Introduction

In 2009, the European Society of Urogenital Radiology (ESUR) published guidelines for the staging of endometrial cancer (EC). The new guidelines recognized magnetic resonance imaging (MRI) as the imaging modality of choice for evaluating disease extent in patients with newly diagnosed EC [1].

More recently, the European Society for Medical Oncology (ESMO) recommended that the initial surgical treatment of patients with EC should be tailored based on the risk of lymph node metastases, addressing the ongoing controversy regarding the role of lymphadenectomy [2,3,4]. The major clinical challenge in the initial management of EC is to distinguish patients who are at intermediate to high risk of lymph node metastases from those at low risk to avoid overtreatment. The ESMO guidelines advise against lymphadenectomy in low risk patients, i.e. grade 1 or 2 endometrioid adenocarcinoma without deep myometrial invasion (MI) [2, 3]. In contrast, lymphadenectomy is suggested or recommended for intermediate and high-risk groups, respectively. Additionally, while further studies are needed, sentinel lymph node (SLN) sampling is now recognized as a potential alternative to lymphadenectomy [5].

MRI can accurately assess the depth of MI and, thus, it is useful to stratify patients into low versus intermediate to high-risk groups before the surgery. Until recently, a combination of T2-weighted (T2WI) and dynamic contrast-enhanced imaging (DCE-MRI) was accepted as the best approach for local staging of EC [6, 7]. Recent evidence suggests that diffusion-weighted imaging (DWI) improves the evaluation of MI. As a result, DWI is now routinely used as an adjunct to T2WI and DCE-MRI [8,9,10,11,12,13]. It remains to be determined whether combined T2WI and DWI is superior to DCE-MRI.

The aim of this manuscript is to present updated ESUR recommendations for the initial staging of EC, reflecting recent clinical and imaging developments. The value of a tailored MR imaging protocol and a standardized imaging report is emphasized.

Materials and methods

The initial ESUR guidelines for staging of EC with MRI were published in 2009 [1]. Over the past year, members of the ESUR Female Pelvic Imaging working group have re-examined the current literature and clinical standards in management of EC, resulting in this update.

We chose the RAND-UCLA Appropriateness Method (RAM) because of its strength in combining evidence-based data and expert judgments to attain consensus on a variety of clinically pertinent questions. RAM was previously used to develop the European Society of Gastrointestinal and Abdominal Radiology consensus guidelines for MR imaging assessment of rectal cancer [14, 15].

The methodological steps were as follows:

  • Step 1: Literature review

Medline (Ovid), EMBRASE (embrace.com) and the Cochrane Library were searched for original manuscripts published between 2007 and 2017 pertaining to MR imaging and staging of EC (Supplementary material – File 1).

  • Step 2: Questionnaire development

A questionnaire consisting of 81 questions was developed by the two lead authors (SN and RF) and later refined with input from three advising members (ES, AR, ITN). The questions in the survey focused on the key MR imaging requirements, integration of functional MR imaging sequences and development of the standardized imaging report.

  • Step 3: Panel selection

The panel was comprised of all members of the ESUR Female Pelvic Imaging working group.

  • Step 4: Survey prior to the first meeting of the panel

The questionnaire was distributed to all members of the panel via electronic mail in November 2016; responses were recorded using a dedicated survey platform (surveymonkey.com).

  • Step 5: Data extraction and analysis

The answers to the questionnaire were collected in January 2017 and analysed by one lead author (SN). On the basis of the answers to the survey, each item was classified as follows: (1) “RECOMMENDED” (at least 80% agreement in favour), (2) “NOT RECOMMENDED” (at least 80% agreement in opposition) or (3) “UNCERTAIN”, i.e. consensus was not reached (less than 80% agreement). The results were presented to and discussed with the ESUR Female Pelvic Imaging working group at the Annual European Congress of Radiology Meeting in March 2017.

  • Step 6: Second survey

Five extra questions were added to the survey and distributed to all members via electronic mail to clarify any potentially conflicting answers that arose after the first survey and the first meeting.

  • Step 7: Second and final meeting of the panel

The members of the ESUR Female Pelvic Imaging working group met again at the Annual European Symposium on Urogenital Radiology in September 2017 with two lead authors (SN and RF) serving as moderators. Final results of the survey were circulated among all members of the panel 2 weeks prior to this meeting and formed the basis of the discussion at the meeting. The focus was on the questions with no consensus among experts.

