Prevalence of adenomyosis features in women scheduled for assisted reproductive treatment, using the Morphological Uterus Sonographic Assessment group definitions

Sara Alson, Povilas Sladkevicius, Ligita Jokubkiene and Emir Henic planned the research protocol. Sara Alson performed all the ultrasound examinations, collected and analyzed the data and drafted the first version of the manuscript that was discussed with Ligita Jokubkiene, Emir Henic and Povilas Sladkevicius whereafter appropriate changes were made. Sara Alson wrote the final version of the manuscript.

The study has been approved by the Regional Ethical Review Board of Lund university, Lund, Sweden, on September 11, 2018, with reference number 2018/555. Informed, written consent was obtained from all participants.

This study was supported by regional research grants from Region Skåne, Sweden.

Of 1224 women potentially eligible to participate in the study, 64 women were excluded. The remaining 1160 women were included in the study (Figure  3 ). Flow chart demonstrating the process of patient inclusion to the study. 2D, two‐dimensional; 3D, three‐dimensional; TVUS, transvaginal ultrasonography. Data are presented as the percentages of all women included in the study. The median age of the 1160 women was 32 years (range, 25–39 years). When divided into three age groups, 339 women (29.2%) were 25–29 years, 499 (43.0%) were 30 to 35 years. Demographic characteristics of the women included in the study are shown in Table  1 . Background characteristics of 1160 women, and for women with vs without at least one feature of adenomyosis at 2D and 3D ultrasound examination. Note : Values are given as n (%) of each group of women or as median (range). Abbreviations: 2D, two‐dimensional; 3D, three‐dimensional; ART, assisted reproductive treatment; BMI, body mass index; PCOS, polycystic ovarian syndrome; TVUS, transvaginal ultrasonography. Women with PCOS are included in this category. Same‐gender couples or single woman. The prevalence of at least one feature of adenomyosis at TVUS was 272/1160 women (23.4%, 95% CI 21.0–25.9). Of these, at least one direct feature was found in 111/1160 women (9.6%, 95% CI 7.9–11.3; Table  2 ). The number of direct features in the different age groups was 12 (3.5%) in women aged 25–29 40 (8.0%) in women aged 30–34 and 59 (18.3%) in women aged ≥35 . Prevalence of direct and/or indirect features of adenomyosis at 2D and/or 3D ultrasonography. Note : Numbers are given as n (%) of the total cohort, with 95% CI. Women may have more than one feature of adenomyosis. Abbreviations: 2D, two‐dimensional; 3D, three‐dimensional; CI, confidence interval; JZ, junctional zone. The most common direct features were echogenic subendometrial lines and buds, found in 82 women (7.1%, 95% CI 5.6–8.5) followed by hyperechogenic islands found in 58 women (5.0%, 95% CI 3.8–6.3; Table  2 ). The most common indirect feature was an irregular JZ found in 157 women (13.5%, 95% CI, 11.6–15.5; Table  2 ). An irregular JZ that was not associated with the presence of a direct feature was found in 31 women (2.7%, 95% CI 1.7–3.6). Of women with at least one feature, 64 (57.7%) women had only one direct feature and 109 (40.1%) women only one indirect feature. The distribution of the total number of adenomyosis features in women studied is presented in Table  S1 . The ultrasonographic characteristics of the different features of adenomyosis are presented in Table  3 . Direct features were most commonly diffuse ( n  = 57, 51.4%), had a mild extent ( n  = 34. 