Limited utility of subendometrial enhancement in assessing the interface between the endometrium and myometrium.

MRI was performed using 3-T MRI units (Ingenia, Philips Healthcare; Magnetom Vida, Siemens Healthineers) with phased-array torso coils. The gynecologic MR protocol included several sequences that were not directly assessed in this study: T2-weighted images in axial, coronal, and sagittal planes; T1-weighted axial images both with and without fat saturation; diffusion-weighted axial images with b values of 0 s/mm² and 1000 s/mm²; and T1-weighted 3D GRE CE images. Before acquiring the protocol-based CE images, dynamic CE MRI was performed to evaluate SEE, using an oblique axial plane perpendicular to the endometrial cavity. Two techniques were utilized to achieve high temporal resolution in dynamic CE MRI: the keyhole technique and the golden-angle radial sparse parallel (GRASP) method 21 , 22 . For the keyhole technique, the scan percentage was reduced to achieve a temporal resolution of 13.3 s, and imaging commenced simultaneously with the administration of 0.1 mmol gadolinium per kilogram of body weight, injected at a rate of 2 mL/s through an antecubital vein. This acquisition lasted 5 min and 9 s. In the GRASP method, 0.1 mmol gadolinium per kilogram of body weight was administered 30 s after the start of image acquisition. Over the course of 5 min and 32 s, a total of 1,940 radial spokes were collected, and the data was reconstructed to achieve an effective temporal resolution of 13 s. The additional parameters used in dynamic CE MRI are summarized in Table  1 . Table 1 Parameters used for dynamic contrast-enhanced MRI. Ingenia Magnetom Vida Sequence THRIVE* GRASP-VIBE** TR (ms) 2.9 3.6 TE (ms) 1.3 1.6 Flip angle ( o ) 10 12 Echo-train length 64 1 Number of excitations 1 1 Acceleration factor 2/1 (P reduction/ S reduction) 9.1 Field of view (cm) 380 * 380 380 * 380 Matrix 128 * 126 256 * 256 Slice thickness (mm) 3 3 *T1 high resolution isotropic volume examination. **Golden-angle radial sparse parallel-volumetric interpolated breath-hold examination. Parameters used for dynamic contrast-enhanced MRI. *T1 high resolution isotropic volume examination. **Golden-angle radial sparse parallel-volumetric interpolated breath-hold examination.

To provide a time series for each slice, the dynamic CE MR images were first sorted by slice location, and then each stack of slices was arranged in ascending order by time point. The MR images, organized into a time series per slice, were uploaded to a dedicated study worklist on the picture archiving and communication system (*******) after an author (***), who did not participate in subsequent image interpretation, removed all personal identifying information. Two board-certified radiologists (*** and ***, with 12 and 19 years of experience in genitourinary imaging, respectively) independently reviewed the dynamic CE MRI. Both radiologists were blinded to the participant details. Image analysis was conducted in two steps. Initially, each radiologist assessed the presence or absence of SEE, which is defined as a thin layer of enhancement between the endometrium and the myometrium 20 . Any thick (≥ 2 mm) layer of enhancement was not considered SEE 23 , 24 (Fig. 2). In the second step, when SEE was identified, the radiologists examined dynamic CE MR images of the entire uterine body to evaluate the reliability of SEE in delineating the interface between the endometrium and the myometrium. This evaluation used a 4-point Likert scale: grade 1 indicated very unreliable SEE (little demonstrable SEE); grade 2, unreliable SEE (poor delineation of the endometrial/myometrial interface); grade 3, reliable SEE (good delineation of the endometrial/myometrial interface); and grade 4, very reliable SEE (excellent delineation of the