  • Step 8: Data reporting

The final data analysis was performed by the two lead authors (SN and RF). Each item was ultimately classified as (1) “RECOMMENDED” versus (2) “NOT RECOMMENDED” if at least 80% agreement among experts or (3) “UNCERTAIN” (no consensus defined as less than 80% agreement among experts).

Results update

The panel included 28 experts from 26 different institutions. Twenty-three members were from centres in Europe: Portugal (n = 2), France (n = 3), Spain (n = 2), United Kingdom (n = 5), Switzerland (n = 3), Germany (n = 3), Austria (n = 1), Sweden (n = 1), Italy (n = 2) and Greece (n = 1). Five panellists were from the three institutions outside Europe: Japan, USA and Brazil.

The panel’s recommendations (based on at least 80% consensus among experts) are summarized in Table 1.

Table 1 Summary of the recommendations based on ≥ 80% agreement among experts
Full size table

Discussion

Update on role of DCE-MRI and DWI

The literature search described in the “Materials and methods” section resulted in 271 relevant publications. A flowchart summarizing the selection process is shown in the supplementary materials. One extra study was added to this list after review of the references from the retrieved studies. Thus, a total of 26 studies (25 + 1 retrieved) were identified that evaluated the accuracy of DCE-MRI and/or DWI for primary staging of EC. No prospective randomized clinical trials or meta-analyses of prospective clinical trials were identified. The majority were well-conducted, non-randomized, single-centre comparative studies and cohort studies and/or retrospective studies.

Role of DCE-MRI

Multiple studies have compared the performance of combined DCE-MRI + T2WI versus T2WI alone to stage EC, focusing in particular on the depth of myometrial invasion (MI) [8, 9, 11,12,13, 16,17,18,19,20,21,22,23,24,25,26] (Table 2). A meta-analysis demonstrated that DCE-MRI had similar sensitivity to T2WI but was more specific for the detection of deep MI [27]. On DCE-MRI, MI was best depicted during the equilibrium phase (2 min 30 s after the injection) [28, 29]. The panel discussed the value of single-phase high spatial resolution contrast-enhanced imaging at 2 min 30 sec versus DCE-MRI. DCE-MRI allows to determine the presence of uninterrupted enhancement of the subendometrial zone which is best seen approximately 35–40 s following contrast injection. This information is useful when fertility-sparing management is being considered because it helps to exclude any MI, a key finding to confirm patient eligibility for conservative management [30]. Moreover, delayed DCE-MRI images (4–5 min after the injection) are optimal for the detection of cervical stromal invasion (CSI). On the basis of the above considerations, the panel recommends performing either DCE-MRI or single-phase contrast-enhanced imaging at 2 min 30 s depending on the provided clinical history (i.e. patient’s age and desire for fertility preservation) and the radiologist’s availability to supervise image acquisition (Figs. 1 and 2).

Table 2 Summary of the recent publications regarding the role of contrast-enhanced imaging for local staging of newly diagnosed endometrial cancer
Full size table
Fig. 1
figure 1

Recommended MR imaging algorithm for patients with new diagnosis of endometrial cancer a or b patients of childbearing age who are being considered for fertility preservation

Full size image
Fig. 2
figure 2

Assessment of the depth of myometrial invasion (MI) in challenging cases

Full size image

Role of DWI

Multiple studies have demonstrated the added value of DWI for EC staging, particularly for assessing the depth of MI [8, 9, 11,12,13, 18, 20,21,22,23, 26, 30,31,32,33,34] (Table 3). DWI is particularly useful in patients who cannot receive intravenous injection of gadolinium-based contrast agents or have tumors that are isointense or hyperintense to the myometrium on contrast-enhanced images [18, 35]. DWI is also useful in evaluating the depth of MI in the setting of concurrent adenomyosis [13]. The ESUR panel recommends including DWI to stage EC; at a minimum the acquisition should include an axial oblique plane with the same orientation as axial oblique T2WI (i.e. perpendicular to the long axis of the uterus).

Table 3 Summary of the recent publications regarding the role of DWI for local staging of newly diagnosed endometrial cancer
Full size table

The ESUR panel endorses the National Cancer Institute and prior ESUR consensus recommendations that advise a minimum of two b values with an optimal high b value of 800 to 1000 s/mm2 [36, 37].

