Novelties in treatment planning
In 2009 the FIGO staging of EC has been updated, and FIGO stage Ia and Ib have been merged, because it has been demonstrated that the prognosis for tumor confined to endometrium or tumor invading the inner half of the myometrium was comparable [
3], thus reducing the subgroups of stage I to Ia (EC confined to endometrium or invading the inner half of the myometrium) and Ib (EC invading the outer half of the myometrium).
Furthermore, it has become more and more evident that prognosis of EC patients depends on multiple different factors, most of which not included in the FIGO staging, such as histological type and grade, age of the patient, size of the tumour, lymphovascular spaces invasion. Therefore, on one hand many multicentric trials and society guidelines have attempted to include multiple prognostic factors in classification of EC [
4]; on the other hand, recent advances in molecular classification of EC by histology has raised the evidence that different outcomes may be related to different expression of specific mutations and alterations, such as TP53, PI3K, KRAS, ERBB, FGFR and others [
4].
Additionally, there has been a long debate about performance of lymphadenctomy. Indeed, lymph nodal involvement is the most common form of extrauterine disease spread, related to deep invasion (> 50%) of the myometrium, grade (G3) or non-endometrioid histologies [
5], and is the strongest predictor of recurrence. After conflicting results, such as those of two randomized trials, showing no therapeutic benefit of lympadectomy [
6,
7], and a retrospective study showing a better overall survival rate in intermediate and high-risk EC patients who underwent systematic lymphadenectomy [
8], sentinel lymph node mapping has been proposed as surgical technique to assess lymph nodal status. This represents an innovative technique to identify lymph node metastases, while reducing the surgical morbidity (lymphoedema, lymphocysts) associated with systematic lymphadenectomy [
9]. It has to be noted that some centres, especially in the US, thanking also the advent of the sentinel node mapping during surgery, do not perform routinely MR for pre-operative staging. However, MRI is specifically recommended during the initial workup of EC patients, in the US as well as in Europe, to establish the origin of a tumor (endocervical or endometrial), as well as when planning a fertility-sparing treatment, in order to confirm the initial stage by excluding myometrial invasion, as well as adnexal or pelvic nodes involvement [
2,
9].
Finally, minimally invasive surgery, including robotic assisted surgery, has been introduced, as associated with significantly fewer moderate-to-severe postoperative adverse events (14% versus 21%) and a lower frequency of hospitalisations > 2 days (52% versus 94%) than laparotomy [
2,
10]. For surgical planning, an important factor not significantly changed overtime, is cervical stromal invasion, that may require radical hysterectomy instead of total hysterectomy (performed in patients with tumours confined within the uterus [
11]), which may represent a helpful information for planning operating time, especially with minimally invasive techniques.
Novelties in imaging evaluation
The most common initial symptom (75%) of EC in postmenopausal women is abnormal uterine bleeding, and this allows the possibility of making an early diagnosis.
Gynaecological examination is usually completed by transvaginal ultrasound (TVUS), which usually represents the first imaging evaluation. TVUS is performed with a dedicated vaginal probe using high frequencies and a small field of view, which allows good local evaluation. Many studies indicate an endometrial thickness of 5 mm as the normal cut-off value in post-menopausal women [
4], but in the presence of bleeding this cut-off value may be lowered to 3 mm [
12]. Three-dimensional TVUS (3D-TVUS) is a novel technique which acquires US images that may be reconstructed in any desired plane, to better depict myometrial invasion in the uterine corners [
13]. However, results are equivocal and 3D-TVUS has not yet proved superior to conventional 2D-TVUS in the assessment of myometrial involvement and identification of cervical infiltration [
14]. When TVUS is performed in a referral centre, it is not always requested to perform an additional imaging examination for evaluation of local extent, such as MR, and the patient may complete the pre-operative staging with a CT scan.
MR with paramagnetic contrast agent is considered an accurate imaging technique for local staging, because the tumoral tissue, the endometrium and the myometrium show different MR signals, especially on T2w images. However, since these differences may be less evident in post-menopausal women, acquisition of dynamic images after injection of contrast medium is recommended [
15]. The European Society of Uro-genital Radiology has suggested in 2009 a dedicated MR protocol for accurate local staging of endometrial carcinoma [
15], based on T2-w acquisitions in three orthogonal planes oriented to the uterine cavity (sagittal, axial oblique and coronal oblique), with additional sequences oriented to the long axis of the endocervical canal. The slice thickness recommended is 3–4 mm with a small FOV of 20–25 cm. In order to detect nodal involvement, coronal T1-w with extended FOV is also suggested. Finally, para-axial dynamic T1-w after intravenous bolus injection of a paramagnetic contrast agent is advocated for optimal evaluation of myometrial invasion.