30.6%) and were located in the inner to middle myometrium in the fundus. The largest median (range) diameter of the largest lesion was 7.0 (3.0–69) mm. Sonographic characteristics of features of adenomyosis at 2D and/or 3D ultrasonography. Some women may have adenomyosis features present in multiple locations. n  = 229, as the extent was not assessed for 43 women. Abbreviations: 2D, Two‐dimensional; 3D, three‐dimensional. Direct features were visible only at 2D TVUS in seven women (0.6%, 95% CI 0.2–1.1) and only at the coronal plane of 3D TVUS in 56 women (4.8%, 95% CI 3.6–6.1). The frequency of adenomyosis features visible only at 2D or the coronal plane of 3D TVUS, presented as percentages of women with direct or indirect features, is presented in Table  4 . Frequency of features of adenomyosis visible either at 2D and/or 3D coronal plane ultrasonography in 272 women with at least one feature of adenomyosis. Note : Numbers are given as n (%), with 95% CI. Percentages are calculated within each group of women with direct or indirect features 2D, two‐dimensional; 3D, three‐dimensional; CI, confidence interval; TVUS, transvaginal ultrasonography. The JZ was unassessable at 2D TVUS in 220 women (19.0%, 95% CI 16.7–21.2) and at 3D TVUS in 43 women (3.7%, 95% CI, 2.6–4.8). In 33 women (2.8%, 95% CI 2.0–4.0), the JZ was not assessable in either 2D or 3D TVUS. In women with any adenomyosis feature, the JZ was unassessable at 2D TVUS in 89 women (7.7%, 95% CI 6.1–9.2) and at 3D TVUS in 26 women (2.2%, 95% CI, 1.4–3.1). After excluding women with myomas, the median (range) thickness of the anterior myometrial wall was 1.4 (0.6–5.6) cm in women with at least one feature of adenomyosis and 1.3 (0.66–2.3) cm in women without such a feature ( P  = 0.013). The median (range) thickness of the posterior myometrial wall was 1.6 (0.6–4.2) cm in women with at least one feature of adenomyosis and 1.4 (0.6–5.8) cm in women without such a feature ( P  < 0.001). In the study population, ultrasonographic signs of endometriosis were seen in 246 women (21.2%, 95% CI 18.9–23.6). Women with endometriosis were more likely to have at least one feature of adenomyosis ( n  = 95/246, 38.6%) than were women without endometriosis ( n  = 177/914, 19.4%; P  < 0.001) and this was also true for at least one direct feature of adenomyosis, n  = 44/246, (17.9%) and n  = 67/914 (7.3%), respectively ( P  < 0.001). Of women with at least one direct feature of adenomyosis, 44/111 (39.6%) had concomitant endometriosis. In women with at least one adenomyosis feature, a negative sliding sign was found in 54 (19.9%) women, of which 50 (18.4%) had concomitant endometriosis at TVUS. Myomas were seen in 177 women (15.3%, 95% CI 13.2–17.3). In total, 11 myomas (0.9%, 95% CI 0.4–1.5) were seen in women with at least one direct feature of adenomyosis and 35 (3.0, [95% CI 2.0–4.0]) in women with at least one indirect feature of adenomyosis. In total, 16 women with myoma (1.4%, 95% CI 0.7–2.1) had an unassessable JZ due to the lesion. The results of a univariate logistic regression analysis suggested that the presence of at least one direct adenomyosis feature was associated with the presence of endometriosis, age, termination of pregnancy and myoma. The odds of having at least one direct feature of adenomyosis was almost three times higher in women with endometriosis than in women without endometriosis and almost twice as high in women with previous termination of pregnancy than in women without (Table  5 ). The presence of at least one indirect feature was associated with previous childbirth and with the presence of endometriosis (Table  5 ). The multivariate analysis and backwards selection procedure showed that the selected variables remained significantly associated with having at least one direct feature, at least one indirect and/or direct feature or at least one indirect feature without any direct feature of adenomyosis. Univariate binary logistic regression analysis of association between demographic and clinical variables and the presence of direct and/or indirect features of adenomyosis at 2D and 3D ultrasound examination. Note : OR was calculated using binary logistic regression analysis. P  < 0.05 is considered statistically significant. Direct features with indirect features. Indirect features with and without direct features. Abbreviations: CI, confidence intervals; OR, odds ratio. More women with direct features of adenomyosis reported dysmenorrhea ( n  = 747/1049, 72.5%) compared with women without such features ( n  = 95/11, 84.1%; P  = 0.008) (Table  S2 ). Women with direct features reported higher VAS scores concerning dysuria and dyspareunia compared with women without direct features, even if ranges were overlapping (Table  S3 ).