Several recent studies have compared DWI and DCE-MRI for the assessment of MI. Some found DWI to be superior to DCE-MRI; for example, Takeuchi et al. reported that DWI had the accuracy of 94% while DCE-MRI had the accuracy of 88% for detecting deep MI [13, 18, 20, 35]. Other studies found no difference in the accuracy between the two techniques [11]. A meta-analysis by Andreano et al. reported no significant difference in the sensitivity or specificity between DWI and DCE-MR for diagnosing deep MI [10]. A more recent and larger meta-analysis by Deng et al. confirmed similar diagnostic performance of DWI to DCE-MRI; however, they also found that combined T2WI + DWI were superior to either DWI or DCE-MRI alone [38].

Tips for MRI interpretation

Diagnosis

EC is typically intermediate in signal intensity (SI) on T2WI and is hyperintense compared with the myometrium.

FIGO MRI stage IA/IB

Stage IA is diagnosed if the tumor invades less than 50% of the myometrial thickness whereas stage IB is present if the tumor involves 50% or more of the myometrial thickness [39].

The depth of MI is best measured on the axial oblique images acquired perpendicular to the endometrial cavity. First, a line is drawn parallel to the presumed inner edge of the myometrium. Then, two lines are drawn: one measuring the entire thickness of the myometrium and the other measuring the maximum tumor extension into the myometrium. The ratio of the two lengths corresponds to the depth of MI [40]. The assessment of the depth of MI may be a challenge (Figs. 3 and 4) if a large endometrial tumor distends and thins the myometrium (Fig. 3), if an endometrial tumor is relatively isointense to myometrium on T2WI (Fig. 4), if a tumor involves a cornu of the uterus where myometrium is physiologically thinner than elsewhere in the uterus (Fig. 4), or if the uterine anatomy is distorted by leiomyomas and/or adenomyosis. In these clinical scenarios, radiologists should be aware of possible tumor overstaging; under these circumstances DCE-MRI and DWI may be of particular value to improve the delineation of tumor margins and to avoid the overestimation of tumor extent.

Fig. 3
figure 3

a and b. Axial-oblique T2WI (a) and fused axial oblique T2-DWI (b) show a large tumor distending the endometrial cavity (white arrows) and compressing the myometrium; a continuous low signal intensity junctional zone is seen on both sets of images. c A smooth uninterrupted band of early subendometrial enhancement on DCE-MRI helps to exclude myometrial invasion (arrows)

Full size image
Fig. 4
figure 4

Axial oblique T2WI (a) shows a large endometrial mass involving the right cornu of the uterus (white arrow). The tumor (black arrow) is isointense to the myometrium making it difficult to detect its margins. On T2WI, the tumor possibly extends to and abuts uterine serosa (white arrow). Contrast-enhanced T1WI (b) and fused axial T2WI-DWI (c) improve the delineation of tumor margins and show tumor extension into the outer half of the myometrium but no involvement of the uterine serosa consistent with FIGO IB disease (white arrow)

Full size image

FIGO MRI stage II

CSI is best assessed by evaluating both the sagittal and axial oblique planes that are acquired parallel and perpendicular to the long axis of the cervix, respectively. CSI is diagnosed when intermediate-SI tumor disrupts low-SI fibrous cervical stroma (CS) on T2WI. On DCE-MRI, CSI is indicated when the normal enhancement of CS is disrupted by a hypo-enhancing tumor, best seen on delayed phase images (4–5 min). On DWI, CSI is suspected when a tumor (high SI on high b value DWI and low SI on the apparent diffusion coefficient (ADC) map) disrupts low-SI CS. Potential pitfalls for assessing CSI on MRI are summarized in Fig. 5.

Fig. 5
figure 5

Detection of cervical stromal invasion

Full size image

FIGO MRI stage III

  • Stage IIIA tumors invade the uterine serosa. They appear as intermediate-to-high-SI lesions that disrupt normally smooth outer contour of the uterus on T2WI. One should be careful not to overcall stage IIIA when the tumor is isointense to the myometrium on T2WI. Contrast-enhanced imaging and DWI can improve the delineation of tumor margins and facilitate MR staging (Fig. 4). Stage IIIA also includes direct tumor spread to the adnexa or ovarian metastases.

  • Stage IIIB tumors involve the parametria or the vagina by either direct invasion or metastatic spread. DWI is particularly useful for detecting small tumor deposits in the cervix and/or vagina.