endometrial/myometrial interface) (Fig. 3). Fig. 2 Subendometrial enhancement (SEE) is defined as a thin layer of enhancement between the endometrium and the myometrium (arrowheads in A ). A thick layer of enhancement (arrowheads in B ) is not considered SEE. Subendometrial enhancement (SEE) is defined as a thin layer of enhancement between the endometrium and the myometrium (arrowheads in A ). A thick layer of enhancement (arrowheads in B ) is not considered SEE. Fig. 3 Reliability of subendometrial enhancement (SEE). In a 30-year-old woman ( A ), both radiologists categorized the reliability of SEE as grade 2 (poor delineation of the endometrial/myometrial interface). In a 35-year-old woman ( B ), radiologist 1 categorized the reliability of SEE as grade 4 (excellent delineation of the endometrial/myometrial interface), while radiologist 2 categorized it as grade 3 (good delineation of the endometrial/myometrial interface). Reliability of subendometrial enhancement (SEE). In a 30-year-old woman ( A ), both radiologists categorized the reliability of SEE as grade 2 (poor delineation of the endometrial/myometrial interface). In a 35-year-old woman ( B ), radiologist 1 categorized the reliability of SEE as grade 4 (excellent delineation of the endometrial/myometrial interface), while radiologist 2 categorized it as grade 3 (good delineation of the endometrial/myometrial interface). The statistical analysis was conducted using MedCalc, version 18.6 (MedCalc Software, Mariakerke, Belgium). We assessed interobserver agreement by employing the weighted kappa (κ) statistic, categorizing κ values as follows: 0.00–0.20 indicates slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81-1.0, near perfect agreement. The chi-square test was utilized to compare the presence of SEE relative to menstrual status. A P-value below 0.05 was deemed to signify a statistically significant difference.

Radiologist 1 identified SEE in 49 out of 110 women (44.5%), while radiologist 2 identified SEE in 43 out of 110 women (39.1%) (Table  2 ). The inter-observer agreement for the identification of SEE was good, with a κ value of 0.74 (95% CI, 0.613–0.866). SEE was more commonly identified during the proliferative phase (56.4% for radiologist 1 and 48.7% for radiologist 2) than in the secretory phase (38.3% for radiologist 1 and 34.0% for radiologist 2) or the menopausal state (38.5% for radiologist 1 and 46.2% for radiologist 2). However, the differences were not statistically significant between the proliferative and secretory phases ( P  = 0.0956 for radiologist 1 and P  = 0.1704 for radiologist 2), between the proliferative phase and menopausal status ( P  = 0.2666 for radiologist 1 and P  = 0.8739 for radiologist 2), and between the premenopausal and postmenopausal stages ( P  = 0.6399 for radiologist 1 and P  = 0.5801 for radiologist 2). Radiologist 1 classified the identified areas of SEE as reliable ( n  = 10) or very reliable ( n  = 2), accounting for 24.5% (12/49), while radiologist 2 classified them as reliable ( n  = 12) or very reliable ( n  = 2), accounting for 32.6% (14/43), in delineating the interface between the endometrium and myometrium. Therefore, among the study population ( n  = 110), areas of SEE deemed reliable for assessing the interface between the endometrium and myometrium were noted in only 12 women (10.9%) by radiologist 1 and in 14 women (12.7%) by radiologist 2. Table 2 Subendometrial enhancement according to menstrual status. Menstrual status Subendometrial enhancement Radiologist 1 Radiologist 2 Proliferative phase 56.4% (22/39) 48.7% (19/39) Secretory phase 38.3% (18/47) 34.0% (16/47) Undetermined 36.4% (4/11) 18.2% (2/11) Postmenopausal stage 38.5% (5/13) 46.2% (6/13) Total 44.5% (49/110) 39.1% (43/110) Subendometrial enhancement according to menstrual status.