Our study showed that one in 10 women scheduled for their first ART had direct or pathognomonic features of adenomyosis at systematic 2D and 3D TVUS examination. The use of 2D and 3D TVUS combined with standardized diagnostic criteria may improve the diagnostics of the disease and can result in an improved counseling and treatment of these women. Our results, with almost a quarter of women having direct and/or indirect features of adenomyosis, are consistent with previous studies 24 , 25 that reported a prevalence of adenomyosis at 2D and 3D TVUS in subfertile women of 24.2% and 19.2%, respectively. It is also in line with a reported prevalence of adenomyosis of 20.9%, albeit in a symptomatic population attending a general gynecological clinic. 26 However, previous studies often included subfertile women already scheduled for embryo transfer 24 , 25 or used fewer or partly different criteria to diagnose adenomyosis than in the present study. 24 , 25 , 27 Several studies did not specifically examine subfertile women. One study found a prevalence of 34.0% in women 18–30 years referred for TVUS due to symptoms, excluding those with endometriosis or infertility. 28 Another study of 78 symptomatic women found a 55.12% prevalence of adenomyosis. 29 These studies included either the JZ thickness and/or the question mark sign as diagnostic criteria, 28 , 29 neither of which is included in the revised MUSA definitions. 9 In addition, there has previously been a lack of a clear and metric definition of asymmetrical myometrial thickening. Different studies may have used different definitions. If strict, direct criteria are used to diagnose adenomyosis instead of any indirect feature, the true prevalence of adenomyosis may be much lower than previously suggested. Other studies might be overestimating the prevalence when using indirect features for diagnosis. Subendometrial lines and buds was the most common direct feature of adenomyosis in our study, found in 82 (7.1%) women. Subendometrial lines and buds also constitute an irregularity, as defined by MUSA. Recent systematic reviews and meta‐analyses found that subendometrial lines and buds was the most accurate individual sonographic MUSA criterion for adenomyosis, 30 and that an irregular JZ had a good discriminative ability for diagnosing the disease. 11 A previous study found that women with endometriosis more often had an altered or infiltrated JZ compared with women without endometriosis. 31 In accordance with this, the present study found that the odds of having at least one adenomyosis feature was almost three times higher in women with endometriosis. However, the fact that some direct features simultaneously can constitute an indirect feature according to the revised MUSA may lead to a perceived overestimation of the number of indirect features. Moreover, the JZ may be irregular due to fibrosis or appear interrupted due to artifacts, which is not necessarily associated with adenomyosis. Likewise, the presence of translesional vascularity is mostly evaluated to exclude fibroids in the presence of other indirect features such as globular uterus or fan‐shaped shadowing and will therefore only appear in conjunction with these. Women who had undergone termination of pregnancy or childbirth had higher odds of having direct features of adenomyosis compared with women who had not. This finding is consistent with theories suggesting that adenomyosis and endometriosis are associated with trauma to the JZ, followed by local estrogen production, hyperperistalsis and subsequent invasion of endometrial epithelial cells to the JZ and myometrium. 32 , 33 Such trauma may consist of pregnancy or curettage, which can damage the JZ. In half of women with direct adenomyosis features, these were only seen in the coronal view of the JZ at 3D TVUS. Previous studies have suggested that 3D TVUS is more accurate than 2D TVUS in detecting adenomyosis. 10 The combination of 2D and 3D TVUS may improve the diagnostic accuracy, due to the visualization of the JZ using 3D TVUS. 11 , 30 At present, it has not been established whether minor alterations of the JZ only visible at 3D TVUS could represent early stages of adenomyosis that may be related to subfertility, 10 or whether these are clinically relevant regarding ART outcome. A strength of our study is its prospective design, with the inclusion of a large population of subfertile women scheduled for their first ART, which we believe is representative of the population of subfertile women being treated at similar clinics. Our study has some limitations. All examinations were performed by one examiner, which may limit the generalizability of the study. 34 However, most women would probably not have agreed to a second examination just for study purposes. Others have found good inter‐ and intraobserver variability in the assessment of the JZ visualization. 35 Diagnostic criteria were well‐defined before study start and any interobserver variability was excluded using the same examiner for all women. Moreover, one cannot rule out an inherent bias that may preclude an accurate assessment of the difference between 2D and 3D TVUS. If, for example, lines and buds are seen at 2D TVUS, the examiner will probably scrutinize the 3D images for a corresponding finding and vice versa. However, as the purpose of this study was to assess the presence of MUSA features using 2D as well as 3D TVUS, we believe that the method used is sufficient and reflects clinical practice. Histopathological confirmation of our findings is lacking. However, the purpose was to describe features of adenomyosis at TVUS examination. Previous studies have demonstrated high sensitivity and specificity for TVUS compared with hysterectomy with histopathology 11 , 26 , 36 in diagnosing the disease. Another potential limitation is not including women aged 40 years or older, who may demand ART as fertility decreases markedly at this age. 37 However, the requirements for eligibility for publicly funded ART in Sweden are non‐negotiable, which is why all women referred to RMC for ART will fulfill these criteria. By not including women >40 years, some age‐related factors that may present a bias are excluded. We chose not to exclude women with myomas. Adenomyosis is believed to be more common in these women, yet diagnosing adenomyosis in this group is not considered more difficult. 26 Excluding them may have introduced a bias. Different phenotypes of adenomyosis may be associated with different symptoms. 38 The ultrasound features in women primarily presenting with, for example, pain or abnormal uterine bleeding may differ from those presenting with subfertility. Therefore, our findings may not be generalizable to other populations. Diagnosing adenomyosis in women scheduled for ART is desirable to counsel women adequately on their chances of a successful ART and to discuss different therapeutic options. To improve ART and obstetrical outcomes 4 , 5 in women suffering from adenomyosis, previous studies have suggested a prolonged pituitary downregulation and subsequent frozen embryo transfer, 4 , 24 , 39 as well as closer monitoring of their pregnancies. 40 However, different features of adenomyosis may have different clinical significance. At present, it has not been established which sonographic features may confer an increased risk of an adverse ART or obstetrical outcome. Knowledge of the prevalence and characteristics of different adenomyosis features is a prerequisite for further studies on this subject.

