Adenomyosis in women undergoing hysterectomy for abnormal uterine bleeding associated with uterine leiomyomas

Abnormal uterine bleeding (AUB) is a frequently reported symptom for non-pregnant women in their reproductive years. The prevalence, based on health care system databases, suggests that up to one-third will be affected at some time in their lifetime [ 1 , 2 ]. While high, these data likely underestimate the true prevalence of these symptoms. Evidence from survey data suggests one of the AUB symptoms, heavy menstrual bleeding (HMB), may have a point prevalence of as high as 50% [ 3 , 4 ]. Several potential causes or contributors to AUB symptoms have been codified in the system developed by FIGO, known as FIGO AUB System 2, or the PALM-COEIN system, initially published in 2011 [ 5 ] and then revised in 2018 [ 6 ]. Adenomyosis and leiomyomas are two common findings in women with AUB in the reproductive years. Leiomyomas have been reported in as many as 70 to 80% of women by the age of 50 using simple ultrasound techniques [ 7 ]. While at least 50% of affected women are asymptomatic, the overall incidence of AUB associated with leiomyomas (AUB-L) is estimated to be anywhere from 14–25% [ 5 , 8 ] and it has been estimated that most of the roughly 600,000 hysterectomies performed annually in the US are for one or a combination of AUB and leiomyomas [ 9 , 10 ]. It is generally accepted that to cause the symptom of heavy menstrual bleeding (HMB), a leiomyoma should be in contact with the endometrium (submucous or SM) [ 11 ], thereby presenting an opportunity for the tumor’s molecular expressions, such as TGF- β 3, to disrupt local hemostasis [ 12 ]. Adenomyosis is defined as the existence of endometrial glands and stroma in the myometrium, typically accompanied by surrounding myometrial hyperplasia and hypertrophy. While adenomyosis is frequently asymptomatic [ 13 ] it is another potential cause or contributor to AUB symptoms (AUB-A) with an estimated prevalence, based on imaging studies, of 20–35% [ 14 , 15 ]. Available evidence suggests that adenomyosis and leiomyomas are commonly found together in women who undergo hysterectomy, typically for AUB, with a reported prevalence ranging from 15 to 57% [ 16 – 20 ]. There is evidence that those women with adenomyosis, in addition to leiomyomas (AUB-A; -L), may have a disproportionate symptom burden, including subjectively increased menstrual bleeding volume and a greater degree of dysmenorrhea [ 21 , 22 ]. Since both adenomyosis and leiomyomas are frequently asymptomatic, they may coexist with other disorders that are the actual causes or contributors to the AUB symptoms, including coagulopathies (AUB-C), ovulatory dysfunction (AUB-O), and primary endometrial disorders (AUB-E). The advent of various uterine-preserving procedural interventions and the use or introduction of a spectrum of medical interventions challenges clinicians as we enter an era where personalized medicine and shared decision-making are beginning to dominate practice. Consequently, there exists a need to distinguish amongst these various causes or contributors to AUB and other symptoms in a fashion that informs treatment decisions. While pelvic ultrasound has been shown to be highly sensitive and specific for the detection of adenomyosis [ 23 – 25 ], available evidence suggests that when leiomyomas are present, both sensitivity and specificity diminish [ 26 , 27 ]. We designed this retrospective, descriptive study to explore the prevalence of adenomyosis in a group of women who underwent hysterectomy for AUB-L, to determine the prevalence of submucous leiomyomas, and to examine the utility of preoperative ultrasound to detect the presence of adenomyosis. The association of histopathological evidence of adenomyosis in cases without a sonographically defined submucous leiomyoma was also examined.