SEE was first described by Yamashita et al. 20 , who reported on dynamic CE MRI of the normal uterus and endometrial cancer. In their study, SEE was observed in 74.1% (20/27) of normal uteri. Recent advances in the temporal and spatial resolution of dynamic CE MRI are expected to increase the frequency of SEE detection. However, to our knowledge, no further studies have investigated the detection of SEE in the normal uterus, underscoring the need for validation studies on this topic. In our study, which included 110 women without endometrial pathology, SEE was observed less frequently than in the previous study, with rates ranging from 39.1 to 44.5% versus 74.1%. Additionally, the identified areas of SEE were reliable for delineating the interface between the myometrium and endometrium in only 24.5% (12/49) of the cases detected by radiologist 1 and 32.6% (14/43) of those detected by radiologist 2. Overall, areas of SEE deemed reliable for delineating this interface were noted in only 10.9% (12/110) to 12.7% (14/110) of the study population ( n  = 110). If SEE is not frequently demonstrated, as indicated by our findings, relying on SEE in dynamic CE MRI may not be effective for assessing myometrial invasion in women with endometrial cancers. Our results also cast doubt on the hypothesis proposed by Yamashita et al. that menstrual status affects the identification of SEE 20 . In their study, SEE was more frequently observed in women during the proliferative (100%) and postmenopausal (92.9%) stages compared to those in the secretory phase (40.0%). Their study, however, had a limited sample size ( n  = 24) and called for follow-up studies to confirm these findings, though none have yet been conducted. In contrast, our study, which included 110 women, found no significant differences in the identification of SEE between the proliferative phase (48.7 − 56.4%) and the secretory phase (34.0 − 38.3%), nor between the premenopausal stage (38.1-45.4%) and the postmenopausal stage (38.5-46.2%). Given the discrepancies between the results of Yamashita et al. study and those of our study, further research involving a larger population is necessary to determine if menstrual status truly influences the identification of SEE. The lower specificity of dynamic CE MRI in detecting myometrial invasion has been a significant challenge in endometrial cancer 10 , 12 , 17 . In a study including 147 women with endometrial cancer, Fujii et al. 12 reported that intact SEE was observed in only 13 of 28 women without myometrial invasion (46.4%) by radiologist 1, and in 9 of 28 women (32.1%) by radiologist 2. Similarly, Cui et al. 17 identified intact SEE in 9 of 17 women without myometrial invasion (52.9%). The current study demonstrates that a reliable SEE, which assesses the interface between the endometrium and myometrium, is infrequently observed in a normal uterus, highlighting the difficulty in overcoming the lower specificity of dynamic CE MRI in detecting myometrial invasion in women with endometrial cancer. Additionally, Nakao et al. reported a low negative predictive value of SEE for myometrial invasion in endometrial cancers 10 . In their analysis of 116 women with endometrial cancer, eight (40%) of 20 women with an intact SEE on dynamic CE MRI were later found to have myometrial invasion. Given the low specificity and negative predictive value of SEE in assessing myometrial invasion in endometrial cancer, it may be necessary to re-evaluate the essential role of dynamic CE MRI in selecting women eligible for fertility-sparing treatment. This reconsideration becomes particularly important considering that no significant difference exists in tumor-myometrium contrast between dynamic CE MRI and single-phase CE MRI 25 , 26 . In the present study, no attempts were made to identify SEE using various time resolutions. Previous research has utilized a range of time resolutions from 7.8 s to 60 s 10 , 12 , 17 , 20 ; therefore, employing a fixed time resolution of approximately 13 s may limit the assessment of SEE in a normal uterus. The findings that reliable areas of SEE for assessing the interface between the endometrium and myometrium are infrequently observed on dynamic CE MRI requires validation in future studies employing various time resolutions. Selection bias is another limitation of this study. We attempted to exclude individuals with pathologies affecting SEE. However, these exclusions were based solely on radiological assessments and not on established pathological or clinical criteria. This approach introduces potential bias and may limit the generalizability of our results. Given that adenomyosis could affect the interface of the endometrium and myometrium, the lack of exclusion of women with adenomyosis may be also a limitation of this study. However, considering that adenomyosis is a common gynecological condition and that many cases may be underdiagnosed on MRI, we believe that not excluding women with adenomyosis based on radiologic assessment may better reflect the incidence of SEE in routine clinical situations. Furthermore, this study involved a relatively small number of participants and was conducted at a single institution. The findings might vary with the inclusion of multiple institutions and a larger participant pool. The present study re-evaluated the utility of SEE in assessing the interface between the endometrium and myometrium. Our findings indicate that reliable areas of SEE for assessing this interface are infrequently observed on dynamic contrast-enhanced (CE) MRI, even when taking menstrual status into account. In conclusion, evaluating SEE on dynamic CE MRI may have limitations in assessing myometrial invasion in women with endometrial cancers. We hope that our experience will inform the decision on whether to use dynamic CE MRI or single-phase CE MRI when performing CE MRI in women with endometrial cancers.

