In 2018, endometrial cancer (EC) emerged as the sixth most prevalent cancer impacting women across the globe, according to the World Health Organization, with 380,000 new cases identified [1]. EC is most often observed in postmenopausal women, with 14% of cases occurring in the premenopausal period and 5% of cases manifesting in females below the age of 40. Key risk factors include obesity, early onset of menstruation, late onset of menopause, excessive exposure to unopposed estrogen associated with nulliparity, diabetes, advanced age ($>$55), and the use of tamoxifen [2]. Unusual vaginal discharge, which can sometimes occur alongside vaginal bleeding and uterine infection, is the most frequent symptom in EC patients and evident in around 90% of instances. The majority of instances are identified in the early stages, primarily owing to irregular vaginal bleeding. However, in advanced-stage disease, the symptoms can resemble those of ovarian cancer, such as abdominal bloating and pelvic or abdominal pain [3]. The gold standard diagnostic method for EC is the endometrial biopsy [4]. In cases of early-stage EC confined to the uterus, surgical staging is advised as the initial treatment. This involves total hysterectomy, bilateral salpingo-oophorectomy, and lymph node (LN) assessment through pelvic and para-aortic lymphadenectomy, particularly in high-intermediate to high-risk EC [5]. For specific histological subtypes such as serous, undifferentiated endometrial carcinoma, and carcinosarcoma, staging techniques involve infracolic omentectomy due to the heightened risk of microscopic omental metastases. For intermediate-high/high-risk patients, lymph node staging is advised. For cases of stage I disease (IV, B), it is acceptable to omit treatment for clear cell and endometrioid carcinoma [5]. An alternative to systematic lymphadenectomy for staging purposes is the use of sentinel lymph node (SLN) biopsy. Additionally, it may be taken into consideration in low/low-intermediate-risk individuals to rule out occult lymph node metastases and determine whether the disease is truly confined to the uterus. Therefore, guidelines from European Society of Gynaecological Oncology - The European Society for Radiotherapy and Oncology - The European Society of Pathology (ESGO-ESTRO-ESP), also approved by International Federation of Gynecology and Obstetrics (FIGO), allow the SLN approach for all patients with endometrial carcinoma [6]. Frozen section examination provides important information about the size of intraoperative tumors and invasion of the myometrium. Invasion is closely related with metastasis to lymph nodes; according to the 2018 FIGO staging, women with invasion of the myometrium (MI) of over 50% (stage IB) require an additional lymphadenectomy and omentectomy [7]. The likelihood of lymph node involvement demonstrably increases to 20% in cases with deep MI, in stark contrast to the 5% associated with superficial invasion [8]. Routine lymphadenectomy does not provide additional benefits in low-risk EC, but it yields a beneficial effect on women with intermediate and high-risk disease who undergo lymph node dissection in both pelvic and para-aortic regions [9]. Since the update of the FIGO staging, progress has been made in comprehending the pathological and molecular characteristics of EC. There is now more data on various histological types, along with biological behavior information. The clarity of findings concerning the molecular and genetic structure of cancer has increased with the publication of The Cancer Genome Atlas (TCGA) data. The objectives of the updated staging system are to refine the definition of prognostic groups and establish sub-stages that guide more tailored approaches to surgical, radiation, and systemic treatments [10]. Where possible, the inclusion of molecular subtypes to the staging criteria in the update helps predict prognosis more accurately. Due to its impact on prognosis determination and adjuvant treatment decisions, comprehensive molecular classification analyses (POLEmut, MMRd, NSMP, p53abn) should be performed in every one of the cases. Molecular subcategory assessment can be conducted on biopsy material; in this case, repeating it on hysterectomy material is not recommended [10]. Hence, a thorough assessment of women with EC, particularly those with MI, employing contemporary imaging techniques, has emerged as a gold-standard as it enables precise staging and enhances the efficacy of surgical intervention [11]. While transvaginal sonography (TVUS) serves as a common method for assessing the depth of MI and identifying cervical stromal invasion, its accuracy in diagnosis has demonstrated considerable variability in various studies. Specific research has yielded results comparable to those obtained using magnetic resonance imaging (MRI) [12].