  • Stage IIIC disease is characterized by the presence of lymph node metastases and is subdivided on the basis of pelvic (stage IIIC1) and/or para-aortic (stage IIIC2) lymph node involvement. The risk factors for lymph node metastases include presence of high-risk histologic subtypes (grade 3 endometrioid adenocarcinoma and non-endometrioid histologic types, i.e. carcinosarcoma, serous carcinoma or clear cell carcinoma), lymphovascular space invasion, deep MI, and CSI [41, 42]. MRI has low sensitivity for the detection of lymph node metastases [43]. The assessment is largely based on size criteria where a short axis diameter of greater than 8 mm in pelvic nodes and 10 mm in para-aortic nodes is taken to indicate tumoral involvement [29, 43]. Other morphological features including round shape, spiculated margins, heterogeneous SI, SI similar to that of the primary tumor, or the presence of necrosis can be used to suggest the involvement in smaller nodes [44]. DWI aids in the detection of lymph nodes owing to their high SI on high b value images. However, there is significant overlap between the ADC values of benign and malignant nodes and therefore the technique is currently used only as an adjunct to T2WI [45,46,47,48,49].

FIGO MRI stage IV

Stage IV disease manifests with direct invasion of the bladder or rectal mucosa (stage IVA) or distant metastases (stage IVB). Preserved fat planes between the tumor and bladder or rectum exclude stage IVA with high accuracy, alleviating the need for cystoscopy or rectosigmoidoscopy.

Key components of MRI report (Fig. 6)

The panel members unanimously agreed on the need for a structured MRI report because structured reporting improves the report quality and facilitates the communication of clinical relevant information to a referring physician [37, 50,51,52,53,54,55]. The recommended structured report is presented in Fig. 6; the elements of the report are summarized as a short mnemonic, “Clinical and MRI Critical TEAM” (Clinical for clinical information, MRI for MI, Critical for CSI, TE for tumor extension, A for adnexa and M for metastasis).

Fig. 6
figure 6

Endometrial cancer structured report: “Clinical and MRI Critical TEAM”

Full size image

Role of MRI in initial clinical decision-making

Role of MRI in selection of patients prior to fertility-sparing therapy

Approximately 5% of women are diagnosed with EC under the age of 40 [56]. If a patient is of childbearing age, desires fertility preservation and has endometrium-confined grade 1 endometrioid adenocarcinoma or premalignant conditions (for example, atypical hyperplasia), conservative medical treatment with progestins (administered orally or via an intra-uterine device) may be an option. Fertility-sparing management is controversial in patients with Lynch syndrome since their disease is due to genetic predisposition and may be less responsive to progestins [57]. In addition, there is evidence that patients with BMI greater than 25 kg/m2 before or after progestogen treatment have a worse response to treatment and a high recurrence rate [58]. MRI is useful prior to initiating conservative management to confirm that the disease is confined to the endometrium.

The eligibility criteria for fertility-preserving management are summarized in Table 4.

Table 4 Eligibility criteria prior to the fertility-sparing treatment in patients with endometrial cancer who desire fertility preservation
Full size table

Role of MRI for initial treatment planning

The standard surgical procedures that simultaneously stages and treats EC include total hysterectomy, bilateral salpingo-oophorectomy with peritoneal washings, and pelvic plus para-aortic lymph node dissection. However, most patients present with FIGO stage I disease and are at low risk for lymph node metastases. The clinical benefit of lymphadenectomy in early-stage EC is controversial. Lymphadenectomy allows complete surgical staging and facilitates adjuvant treatment selection, potentially reducing the morbidity of unnecessary radiation therapy. However, lymphadenectomy carries a 7–10% risk of lymphocele development and a 23% risk of lower-extremity lymphedema [59]. Several recent large prospective trials showed no survival benefit after lymphadenectomy in patients with early-stage grade 1 and 2 endometrioid adenocarcinoma [60,61,62,63]. Therefore, in patients with clinical stage I disease, the need for lymphadenectomy may be determined based on the presence of risk factors that increase the likelihood of lymph node metastases and subsequent recurrence [63,64,65,66].

The ESUR endorses the ESMO recommendations to stratify stage I EC into four risk categories [2, 4]. Accordingly, lymphadenectomy is not recommended in the low-risk group, i.e. stage I grade 1 or 2 endometrioid adenocarcinoma with less than 50% MI [2, 4]. Lymphadenectomy is suggested or recommended for all other patients with newly diagnosed EC. In this schema, preoperative information regarding the depth of MI and histologic subtype is essential to tailor the surgical approach. MRI can assess the depth of MI, while histologic type and grade are determined by endometrial sampling [67].