The hysterectomy database (HD) was searched to identify cases of AUB-L performed on women 18–52 years of age who were in the HMO for at least 3 years before the performance of the procedure. The resulting 1,243 cases were reduced to 400, disproportionately sampling those aged 40–44 years over those 45–52. Of this sample, 14 had no documentation of leiomyomas histopathologically, and 16 had no preoperative ultrasound report in the database ( Fig 1 ). *AUB: Abnormal uterine bleeding. **AUB-L: Abnormal uterine bleeding associated with leiomyomas. The subject selection process is summarized. Ultimately, 400 records were obtained for review; after evaluation, 30 were excluded for the absence of fibroids on the pathology report [ 14 ] or the pelvic ultrasound report (n = 16), leaving 370 available for analysis. The demographic features of this cohort and the subgroup evaluated by Radiologist 2 are displayed in Table 1 . Almost half of the subjects were between 40 and 44, and the mean age was 41.9 years. Notably, nearly 50% of the participants identified as Hispanic, almost a quarter were Black, and just under 20% were White. The mean BMI was 31.4, with 51.8% of these subjects categorized as obese Class 1, 2 or 3. While the inclusion criteria defined the population to include those with an AUB diagnostic code at least 30 days before hysterectomy, a thorough chart audit showed that 277/370 (74.9%) of the cohort had AUB diagnostic codes entered at least 180 days before the surgical intervention. Table 2 demonstrates the pre-hysterectomy medical and surgical interventions reported for the overall cohort and those with and without adenomyosis. The most common medical interventions were gonadotropin-releasing hormone analogs (36.8%), NSAIDS (62.4%), and progestin therapy, implantable (1.4%), oral (26.2%), or intrauterine (31.6%). Given that D&C and sterilization are not considered therapeutic for AUB, procedural interventions included laparoscopy (15.9%), hysteroscopic myomectomy (9.5%), and myomectomy performed via an abdominal approach (6.8%). There were no differences in the frequency of any pre-hysterectomy intervention between those women with and without histopathological evidence of adenomyosis. 1 Fisher Exact p-value; Adenomyosis was identified in 170 of the 370 evaluable cases for a prevalence of 45.9% ( Table 1 ). There was no difference in the histopathological prevalence of adenomyosis based on the presence or absence of a submucous myoma ( Table 3 ). The presence of endometriosis was documented based either on histopathology or the description in the operative report. Endometriosis was present in 13.5% of the cohort, with no significant differences between those with and without a submucous leiomyoma. 1 Based on FIGO Level 2 criteria (submucous or “other”). A review of the imaging used by the HMO’s clinicians demonstrated that MRI was uncommonly employed, whereas pelvic ultrasound was performed in almost all of the subjects identified in the database. Consequently, the radiologists evaluated pelvic ultrasound images to determine the relevant features of the leiomyomas in this cohort of women. Radiologist 2 reviewed the pelvic ultrasound images from a subgroup of 155 records of subjects with a similar demographic distribution to the overall cohort and a similar histopathological prevalence of adenomyosis ( Table 1 ). Of these, 16 had no identifiable ultrasound images, making 139 evaluable for subgroup analysis (Table 5). A total of 36, or 25.9% of the ultrasound studies, were designated “can’t determine” by the radiologist. This left 103 subjects for which the reviewer determined the presence or absence of sonographic features suggestive of adenomyosis ( Table 4 ). The categories “no” and “unlikely” were conflated to indicate that such features were not present, while ‘likely” suggested the presence of adenomyosis. Using these categories, the sensitivity and specificity were 47.4% and 50.0%, respectively, while the positive and negative predictive values were 54.0 and 43.4%. 1 Only 74.1% of these images were evaluable by the radiologist for the presence of adenomyosis, so the utility of these ultrasound images was even less than reflected in the table data. Table 5 demonstrates the relationship between adenomyosis and a preoperative complaint of either AUB only or AUB and chronic pelvic pain. The chronic pelvic pain elements comprised one or a combination of dysmenorrhea, dyspareunia, and acyclic pelvic pain or pressure. Those women with pain and AUB were no more likely to have adenomyosis identified in their hysterectomy specimens. 1 Comparison of the diagnosis of adenomyosis in women presenting with AUB only and AUB with chronic pelvic pain.

This work is consistent with the results of others demonstrating that adenomyosis is frequently found in women deciding to undergo hysterectomy for AUB-L. However, the role of adenomyosis in the generation of AUB symptoms, including HMB, is unclear from these data. Histopathological or imaging-based, well-defined phenotyping for disease burden might identify features more likely to contribute to symptoms. Similar to the findings of other investigators, 2-D pelvic ultrasound appears to be of limited value for the diagnosis of adenomyosis in the presence of leiomyomas: MRI is probably more precise and sensitive but was uncommonly utilized and, consequently, not evaluated in this work. Nevertheless, if there is suspicion that adenomyosis may be contributing to the patient’s symptoms, MRI should be considered as a more appropriate evaluation method in a way that better informs counseling regarding medical or surgical interventions. Furthermore, to be optimally useful, databases such as the HD would benefit from the inclusion of clinical features assimilated in a fashion that facilitates a more complete exploration of the contributors to AUB symptoms. Such an approach could facilitate the research necessary for clinicians to design more personalized treatment strategies.