After publication of the abovementioned guidelines, many advances in knowledge of EC behaviour and in imaging technology have been achieved, with the consequence of some novelties in patient management.
In literature, several articles have been published about the role of 3 T MR in the preoperative evaluation of depth of myometrial infiltration and accuracy has been reported equivalent to that of 1.5 T MR [
16]. MR imaging using 3 T equipment, with a higher Signal-to-Noise Ratio (SNR) than 1.5 T examinations, may lead to a more advanced temporal and spatial resolution, although it is also known that at 3 T the quality of abdominal imaging may be influenced by some artefacts. In fact, there are some known restrictions to the use of a higher magnetic field, such as an increased T1 relaxation time, a reduced T2 relaxation time and a higher magnetic susceptibility and chemical shift artifacts [
17].
At the time of publication of these guidelines, Diffusion-Weighted Imaging (DWI) was indicated as a promising tool in assessing gynecologic cancers, providing information on tumour cellularity and microenvironment. After some years of experience, DWI is currently widely used as an adjunct to T2w and dynamic imaging in routine clinical practice [
18]. Indeed, it is now well known that DWI is a helpful tool in detecting EC when endometrial biopsy is technically impossible, due to cervical stenosis, as well as when histopathologic results are inconclusive, because ADC values are significantly lower in cancer than normal endometrium or benign polyps [
19,
20]. A recent systematic review and meta-analysis evaluating pre-operative staging of myometrial invasion in EC patients has shown that in different studies published between 2008 and 2013, DWI sequences were performed with different high b-values ranging between 500 and 1000, where the highest b-value (b = 1000) was mainly on 3 Tesla machines [
21‐
23], whereas on 1.5 Tesla machines the highest b-value used was more frequently b = 800 [
24,
25], with few exceptions [
26].
It is also under consideration if the combination of T2w images and DWI may be superior to dynamic contrast-enhanced MR. Indeed, a meta-analysis including 9 studies (442 patients) comparing the diagnostic accuracy of dynamic contrast-enchanced (DCE) MR imaging and DWI to predict deep myometrial invasion, showed no differencies in pooled sensitivities and specificities of the two sequences [
27]. However, in a more recent meta-analysis including 15 studies and 849 patients, the specificity of T2w images plus DWI was superior to that of DCE MR imaging [
28].
Evidences suggest that fusion T2-DWI imaging at 3-T has an 88% accuracy in the assessment of the depth of myometrial invasion, with a similar diagnostic performance as DCE MRI [
29]. In addition, the use of a reduced FOV for DWI imaging may improve the local staging accuracy of MRI in the assessment of the depth of myometrial invasion; in particular for patients with coexisting adenomyosis, it can be a helpful alternative diagnostic tool to 3D DCE MRI [
30]. Field-of-view (FOV) optimized and constrained undistorted single shot (FOCUS) is a new sequence clinically applied by using a two dimensional spectral spatial excitation radiofrequency pulse, which limits the signal to a small FOV, thus excluding possible sources of artifacts that are outside the region of interest [
30]. This may avoid local field homogeneity conditions, and minimize off-resonance artifacts due to B0 inhomogeneity, susceptibility gradients, eddy currents, and chemical shift [
31,
32]. Reduced FOV DW imaging with high spatial resolution may enhance the lesion detection, and the evaluation of tumor extent in endometrial cancer. Indeed, severe geometric distortion at air-tissue interfaces around the uterus, ovaries and rectum, can be minimized on reduced FOV DW imaging. However, since the small FOV may be a limit for the detection of lymph node metastases, peritoneal dissemination and/or bone metastases, reduced FOV DW imaging is considered as an additional sequence to the traditional large FOV DW sequence, for the evaluation of local tumor extent [
30].
Bhosale et al. reported the feasibility of reduced FOV DW imaging in the assessment of myometrial invasion in patients with clinical FIGO stage I endometrial cancer. In this study the authors demonstrated the feasibility of IVIM parameters as imaging markers of MSI status (Micro-Satellite Instability), defined as a germline mutation of mismatch repair system (DNA polymerase) that leads to a higher risk to develop EC. Furthermore, compared with T2-weighted imaging plus DCE MR, reduced FOV DW imaging yielded greater specificity and accuracy for deep myometrial invasion determined by one reader, and greater the sensitivity for deep myometrial invasion determined by another reader [
31].