The panel recommends a tailored MR imaging assessment that is closely aligned with the ESMO guidelines. MRI protocol should be tailored according to the tumor histology and grade, patient preferences with regard to fertility preservation and the radiologist’s availability to monitor image acquisition (Fig. 1). Briefly, patients with grade 3 endometrioid adenocarcinoma and non-endometrioid histologies (carcinosarcoma, serous carcinoma or clear cell carcinoma) are at high risk of extra-uterine spread including lymph node metastases. In this group, assessment for MI or CSI with MRI is less important, while the detection of extra-uterine disease is critical for treatment planning. DWI through the entire abdomen and pelvis should be performed; contrast-enhanced imaging is recommended but DCE-MRI is not required. In patients with grade 1 or 2 endometrioid adenocarcinoma, MRI should focus on evaluating the depth of MI and presence of CSI. If image acquisition is unsupervised, DWI and contrast-enhanced are recommended to assess MI and CSI. If image acquisition is supervised, DCE-MRI may only be necessary as an adjunct to DWI in challenging cases. In patients of childbearing age who desire fertility preservation and have grade 1 endometrioid cancer, DCE-MRI should be added to T2WI and DWI because the presence of intact subendometrial enhancement is useful to confirm endometrium-confined disease.

Future research directions

DWI

Histologic grade is determined preoperatively using endometrial sampling but is subject to sampling error [68,69,70]. Several studies investigated the role of DWI with ADC for assessing tumor grade, but the results are inconclusive so far [9, 18, 20, 71,72,73,74,75,76,77,78,79,80,81,82,83,84,85]. Some found an association between low ADC values and high-grade histology [9, 72, 77, 79], while others did not [71, 73, 74]. FOCUS DWI (field of view (FOV) optimized and constrained undistorted single-shot DWI) [8, 34, 86] is a new approach to DWI acquisition that uses the reduced FOV in the phase encoding direction to minimize artefacts; the results of using FOCUS DWI to assess disease extent in patients with uterus-confined disease are encouraging [8].

PET/CT and PET/MRI

18F-FDG PET/CT is superior to MRI for N and M staging [87]. However, the yield of PET/CT is low in early-stage EC as a result of the low prevalence of lymph node metastases [88]. The present ESUR guidelines endorse the National Comprehensive Cancer Network guidelines that recommend PET/CT for initial staging only in the presence of clinical suspicion for extra-uterine spread [89].

It is hoped that PET/MR will allow the integration of morphologic, functional and metabolic information to facilitate the evaluation of both local and distant extent of disease.

Other new MRI tools

Radiomics is the emerging field that correlates image-based texture features with clinically relevant oncologic outcomes. The role of texture analysis (TA) in EC is an area of active investigation [90]. Preliminary data suggest that quantitative texture features may be useful for evaluating the depth of MI, detecting lymphovascular space invasion and identifying high-grade histology. For evaluating the depth of MI, quantitative texture features show similar accuracy to subjective interpretation by experienced radiologists. Tumor size and tumor volume as determined on MRI are also useful for evaluating the depth of MI and detecting the presence of lymphovascular space invasion [9, 76, 91]. These preliminary findings require further confirmation and external validation. The development of robust auto-segmentation techniques is an active area of research and would facilitate the application of TA in daily clinical practice.

Furthermore, advances in molecular profiling of EC may facilitate new research in radiogenomics. Radiogenomics refers to the study of the relationships between imaging phenotypes and genomic signatures. The Cancer Genome Atlas Research Network has recently improved the characterization of the EC molecular landscape whereby four molecular subtypes have been described: (1) POLE, the smallest group with excellent prognosis, (2) microsatellite unstable tumors, (3) copy-number low microsatellite stable tumors and (4) copy-number high tumors with TP53 mutations. The last group includes serous carcinomas and is associated with a poor prognosis [5, 92,92,93,95]. As such, future radiogenomic studies may lead to path-breaking changes in EC imaging.

Summary

MRI is now widely accepted as the imaging modality of choice for initial staging of EC. Recently, DWI has emerged as a promising tool to facilitate the assessment of local disease extent. These updated ESUR Female Pelvis Imaging working group guidelines build on the 2009 version and the recent ESMO guidelines to reflect new imaging and clinical developments in the field. This update recommends an algorithmic approach to MRI acquisition, addresses patient evaluation prior to fertility-sparing management and proposes a structured MRI report to facilitate effective communication between radiologists and their referring physicians.