The aim of our study was to contrast the preoperative TVUS and MRI findings with the intraoperative frozen section and the postoperative pathologic results of 321 patients diagnosed with EC.

This retrospective study was conducted utilizing clinical and pathological data from 370 individuals who underwent hysterectomy for EC in a tertiary gynecologic-oncology clinic between January 2012 and December 2022. Excluded from the study were instances without preoperative biopsy, TVUS, or MRI, as well as cases featuring evident extrauterine lesions. Cases with a diagnosis of involvement of organs outside the uterus during surgery were also excluded. Additionally, cases suspected of atypical endometrial hyperplasia before surgery and subsequently diagnosed with definite EC were included in the study, whereas cases with a definite diagnosis of atypical endometrial hyperplasia were excluded [non-endometrioid EC (n = 25), no imaging (n = 8), extrauterine involvement (n = 7), endometrial hyperplasia (n = 9)]. The final study was comprised of 321 patients.

The TVUS assessments were conducted by proficient gynecologists in our gynecology clinic or the tertiary center based on the patient’s symptoms. Effectively identifying and assessing endometrial pathology in symptomatic women can be achieved through the economical use of two-dimensional transvaginal ultrasound (2D-US) as the primary assessment. This is accomplished by measuring endometrial thickness (ET), which has been demonstrated to reliably predict EC [13]. Assessment of the endometrium, utilizing two distinct layers and focusing on maximal anteroposterior dimensions, reveals a threshold of 4 to 5 mm, beyond which the likelihood of pathological alterations significantly increases [14]. Patients’ ultrasound measurements were recorded in hard disk drives. All TVUS examinations were performed by proficient gynecologist-oncologist, or expert in gynecologic ultrasound. The ultrasound equipment used was the SonoAce R3, Elite (Samsung Medison, Seoul, South Korea). MI and cervical stromal invasion (CSI) depth were subjectively assessed. MRI, with its excellent soft tissue contrast and multiplanar capability, surpasses computed tomography (CT) in assessing the depth of MI, cervical invasion, and early parametrial invasion. CT uses ionizing radiation, and its low soft tissue contrast resolution is less effective than MRI in distinguishing between the tumor in the uterine corpus, cervix and normal soft tissues. While MRI is considered the best alternative for patients with contrast allergies or renal insufficiency, CT is more sensitive than MRI in overall detection of tumor spread beyond the uterus [15].

The magnetic resonance images were obtained utilizing a 1.5 Tesla MRI device (Solo, Siemens, Munich, Germany) equipped with a 16-channel phased-array coil system. Patients were positioned supine posture. The imaging procedure comprised T1A (repetition time/echo time (TR/TE): 609/19 ms, slice thickness/gap: 5/1, matrix: 256 $\times$ 128, field of view: 30 cm, 1 excitation) and T2A axial short tau inversion recovery (STIR) (TR/TE: 4000/46 ms, slice thickness/gap: 5/1, matrix: 256 $\times$ 256, field of view: 18–20 cm), along with T2A axial (TR/TE: 7220/103 ms) images. Pre-surgical assessment employed dynamic contrast-enhanced MRI with fat-saturated T1-weighted gradient-echo sequences (TR/TE 4.81/1.77 ms, flip angle 15°) before and at 30, 60, and 120 seconds following intravenous injection of 0.1 mmol/kg meglumine gadoterate (Magnescope ®, Guerbet, Tokyo, Japan). Two radiologists independently interpreted each patient’s MRI data, resolving any discrepancies through joint discussion to arrive at a final diagnosis.

Concurrently, frozen section (FS) analysis based on macroscopic examination, was performed by pathologists on the area exhibiting the deepest apparent MI. A gynecology-specialized pathologist, unaware of the MRI findings, examined one or two frozen section slides under a microscope. Definitive post-surgical diagnosis was confirmed using formalin-fixed paraffin-embedded sections.