A retrospective, descriptive, cross-sectional study was designed to query the Hysterectomy Database (HD) populated by a physician entry into the Hysterectomy Registry and associated electronic medical record (EMR) at Kaiser Permanent Southern California (KPSC). In this retrospective data-only study, the research involved minimal risk to participants and involved no procedures for which signed consent is usually required; thus, we received a waiver of written informed consent from the KPSC Institutional Review Board. This study was approved by the Health Maintenance Organization’s (HMO) Institutional Review Board, was conducted in accordance with ethical principles of the current Declaration of Helsinki, and was consistent with the International Conference Harmonization Good Clinical Practice (ICH-GCP) and Good Epidemiology Practices (GEP) and applicable regulatory requirements. The source population comprises members of a large integrated healthcare delivery system that provides comprehensive care for over 4.5 million diverse health plan members across southern California. The primary data source for the overall cohort assembly and characterization was from the HMO’s Hysterectomy Registry/Database EMR form(s) populating the HD, including the EMR chart data. Part of a systemic policy regarding EMR documentation of benign hysterectomy cases, relevant fields captured in each of the registry entries include pre-operative diagnosis (endometriosis, adenomyosis, and uterine fibroids), history and duration of AUB symptoms, and history and duration of pelvic pain. As part of the operative note, findings are easily captured immediately postoperative with checkboxes documenting the surgeon-observed presence of endometriosis and leiomyoma, among other variables. The first step was to identify patients aged 18–52 years, with at least 3 years of registration in the HMO, who underwent hysterectomy between January 1, 2018, and December 31, 2019, with a pre-operative (clinical) diagnosis of chronic AUB associated with uterine leiomyomas (AUB-L). To exclude individuals with acute heavy menstrual bleeding, to be eligible, it was necessary to have at least one AUB diagnosis code in the EMR system 30 days before the hysterectomy. The electronic medical record (EMR) was also searched to determine the proportion of cases with an AUB diagnostic code entered 180 or more days before the hysterectomy. This study data collection and analysis was conducted from January 2020- December 2022. Cases were selected from the registry database if they had a preoperative diagnosis of AUB and leiomyoma and confirmed via chart review documenting the presence of UF along with additional data on fibroid type, location, and size. The cohort of reviewed cases required a post-operative diagnosis of uterine leiomyomas as documented in the related pathology report. Those with a history of gynecologic malignancy, adnexal mass, or symptomatic pelvic relaxation were excluded from the study. A research analyst obtained a stratified random sample of 400 hysterectomy cases from the HD where “leiomyoma” was identified in the preoperative diagnosis field. While random sampling of all age strata was performed and data stratified by age category, it was felt essential to skew the cohort composition such that older women in the later reproductive years were underrepresented. This meant that age strata 40–44 years were randomly selected in a 2:1 radio compared to those aged 45–52 years. All study staff had access to the medical records of the identified cohort from the database for chart abstraction and review of radiology reports. However, during analysis, the data set was aggregated and fully anonymized. Research Associates conducted chart abstraction tasks utilizing REDCap, a secure web application for building and managing online surveys and databases. The purpose of the chart abstraction was to further assess HD data quality by comparing those data housed in the registry with those from the EMR. The chart abstraction process documented the clinical course of patients with a history of AUB-L who underwent a hysterectomy, including diagnosis, initial symptoms, comorbidities, fertility status, and medical and surgical treatments. The chart abstractors collected relevant data for up to 3 years before the hysterectomy procedure date. The EMR chart review included outpatient, inpatient, and emergency data, pharmacy records, and searchable physician notes. The abstractors also documented the preoperative and postoperative diagnoses and radiologic diagnoses, including ultrasound (US), computerized tomography (CT), and magnetic resonance imaging (MRI). Operative notes and pathology reports associated with the hysterectomy were reviewed to confirm the post-operative diagnosis of uterine leiomyomas and identify endometriosis and/or adenomyosis. The imaging reports were reviewed to determine the size and location (anterior, posterior, lateral, fundal, pedunculated) of the largest fibroids up to three in number. Two health plan radiologists unaware of the final post-hysterectomy pathology diagnosis reviewed images (not reports) after the chart abstractors completed their work to identify missing data on the imaging-based location of the three largest fibroids. Radiologist 1 reviewed images from 231 cases that had missing US report data for the location and diameters of fibroids and recorded these metrics in case report forms. Radiologist 2 was given a unique set of records, also with missing data, and reviewed the US images reporting location, FIGO type, and uterine and leiomyoma volume, and specifically determined if and how many submucous fibroids were found regardless of size. The pelvic ultrasound images of this subset were also reviewed for the presence of adenomyosis using the following criteria: Asymmetrical myometrial thickening, indistinct endo-myometrial interface, echogenic linear striations, nodules extending from the endometrium into the myometrium. The reviewer was asked to select from a scale that ranged from absent features correlating to adenomyosis to unlikely to likely with a category for “can’t determine” if the file was unreadable for technical reasons or if the leiomyoma presence obscured interpretation. The characteristics of the samples of the two radiologists were compared using the χ2 test or the Fisher exact test for categorical variables and the Kruskal–Wallis test for continuous variables, as appropriate. The association between adenomyosis present and absent and FIGO type was assessed using the χ2 test for independence. Two-sided P values < 0.05 were considered statistically significant. The sensitivity (percent with radiologically likely adenomyosis who also had histopathologically-determined adenomyosis) and specificity (percent with unlikely or no adenomyosis on imaging who also had histopathological adenomyosis) were calculated to compare the radiologic diagnosis and the gold standard of histopathological examination of the uterus. To further understand these differences, the positive predictive value (PPV) was calculated (percentage with histopathological adenomyosis among those with likely adenomyosis on radiological imaging) as well as the negative predictive value (NPV), the percentage with no histopathological evidence of adenomyosis among those with radiologically unlikely/no adenomyosis. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC).