Endometriosis is a disease characterized by the presence of endometrium-like tissue outside the endometrium and myometrium, usually accompanied by inflammatory changes (1). Endometriosis can affect various structures: The peritoneum, the ovaries, the intestinal wall, the urinary bladder, or extra-abdominal structures. Deep infiltrating endometriosis (DIE) is a subtype of endometriosis, characterized by the presence of endometrial-like tissue in the abdominal cavity, which spreads on or under the peritoneal surface and can infiltrate adjacent organs (1).

Endometriosis can cause various symptoms, including chronic pelvic pain, dyspareunia, fatigue, and infertility (2). About 10%−15% of women of childbearing age and 35%−50% of women with pelvic pain and/or infertility are affected by endometriosis (3). Laparoscopy has traditionally been the method of choice for diagnosing endometriosis. Recently, however, there has been a growing body of research highlighting the value of imaging (MRI and transvaginal ultrasound) in the diagnosis of endometriosis (4–6). The recommendation of imaging in current guidelines (2) and the advantages of MRI (large examination field, non-invasiveness, standardization, little operator dependency) give reason to expect an increasingly broad application (4).

As with many other radiological examinations, the description and interpretation of incidental findings (IFs) on MRI for endometriosis can present a challenge. No data on IFs in endometriosis MRI are currently available (level of evidence: n/a) (7). IFs are defined as findings beyond the primary clinical indication of a study and may be clinically relevant but do not necessarily have to be (8–10). IFs include both insignificant marginal findings and false positive findings. IFs can lead to uncertainty among radiologists, referring physicians, and patients. The main reasons for this are a lack of information about the frequency and relevance of IFs and difficulties in differentiating relevant findings from physiological changes and normal variants. The radiological reporting of IFs may lead to further diagnostic examinations and medical interventions. These additional measures can be helpful and potentially life-saving but can also be unnecessary, costly, and risky. Therefore, an adequate strategy for the disclosure of IFs must be chosen. In order to develop such a strategy, however, data on the types and the frequency of findings are required in the first instance. As there is currently a lack of data on IFs in endometriosis MRIs, potential negative effects of this knowledge gap on patient treatment and outcomes are possible.

Challenges in the scientific analysis of IFs are that the patient's history and radiological reports are often available only in non-standardized form and that the interpretation of the images is subject to variability. In addition, a very specific imaging question (e.g., endometriosis) leads to a higher rate of IFs than a broader question (e.g., pelvic pain). A retrospective study of 1,040 abdominal CT scans for different indications revealed relevant IFs (leading to further imaging, clinical evaluation, or follow-up) in 18.8% of patients (11). As the rate of IFs in this study was based on a review of the radiology reports without a review of the images, it can be assumed that the rate of IFs was underestimated (9).

An association between endometriosis and various other diseases has been suspected, including gynecologic diseases, gastrointestinal diseases, immunological-related/autoimmune diseases, cardiovascular diseases, and cerebrovascular diseases (12, 13). This association would give reason to expect an increased rate of IFs in MRI examinations positive for endometriosis. However, no data on IFs in endometriosis MRI are currently available (7).

The objective of the present study was therefore to analyze the prevalence and distribution of IFs identified on pelvic MRI for endometriosis, including overview sequences from the kidneys to the pubic bone. In addition, differences in the frequency of IFs between patients with and without evidence of DIE on MRI and between patients with and without administration of gadolinium based contrast agents (GBCAs) were to be examined.

This study was approved by the institutional review board. Due to the retrospective design of the study, informed consent was waived.

A total of 1771 IFs were noted in the study cohort of 303 patients [mean age, 33.4 years ± 8.3 (standard deviation); median age, 33; Figure 1]. IFs were recorded in 299/303 examinations (98.7%). 11 patients had one IF, 30 patients had two IFs, 31 patients had three IFs, 41 patients had four IFs and 186 patients had five or more IFs. The mean number of IFs per patient was 5.8, and the median number of IFs per patient was 6.

This study analyzed the prevalence and distribution of IFs detected on pelvic MRI for endometriosis, including overview sequences from the kidneys to the pubic bone. Our findings show that IFs with high clinical relevance are common, and IFs with moderate and low clinical relevance are very common with prevalences of 25/303 (8.3%; 95% CI 5.4–11.9%), 134/303 (44.2%; 95% CI 38.5–50.0%), and 298/303 (98.3%; 95% CI 96.2–99.5%), respectively. The most frequent individual IFs were ARO (200/303 patients, 66.0%) and lumbar disc desiccation (146/303 patients, 48.2%). The overall prevalence of IFs, the prevalence of IFs grouped by clinical significance and the mean number of IFs per patient did not differ significantly between patients with and without DIE diagnosis on MRI (p = 0.064 to p = 0.922). For three individual IFs, significantly higher prevalences were found in patients with than in patients without evidence of DIE on MRI (ARO, fissures of the annulus fibrosus, gallstones).