The average age of the women with EC was 63.9 $\pm$ 9.5 years, and 86.3% of the patients were postmenopausal. The most common complaint among the patients presenting to the clinic was postmenopausal bleeding (83.5%). Early-stage EC was observed in 88.4% of the patients, low-grade cancer was observed in 39.9% of the patients, and lymph vascular space invasion (LVSI)+ was observed in 15.3% of the patients. The findings are summarized in Table 1.

MI was noted and graded as 50% or $>$50% of myometrial involvement. TVUS revealed more than 50% MI in 23 patients (7.2%), with the final pathology showing below 50% MI in 18 cases and above 50% MI in 5 cases. On MRI, 12 patients (3.7%) were identified with MI above 50%, and in the final pathology, MI was below 50% in 7 cases and above 50% in 5 cases. A statistically significant difference was identified between the preoperative TVUS and MRI findings of the patients and the final results (p 0.001, p 0.016). The agreement of the results was low (kappa 0.340–0.579). The agreement was almost perfect in the frozen section results. The specificity of TVUS, MRI, and FS was 94.3%, 97.8%, and 99.7%, respectively. The results are summarized in Tables 2,3.

A statistically noteworthy disparity was observed in the concordance rates of the MRI and ultrasound results with the final pathology (p = 0.001). The results are summarized in Table 4.

In our investigation, we conducted an assessment and comparing the diagnostic accuracy of TVUS, MRI, and frozen sections for the identification of deep MI in women diagnosed with EC, who underwent operative staging. According to our results, MRI is not necessary in cases where invasion is confirmed, but it is necessary in cases where invasion is not detected by TVUS. According to the kappa test, MRI had a lower rate of missing invasion compared to TVUS.

In contrast to MRI, TVUS is a more cost-effective technology extensively employed in gynecology, without any contraindications for its utilization. It can be conducted during the initial oncologic visit, leading to cost reductions, decrease in preoperative diagnostic evaluation time, and alleviation of patient anxiety.

To evaluate the extent of MI, contrast-enhanced T${}_1$-weighted imaging (T${}_1$WI) incorporating dynamic contrast-enhanced (DCE) magnetic resonance (MR) imaging is deemed more effective than T${}_2$-weighted imaging (T${}_2$WI) in determining deep MI [20]. In contemporary practice, MR examination, comprising T${}_2$WI and DCE MR imaging on different planes is regarded as the norm for the preoperative evaluation of MI in those with EC [20]. Nevertheless, contrast-enhanced T${}_1$WI imaging necessitates the use of gadolinium-containing agents, that present contraindications like significant renal impairment prone to nephrogenic systemic fibrosis, hypersensitivity to gadolinium-containing agents, or during pregnancy [21]. Beddy et al. [22] stated that diffusion-weighted imaging (DWI) exhibited greater diagnostic precision in evaluating the MI depth, compared to DCE MR imaging. DWI demonstrates a greater diagnostic accuracy in identifying deep MI, without noteworthy variance between sensitivity and specificity compared to DCE MR imaging, as documented by a meta-analysis [23]. Consequently, DWI serves as an alternate option to contrast-enhanced MRI when assessing MI of EC. However, the comparatively limited image detail of conventional DWI and signal alteration from vulnerability artifacts due to gas in the intestines may lead to misinterpretation of MI [24]. MRI is widely acknowledged as the optimal imaging method for preoperative staging of EC, exhibiting high sensitivity in evaluating the existence of deep MI [25]. Nonetheless, MRI comes with certain constraints since limited reader expertise may compromise its value and there is comparatively high interobserver variability [26]. Several pieces of evidence suggest that T${}_2$WI can be synergistically empowered by the joint input of both DWI and T${}_1$WI dynamic contrast-enhanced (DCE) sequences for deep MI [27]. While concise unenhanced protocols exist, DWI’s accuracy potentially rivals or surpasses DCE for EC staging [28]. Additionally, the use of a contrast agent extends the examination duration, escalates its financial cost, and presents uncertain long-term consequences in clinical contexts [29]. Acknowledging these points, the European Society of Urogenital Radiology’s guidelines emphasize that conclusive evidence regarding the superiority of combined T${}_2$WI and DWI over DCE imaging for prostate cancer staging remains elusive. Harnessing the power of machine learning, radiomics sheds new light on image analysis by quantifying data within medical scans. This cutting-edge approach allows us to decipher the intricate landscape of a tumor, revealing its internal variations. In the realm of oncology research, radiomics reigns supreme, serving as a versatile tool for diverse tasks, from predicting a tumor’s grade and lineage to guiding diagnosis, tailoring treatment plans, and monitoring response to therapy [30]. Bolstering diagnostic precision and treatment response evaluations while alleviating the need for invasive procedures, MRI-based radiomics in EC unveils promising potential, as evidenced by a recent study highlighting its surgical utility [31].