This study found that one in 10 women scheduled for their first ART have direct or pathognomonic signs of adenomyosis at 2D and 3D TVUS. Systematic 2D and 3D TVUS examinations, using standardized criteria for diagnosing adenomyosis, should be considered in daily practice. An improved understanding of the disease spectrum and ultrasonographic appearance may lead to an enhanced counseling and individualized care of women with different features of adenomyosis. Future studies will reveal the clinical relevance of our findings regarding ART and pregnancy outcomes.

Adenomyosis is a benign, estrogen‐dependent disease, characterized by ectopic endometrial glands and stroma within the myometrium. 1 Data suggest that adenomyosis may have a negative impact on fertility 2 , 3 and assisted reproductive treatment (ART) outcome, 4 , 5 but evidence is conflicting. 4 , 5 , 6 More data regarding the prevalence and characteristics of adenomyosis, as diagnosed by noninvasive methods in subfertile women, is needed. 7 , 8 Transvaginal ultrasound examination (TVUS) is the first‐line noninvasive method to diagnose adenomyosis. 9 Further, the use of three‐dimensional (3D) ultrasound has enabled an enhanced visualization of the junctional zone (JZ), which may improve the diagnostic accuracy. 10 An improved agreement between different diagnostic modalities without the need for hysterectomy 10 , 11 , 12 has enabled studies on the prevalence of adenomyosis, which ranges from 5% to 70%. 13 , 14 Heterogeneity in study design and different criteria used to describe adenomyosis could explain some of these variations. Some ultrasonographic signs of adenomyosis are common but unspecific, whereas more specific signs are less prevalent. 15 The Morphological Uterus Sonographic Assessment (MUSA) group has suggested a uniform system to describe different features of adenomyosis. 16 , 17 In a revised paper, the MUSA group propose defining ultrasonographic features of adenomyosis as direct or indirect. 9 Direct features are pathognomonic for the disease, whereas indirect features are secondary to ectopic endometrium and only suggestive of adenomyosis. At present, the prevalence and sonographic characteristics of adenomyosis in subfertile women, using the revised MUSA definitions, is not completely known, nor has it been established whether the use of 3D TVUS improves the diagnostics when implementing the revised MUSA definitions. Studies that describe the prevalence of different ultrasonographic adenomyosis features in subfertile women, adhering to standardized diagnostic criteria, would facilitate further studies on the potential impact on fertility and ART outcome. 7 , 8 The primary aim of this study was to determine the prevalence and ultrasonographic characteristics of different features of adenomyosis, using the MUSA definitions, at two‐dimensional (2D) and 3D TVUS examination in a large cohort of women planned for their first ART.