To date, no data are available on the presence of IFs on pelvic MRI for endometriosis. Several recent studies have investigated the frequency of incidental findings in prostate MRI, although the comparability with our results is obviously reduced due to the different patient population. Cutaia et al. found IFs in only 52.7% and Sherrer et al. in only 40.2% of patients on prostate MRI despite the older age of the patients (mean age 67.1 and 63.3 years, respectively) (17, 31). These lower prevalences of IFs are most likely attributable to the smaller field of view of prostate MRI. Consequently, changes of the hip joints and the lumbar spine were not included in these studies. MRI for endometriosis is performed with a larger field of view, so that the entire pelvis and the lower part of the lumbar spine are included in the scans. When sequences are included for an overview of the kidneys and urinary tract, parts of the liver and other upper abdominal organs may also be visible. In the ESUR guideline for the MR imaging of endometriosis, four of the eight participating centers stated that their MRI protocol contains a T2-sequence from the kidney to the pubic bone, and a corresponding recommendation is suggested to enable a systematic visualization of kidneys and potential analysis of the right iliac fossa (16).

Even minor findings can pose difficulties for radiologists and referring physicians in everyday practice and cause uncertainty. Particularly in a young patient population, the question regularly arises as to when findings should be considered pathological, a normal variant and/or be communicated. The most common IF we found was ARO in 200/303 patients (66.0%). As this finding has received little attention to date, no reports are available on the prevalence in non-musculoskeletal pelvic MRI examinations. The importance of this very common finding lies primarily in not interpreting it as pathological, as Valente et al. (23) have pointed out: In their 2021 study, they found ARO in 96% of 75 asymptomatic patients (mean age, 47.7 years). Consequently, the diagnosis of osteoarthritis should not be made solely based on the presence of ARO.

Another very common observation in our study collective were degenerations of the lumbar spine, despite the low average age of the patients. Annular fissures of the intervertebral discs were detected significantly more frequently in patients with than in patients without DIE diagnosis on MRI (p = 0.006). This observation could be explained by the association between lumbar disc degeneration and comorbidities related to systemic inflammation reported by Lambrechts et al. (32), although we could not find significant differences in the prevalence of disc desiccation between patients with and without DIE on MRI (p = 0.344). It is currently unclear whether degenerations of the spine as IFs should be reported by radiologists and communicated to patients. As Brinjikji et al. (33) stated, disc degeneration has a higher prevalence in adults with low back pain than in asymptomatic individuals. On the other hand, routine MRI reports have been found to produce a negative perception and poor functional outcomes in low back pain (34).

In our study cohort, a significantly higher prevalence of three types of IFs was found in patients with DIE diagnosis on MRI (ARO, fissures of the annulus fibrosus, gallstones). An association between endometriosis and various comorbidities has been suspected, including gynecologic diseases (fibroids, adenomyosis, ovarian cancer), gastrointestinal diseases (irritable/inflammatory bowel disease), immunological-related/autoimmune diseases (rheumatoid arthritis, psoriasis, osteoarthritis, asthma, allergy), cardiovascular and cerebrovascular diseases (12, 13). Causal mechanisms are considered to be endometriosis-induced local and systemic inflammation, immune dysregulation, hormonal changes, and treatment sequelae. The results of our study provide a potential indication of an association of the three IFs mentioned above with endometriosis, although no statistically significant differences in the overall prevalence and mean number of IFs (with/without DIE on MRI) and no associations comparable to the age dependence of IFs could be demonstrated (Supplementary Tables S1–S3).

We found a significantly higher mean number of IFs per patient in contrast-enhanced than in non-contrast MRIs (p = 0.002) without significant difference in the overall prevalence of IFs (p = 0.579). However, the decision on GBCA administrations had been made depending on the findings of the non-contrast sequences and the presence of ancillary questions (14). Therefore, the higher rate of IFs must be seen as a reason for the application rather than a consequence of GBCA administrations. DIE had not been found significantly more frequently in contrast-enhanced MRIs, which is consistent with the current ESUR guidelines for endometriosis MRI that do not routinely recommend GBCA administration (16).

Various guidelines for the management of IFs in clinical imaging and research have been established recently (35–48). Due to the extensive use of imaging in modern medicine, there is an ongoing need for standardization of the management of IFs (49, 50). Further assistance for radiologists through artificial intelligence (AI) may be expected in the future (51, 52). Radiologists must nonetheless familiarize themselves with IFs to properly determine consequences and provide guidance (53, 54). Detecting an IF does not necessarily imply that it should be reported. For IFs with moderate or high clinical significance, a description in the radiologic report is warranted, if available, with reference to current guidelines on the management of the findings. Appropriate wording should be used so as not to cause unnecessary further investigations or patient distress. However, the majority of IFs has low clinical significance and a description in the radiology report is often unnecessary and not beneficial to the patient, e.g., in non-pathological findings such as ARO or in age-typical degeneration of the spine.

There are some limitations of our study. The study was conducted retrospectively at a single tertiary care center. Diagnoses were mainly based on the review of the imaging findings and radiological reports. Since not every manifestation of DIE is detectable on MRI, our results comparing the frequency of IFs between patients with and without evidence of DIE on MRI may have somewhat limited generalizability. Further studies to externally validate our results are warranted.

In conclusion, incidental findings are found very commonly on pelvic MRI for endometriosis, including overview sequences from the kidneys to the pubic bone. Many incidental findings have no, only minor, or uncertain consequences. Although less prevalent, radiologists should be aware of findings with high clinical significance.