The standard approach for EC is a total abdominal hysterectomy, bilateral salpingo-oophorectomy and peritoneal washing cytology. As MI deepens ($>$50%), the likelihood of metastasis to lymph nodes increases, requiring the inclusion of lymphadenectomy to the standard surgical procedure. Thorough preoperative evaluation can effectively pinpoint individuals with confirmed lymphatic spread, enabling focused postoperative intervention and potentially reducing the adverse effects associated with superfluous radiotherapy [32]. Additionally, LND can eliminate metastatic lymphatic disease. On the other hand, routine lymphadenectomy has been shown not to provide a benefit in disease-free and overall survival [33]. This situation has led surgeons to conduct preoperative assessments to evaluate the requirement for lymphadenectomy in cases of EC, and the most logical approach has been to determine the depth of MI.

There are extensively reported challenges in identifying deep MI. In a previous study, 100 cases of endometrial carcinoma were examined [33]. The deep MI was reevaluated, and morphological features that complicated the assessment of MI were investigated. Initially diagnosed as invasive, almost all endometrial cancers were reevaluated as noninvasive in the study. When irregular endomyometrial connections, exophytic tumors, and adenomyosis were investigated to determine if the spread of stromal metaplasia, noninvasive patterns, and MI patterns differed in cases with and without measurement inconsistency, it was observed that irregular endomyometrial connections, exophytic tumors, and adenomyosis commonly coexisted, and they were more prevalent in cases with deep MI inconsistency. MI patterns other than the traditional destructive pattern were found to be rare enough in numerous cases not to affect deep MI measurement. Deep MI measurement is generally uncomplicated, but focus should be given to instances involving exophytic tumors, irregular endomyometrial connections, adenomyosis, and widespread stromal smooth muscle metaplasia. We share the same opinion in our evaluations [34].

