and historical features ( 16 , 17 ), but these studies depend on a physician diagnosis of endometriosis or not rather than a standard laparoscopic or other clinical evaluation for all subjects. The lack of universal laparoscopy in these studies, then, results in bias in that diagnosis would be made only in people with symptoms that would prompt surgical evaluation by physicians. While it is theoretically possible that otherwise unappreciated signs or symptoms might be useful for patients with a clinically subtle presentation, the current role of such questionnaires for the REI physician would seem very limited.