Endometrial cancer is the most common gynecologic malignancy in developed countries 1 . With the increasing rates of obesity, nulliparity, and polycystic ovarian syndrome, there has been a corresponding rise in the incidence of endometrial cancer among younger women 2 – 4 . For these younger patients with low-grade endometrioid adenocarcinoma and no evidence of myometrial invasion on imaging, fertility-sparing progestogen therapy can be considered 5 – 9 . Uninterrupted subendometrial enhancement (SEE) is a crucial imaging finding used to identify women who are candidates for fertility-sparing treatment 10 – 15 . This finding underpins the preference for dynamic contrast-enhanced (CE) MRI in women opting for conservative management 12 , 16 , 17 . Moreover, with the FIGO 2023 staging system reintroducing the distinction between minimal myometrial invasion and no myometrial invasion, the preference for dynamic contrast-enhanced (CE) MRI to evaluate SEE may further increase 18 . However, the lower specificity of dynamic CE MRI in detecting myometrial invasion presents a significant challenge in the management of endometrial cancer 10 , 12 , 17 , 19 . Our experience suggests that SEE is not commonly observed on dynamic CE MRI, which may contribute to its reduced specificity in detecting myometrial invasion in women with endometrial cancers. To date, no studies have been conducted to determine the frequency of SEE in a normal uterus since it was first described by Yamashita et al. 20 . Therefore, this study aimed to re-evaluate the utility of SEE in assessing the interface between the endometrium and myometrium in uteri without any endometrial pathology.

This prospective study was approved by the institutional review board of Seoul Metropolitan Government Seoul National University (SMG-SNU) Boramae Medical Center (institutional review board [IRB] No. 20-2023-77) and carried out in accordance with the Declaration of Helsinki of the World Medical Association. All participants were informed that they would undergo dynamic CE MRI in addition to protocol-based single-phase CE MRI, without any additional administration of contrast material. Written informed consent was obtained from all participants. A total of 233 consecutive women who underwent pelvic CE MRI for gynecologic issues were prospectively enrolled in this study between June 2023 and September 2023. Of these, 205 women met the inclusion criteria: (1) MRI was performed due to suspicion of gynecologic issues excluding endometrial or cervical pathologies, (2) no history of hysterectomy, pelvic radiation therapy, or chemotherapy, (3) consent was provided for participation in the study. One author (***, with 9 years of experience in genitourinary imaging) who had not participated in the image analysis reviewed images to confirm that they were suitable for the evaluation of SEE and excluded 95 women based on the following reasons: (1) myoma ( n  = 65) and endometriosis ( n  = 3), which cause distortion of the endometrial cavity, making it difficult to obtain oblique axial images; (2) unrecognized hysterectomy state ( n  = 5) and subtotal hysterectomy state ( n  = 1) before the examination; (3) endometrial compression by myoma, which could cause hemodynamic change ( n  = 10); (4) suspected endometrial pathology ( n  = 2); (5) presence of hydrometra ( n  = 1); (6) technical failure to obtain oblique axial images ( n  = 8). Consequently, 110 women (age range, 14–81 years; mean age, 39 years) were included in the final analysis (Fig.  1 ). Before undergoing the MR examination, participants were asked about their menstrual history, including cycle regularity, duration, and any noted changes or abnormalities. Menstrual status was then classified as proliferative phase, secretory phase, menopausal status, or undetermined, and this classification was recorded for subsequent analysis. Fig. 1 The flow of selection of study population. The flow of selection of study population.