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

This was a prospective, observational cross‐sectional study at the Reproductive Medical Center (RMC) at Skåne University Hospital, Malmö, Sweden, a tertiary care subfertility clinic treating couples from the Swedish Southern Healthcare region. Women scheduled for their first ART were consecutively invited to participate in the study between December 2018 to May 2021, as previously described. 18 Eligible for publicly funded ART are all subfertile, non‐smoking women between 25 and ≤39 years without common children with the present partner. A body mass index (BMI) ≤30 kg/m 2 or a >10% weight loss for BMI <30 to ≤35 kg/m 2 , is also required. Women with current hormonal treatment or previous uterine surgery were excluded, as this may alter the sonographic appearance of the myometrium. Women with previous surgical destruction of superficial endometriotic lesions were also excluded. They may wrongly have been classified as not having endometriosis, even if only the morphological changes can be removed surgically, not the disease per se. All women filled in a questionnaire regarding previous pregnancies and the presence of typical symptoms (dysmenorrhea, pelvic pain, dyspareunia, dyschezia, dysuria, hematochezia or hematuria), and underwent a systematic 2D and 3D TVUS examination by the first author. The severity of symptoms was reported on a 100‐mm visual analog scale (VAS). The left extreme indicated absence of pain, and the right extreme the worst possible pain. All women were examined in the lithotomy position using a Voluson 10 Expert (GE Medical systems, Zipf, Austria) high resolution ultrasound machine equipped with a 5–9 MHz transvaginal transducer (RIC5‐9D). Following acquisition, ultrasound images, video clips and volumes were stored on the imaging and information management systems Syngo® Dynamics (Siemens Medical Solutions Health Services, Malvern, PA, USA) and ViewPoint with the integrated 4D view software (GE Healthcare, Munich, Germany) for subsequent retrieval and offline analysis. All features of adenomyosis were assessed and documented while live scanning. Measurements were made from the images immediately after the examination. An offline re‐evaluation of 2D and 3D images was performed separately. The systematic TVUS examination included a dynamic 2D and 3D assessment of the uterus and endometrium in three orthogonal planes. The myometrium and JZ were evaluated first in a sweep of the uterus in the sagittal and transverse sections and then in the mid‐sagittal plane of the uterus in the 2D images. Thereafter, 3D volumes of the entire uterus were acquired in the sagittal plane in a standardized manner, 10 , 19 with a sweep angle of 120°. The uterus was assessed in the multiplanar view, which allows visualization of the rendered coronal plane. To assess the whole uterus by 3D ultrasound, this plane was moved up to the anterior wall and down to the posterior wall simultaneously observing the myometrium and JZ for features of adenomyosis. Each interstitial portion of the tubes was visualized, albeit not always simultaneously, when scrolling through and tilting the rendered 3D volume. The sagittal plane was examined in the same way. Power Doppler (fixed preinstalled settings: frequency, 5–9 MHz [‘normal’]; pulse repetition frequency, 0.3–0.6 kHz; gain, −4.0; wall motion filter, ‘low 1’ [40 Hz]) was used to evaluate trans‐ or intralesional vascularization of the myometrial tissue and to distinguish between a myometrial cyst and a vascular component. The myometrium was assessed at 2D and 3D TVUS. Features of adenomyosis were documented using the MUSA definitions. 17 At reassessment of the images, all features were classified into direct or indirect as suggested in the revised MUSA definitions. 9 , 17 Direct features include myometrial cysts, hyperechogenic islands and echogenic subendometrial lines and buds (Figure  1 ). Indirect features are globular uterus, asymmetrical myometrial thickening, fan shaped shadowing, translesional vascularity and irregular or interrupted junctional zone (Figure  2 ). Irregularities were assessed subjectively to obtain a general impression of the JZ including focal thickness. Lines and buds or myometrial cysts in the JZ constitute an irregularity and hence were reported as direct as well as indirect features if present. The JZ was categorized as interrupted when any proportion of the JZ could not be visualized in either 2D or 3D TVUS images in any plane. 9 The JZ was classified as not assessable when no portion of the JZ was visible, due either to myomas or to poor image quality. The anterior and posterior myometrial walls were measured in the midsagittal plane at the thickest point from the external uterine serosa to the basal line of the endometrial‐myometrial interface. The ratio between the thickest to the thinnest anterior and posterior myometrial walls was calculated. More than 5‐mm difference in wall thickness or a ratio well above one, unrelated to myomas, was considered to indicate asymmetrical myometrial thickening. 