Recently there has been a growing need for a non-invasive imaging method for the accurate diagnosis and differentiation of endometrial malignancy. With this regard, it has been demonstrated that DW signal and ADC values can be influenced not only by molecular proton diffusion, but also by microcirculation or blood perfusion. Some Authors have reported the feasibility to study the tissue characteristics of endometrial cancer using IVIM model at 3 T and to assess their diagnostic potential [
32]. In this study, ADC values derived from a mono-exponential model of DWI containing perfusion effect were demonstrated to limit the reliability of ADC in characterizing lesions [
33]. The IntraVoxel Incoherent Motion (IVIM) has been used to estimate perfusion in tissues, as blood flow in randomly oriented capillaries mimics a pseudo-diffusion process. By using IVIM-based perfusion MRI, microcirculation or perfusion effects can be distinguished from true tissue diffusion, when using a sufficient b value sampling and bi-exponential curve fit analysis with the intravoxel incoherent motion (IVIM) model [
34]. IVIM MRI can be considered as an extension of DWI that enables the simultaneous acquisition of both microcirculatory and diffusivity information. The capillary network lacks the spatial orientation and the molecular motion within the microcirculation; this can be viewed as “pseudo-diffusion”. This pseudo-diffusion is significant at the lower b-value range (< 200 s/mm2) and the following two quantitative perfusion parameters can be derived from analysing the steeper part of the diffusional signal decay curve: the fractional perfusion related to the microcirculation, and the pseudo-diffusion coefficient that measures the incoherent perfusion-related diffusion within the microcirculation. The primary advantage of IVIM is that it enables the simultaneous acquisition of diffusion and perfusion parameters and therefore can provide both measurements without the requirement of a further co-registration processing step [
18,
35]. The perfusion characteristic differs between tissue types in the female pelvis: tumoral perfusion affects the sensitivity to chemo-radiotherapy, whereas altered vasculatures and variable levels of aniogenic factors within gynecological disease are thought to determine therapeutic success following interventional therapy. Endometrial cancer has low perfusion and high diffusion IVIM characteristics with promising potential for early non-invasive diagnosis, hence proving potentially useful for tissue differentiation.
CT has a low sensitivity (83%) and specificity (42%) in detecting and evaluating myometrial involvement, as well as in assessing cervical stromal invasion [
36]. However, dual-energy CT evaluation of deep myometrial invasion has been suggested [
37], and CT is widely recognized as helpful in assessing distant metastases.
PET/CT is more sensitive than CT or MR imaging for detection of nodal metastases [
38,
39]. Furthermore, in patients for whom biopsy demonstrates high-risk histology, PET/CT may be used to identify unsuspected distant disease that would obviate the morbidity of a staging operation. Two meta-analyses have indeed shown that PET/CT is highly sensitive and specific for distant nodal metastasis [
40‐
42], although a significant percentage of lymph nodes still go undetected by PET/CT. In a direct comparison of MR imaging and PET/CT for initial staging, PET/CT was found superior for detection of metastatic lymph nodes (sensitivity 70% for PET/CT vs 34% for MR imaging) [
38].
Recently introduced PET/MR scanners acquire MR and PET data either simultaneously or sequentially. In gynaecologic cancer patients, 18F-FDG PET/MR protocols are intended to provide a sole examination to solve the problems of treatment planning, as well as for evaluation of prognosis and surveillance. Indeed, in simultaneous PET/MR scanners, during the PET acquisition, whole-body Dixon images, the anatomically descriptive half-Fourier acquisition single-shot turbo spin-echo images, and fluid-sensitive inversion recovery images and DWI are co-acquired; these acquisitions are then completed for local staging by a dedicated pelvic MR including dynamic intravenous gadolinium administration. MR imaging and PET are complementary in the initial staging of endometrial cancer because MR is more accurate for evaluation of local extent of the tumour, whereas PET is more accurate for distant metastases. Lesions detected on PET or DWI can be precisely localized and characterized on the conventional MR sequences, thereby allowing for improved sensitivity without a loss of specificity. In a direct comparison of fused PET/MR imaging, MR imaging, and PET/CT data, PET/MR imaging and MR imaging alone were both more accurate for local staging of endometrial cancer (T-stage accuracy 80% for PET/MR imaging or MR imaging alone vs 60% for PET/CT) [
39].
For association with prognosis after therapy, both PET and MR imaging features of endometrial cancer have been evaluated [
40]. The preoperative maximum SUV has been correlated with the presence of high-risk or low-risk disease after surgical staging [
43]. Accordingly, tumors with high-risk features, such as deep myometrial invasion, cervical invasion, lymphovascular space involvement, and lymph node metastases, showed high maximum SUV and low minimum ADC [
44].