When reviewing the literature, it has been observed that 3D TVUS, unlike 2D TVUS, exhibits good diagnostic accuracy in evaluating deep MI and cervical invasion, with sensitivity and specificity comparable to MRI [35]. In a meta-analysis covering 2773 patients, two dimensional-transvaginal ultrasound (2D-TVS) was evaluated for assessment of deep MI after operation, with a sensitivity of 82% and a specificity of 81% [36]. However, in a recent study on early-stage low-grade EC patients, 2D-TVS demonstrated lower sensitivity at 69% and higher specificity at 87%, while MRI showed values of 51% and 91%, respectively [37]. The known disadvantages of MRI include being time-consuming and expensive. However, it provides additional advantages in identifying high-risk patients concerning cervical stromal invasion, MI, and lymphatic involvement. In a meta-analysis, the sensitivity for detecting high-risk EC was determined to be 80.7% [38]. The gold standard method for determining MI is the histologic examination of hysterectomy material in frozen sections. In a recent study, the concordance for early-stage low-grade frozen sections and the final pathology was reported as 92.3% for the histologic type, 77% agreement regarding tumor grade, 82% confirmation of the depth of MI, and a 100% match in terms of tumor size [39]. In a study involving 378 patients comparing preoperative MR and FS concordance, the concordance rate of FS with MI was superior to that of FS with MRI. Despite its limitations, MRI remains a valid tool for informing preoperative strategies, particularly in the context of lymphadenectomy for endometrioid adenocarcinoma of grade 1 or 2 [40]. In another study, transvaginal ultrasound demonstrated a sensitivity of 88.64%, specificity of 90.48%, positive predictive value of 95.12%, and negative predictive value of 79.17% in when determining the extent of MI. Magnetic resonance imaging showed a sensitivity of 63.64%, specificity of 95.24%, positive predictive value of 96.55%, and negative predictive value of 55.56%. The study concluded that when performed by experienced specialists, transvaginal ultrasound, being inexpensive and easily applicable, is more advantageous in determining the depth of MI [7]. In our study, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and kappa values of TVUS were 100%, 94.3%, 21.7%, 100%, and 0.34, respectively. For MRI, these values were 100%, 97.8%, 41.7%, 100%, and 0.579, and for FS, they were 100%, 99.7%, 83.3%, 100%, and 0.908, respectively. The highest concordance in our study was achieved with FS; although TVUS showed similar sensitivity to MRI, it was lower than FS examination.

The general limitation of this study arises from its retrospective design, and the abundance of cases may partially alleviate this limitation. The absence of sentinel lymph node biopsy with current findings restricts contemporary approaches. Difficulties in evaluating MI, especially with ultrasound, are encountered. It is well known that most women with EC are obese, often leading to the uterus being in an upright position. It is acknowledged that assessing the endometrium is challenging when it is at a 0° angle to the probe. However, using a free hand and applying slight pressure on the abdomen to manipulate the uterus can overcome this situation. Another issue is the presence of large tumors mimicking cervical invasion by extending towards the cervical canal. These cases can often be excluded using dynamic examination techniques, evaluating the tumor’s mobility over the internal cervical os, differentiating a tumor truly invading the cervix from one mimicking cervical invasion. Additionally, blood vessels entering the tumor at the internal cervical os and showing cervical stromal invasion can be examined. We believe that MRI-based radiomics, introduced in recent years, has the potential to guide surgery.

It is our opinion that with further advancements in imaging technologies, concordance of TVUS may approach that of frozen sections. Finding a sensitivity close to FS, contrary to many studies in the literature, is one of the strengths of our research. We suggest that conducting studies with multicenter collaboration, prospective planning, and the increasing integration of new imaging methods into practice will yield better results.

Systematic lymphadenectomy is an advised strategy in FIGO stage IB EC cases when the myometrial invasion depth is greater than 50%. In addition to the gold standard FS, TVUS and MRI methods have a significant role in assessing the extent of invasion preoperatively. Considering the expense of MRI, the practical nature and accessibility of TVUS are notable advantages. In preoperative assessment, TVUS has been found to provide acceptable sensitivity close to MRI and FS in guiding intraoperative decisions for lymphadenectomy.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

SÖ, FŞ, ÖA designed the research study. OD, ÖA, AKK performed the literature search and analyzed the data. FŞ, SÖ, ÖA and AKK preparation, creation and/or presentation of the published work by those from the original research group, specifically critical review, commentary or revision—including pre- or post-publication stages. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

This retrospective study received ethical approval from the Istanbul Prof. Dr. Cemil Taşcıoğlu City Hospital Clinical Research Ethics Committee (approval no. 208). Written permission was obtained from the institutions where the research was conducted, and informed consent was obtained from the patients. The study was conducted in accordance with the principles of the Declaration of Helsinki.

We would like to express our gratitude to all those who helped us during the writing of this manuscript. Thanks to all the peer reviewers for their opinions and suggestions.

This research received no external funding.

The authors declare no conflict of interest.