9 Translesional vascularity was defined as the presence of blood vessels perpendicular to the uterine cavity or crossing the lesion. The location of the different features of adenomyosis was described as anterior, posterior, fundus, lateral right or lateral left. Some women had lesions in multiple locations; if more than two, the location was described as global. The adenomyosis features were described as either focal (when normal myometrium surrounded >25% of the circumference of the lesion) or diffuse (if normal myometrium surrounded <25% or if it was difficult to differentiate focal from diffuse adenomyosis). Features were classified as “mixed‐type” if both focal and diffuse adenomyosis were present in different locations. In addition, features were classified as either cystic (with myometrial cysts ≥2 mm) or non‐cystic and the presence of an echogenic rim was documented. The uterine layer involvement was assessed: the inner myometrium (JZ, Type 1), the middle myometrium (Type 2), the outer myometrium (Type 3) or several layers involved (Type 1–2, Type 2–3 or Type 1–3). If asymmetry was the only feature present, no myometrial layer was assigned. The presence of sliding sign 20 was recorded to assess serosal involvement. 17 A negative sliding sign indicated fixed viscera. The extent of the disease was assessed subjectively based on the estimated affected proportion of the uterus, and classified as mild (50%). For focal lesions, the largest diameter of the largest lesion was measured. In the case of diffuse lesions, the myometrial wall thickness was measured. The uterine length was measured in the midsagittal plane from the external serosa of the fundus to the internal cervical os. The uterine height was measured in the same plane, anteroposteriorly from the outer serosal surfaces. The width was measured in the transverse plane from the right to the left outer serosal surface. Direct features of adenomyosis at 2D and 3D transvaginal ultrasound examination. The top row demonstrates direct features of adenomyosis in the sagittal view of the uterus at 2D TVUS. The bottom row shows direct features in the coronal view of the uterus JZ at 3D TVUS, corresponding to the 2D images on the upper row. The arrows mark the specific adenomyosis features, but several features may be seen in the images. (A,D) Hyperechogenic islands, (B,E) myometrial cysts, (C F) subendometrial lines and buds. 2D, two‐dimensional; 3D, three‐dimensional; JZ, junctional zone; TVUS, transvaginal ultrasonography. Indirect features of adenomyosis at 2D transvaginal ultrasound examination. Indirect features of adenomyosis are shown in the sagittal view of the uterus at 2D transvaginal ultrasound examination. The arrows mark the adenomyosis feature, but more than one feature may be seen in the images. (A) Globular uterus, (B) fan‐shaped shadowing, (C) asymmetrical myometrial thickening, (D) translesional vascularity, (E) interrupted junctional zone, (F) irregular junctional zone due to subendometrial lines and buds. 2D, Two‐dimensional. The adnexa and pelvis were assessed for endometriomas, defined as unilocular cysts with ground glass echogenicity 21 or as deep endometriotic lesions, which were reported using the International Deep Endometriosis Assessment group terminology. 18 , 22 The primary outcome was the prevalence of different features of adenomyosis at 2D and/or 3D TVUS, according to the MUSA definitions 16 , 17 and classified into direct or indirect features as described in the revised MUSA definitions. 9 Secondary outcomes were any difference in the diagnostics when using 2D or the coronal plane of 3D TVUS, the ultrasonographic characteristics of any features of adenomyosis 23 (location in the uterine wall, uterine layer involvement, extent, differentiation and size of the lesion) as well as any association between adenomyosis features and characteristics such as age, previous childbirth, termination of pregnancy, miscarriage or the presence of endometriosis and/or myomas. For statistical analyses, we used the statistical package IBM Corp. released 2020 (IBM SPSS Statistics for Windows, Version 29.0. Armonk, NY, USA). Assuming a prevalence of adenomyosis of 24% in our population, 24 we needed to recruit at least 281 women to achieve a precision of 5% with a confidence level of 95%. Median values were presented with range. The Chi‐square or Fisher's exact tests were used to compare proportions between groups. For skewed data, the Mann–Whitney U ‐test was used to compare median values between the two groups. Prevalence was reported as percentage with 95% confidence interval (CI). Univariate binary logistic regression analysis was used to assess whether clinical or demographic factors were associated with the occurrence of at least one direct feature of adenomyosis, at least one direct and/or indirect feature or at least one indirect feature of adenomyosis without any direct feature. The size of the association was quantified by odds ratios (OR). Values were given with 95% confidence intervals (CI) and P  < 0.05 was considered to indicate a statistically significant association. Variables that showed some association with the presence of features of adenomyosis on univariate analysis were adjusted for in a multivariate analysis. A backwards selection procedure was then performed to omit any variables not independently associated with the presence of at least one direct feature, at least one indirect feature or at least one indirect feature without any direct feature of adenomyosis.

Table S1. Table S2. Table S3.