{"paper_id":"a3cafb07-c64d-4750-a46d-5b9c07a6e02c","body_text":"Endometrial cancer is the leading gynecological cancer in high-income countries and\nthe sixth most common in women worldwide ( 1 , 2 ) . More than 90% of all cases of endometrial\ncancer occur in women > 50 years of age, whereas only 4% occur in women < 40\nyears of age ( 2 ) . Endometrial cancer is typically diagnosed at an\nearly stage (80% of cases being diagnosed in stage I) during the investigation of\natypical bleeding in postmenopausal women, with five-year survival rates of over 95%\nin such cases ( 3 ) . However, overall five-year survival varies widely,\nfrom 20% to 95%, the prognosis mainly depending on three\nfactors ( 4 ) : histological subtype and grade; local tumor stage\nat diagnosis; and the presence or absence of lymph node metastases. Although the\ngrade and histological subtype of endometrial cancer can be diagnosed through\nendometrial sampling, tumor staging is traditionally performed intraoperatively,\naccording to the International Federation of Gynecology and Obstetrics (FIGO)\nguidelines, which include the use of total abdominal hysterectomy, bilateral\nsalpingo-oophorectomy, peritoneal lavage, and pelvic/para-aortic lymphadenectomy,\ndepending on the findings at intraoperative staging ( 5 ) .\nThe depth of myometrial invasion represents the morphological feature with the\ngreatest prognostic value in endometrial cancer, correlating with tumor grade, lymph\nnode metastases, and overall patient survival ( 4 , 6 ) . Tumor invasion to greater than 50% of the\nmyometrial thickness translates to a six to seven times greater risk of pelvic and\npara-aortic lymph node metastases, and patients should be considered candidates for\na more aggressive surgical approach ( 7 ) . However, the value of routine lymphadenectomy\nin early endometrial cancer remains controversial. Two large prospective studies,\ncollectively including approximately 2000 women, demonstrated no benefit of pelvic\nlymphadenectomy in terms of overall and recurrence-free survival in women with\nearly-stage (IA) preoperative endometrial cancer, in comparison with less invasive\nsurgical approaches ( 8 , 9 ) . Accurate preoperative assessment of the depth of\nmyometrial invasion and pathological staging is therefore crucial to the planning of\nthe appropriate surgical approach ( 10 ) .\nMagnetic resonance imaging (MRI) has proven to be the most effective technique in the\npreoperative evaluation of endometrial cancer, showing great accuracy in the\nassessment of myometrial invasion, cervical stromal invasion, and lymph node\nmetastases ( 11 , 12 ) . Although MRI was not formally incorporated into\nthe revised FIGO staging system for endometrial cancer, it is widely used to assess\nthe stage of the disease and to plan the appropriate therapeutic\napproach ( 4 , 8 ) . The standard MRI protocol includes high-resolution\nT2-weighted imaging (T2WI), in various planes, and multiphase dynamic\ncontrast-enhanced (DCE)-MRI ( 13 ) . However, there is no consensus regarding the\nbest protocol, and recent studies have produced contradictory results, demonstrating\nno significant added value of DCE-MRI, either in the assessment of myometrial\ninvasion or in the staging-nor have there been any reported differences between T2WI\nand DCE-MRI+T2WI in terms of interobserver agreement ( 4 , 6 ) .\nSimultaneously to the technical improvements and growing general interest in the use\nof diffusion-weighted imaging (DWI) in the evaluation of the female pelvis, studies\nhave obtained encouraging results with the use of DWI in the preoperative assessment\nof endometrial cancer ( 11 , 13 ) . However, further studies are needed in order to\ndemonstrate its added value and to consolidate its use in clinical practice. DWI is\na functional imaging technique that provides information about water mobility,\ntissue cellularity, and the integrity of cellular membranes ( 14 ) . On DWI, endometrial\ncancer demonstrates restricted diffusion in comparison with that of normal\nmyometrial tissue, resulting in high signal intensity at high b-values (500-1000\ns/mm 2 ) and low apparent diffusion coefficient (ADC)\nvalues ( 2 , 14 ) . ADC values are also significantly lower in\nendometrial cancer than in normal endometrium or in benign conditions such as\nendometrial polyps, leiomyomas, and endometrial hyperplasia ( 15 , 16 ) .\nThe purpose of this study was to evaluate the added value of DWI (as an adjunct to\nT2WI) in the preoperative assessment of endometrial cancer. Our main hypothesis was\nthat DWI+T2WI would have greater diagnostic accuracy in the assessment of the depth\nof myometrial invasion in endometrial cancer than would T2WI alone or\nDCE-MRI+T2WI.\n\nThis was a retrospective, single-center study including 44 women with a median\nage of 68 years (range, 44-78 years). All of the women had surgically confirmed\nprimary endometrial cancer and had undergone pelvic MRI (including T2WI,\nDCE-MRI, and DWI) as part of their preoperative evaluation. Patients who had\nundergone preoperative MRI and/or underwent surgery at other institutions were\nexcluded. The institutional review board approved this study and waived the\nrequirement for written informed consent.\nThe surgical procedures were performed at a dedicated oncology center by surgeons\nwith more than 10 years of experience in the treatment of gynecological\nmalignancies. All of the operations were performed 3-8 weeks after the\npreoperative MRI examination. The result of the histopathological examination,\nwhich was performed by a pathologist who specialized in gynecology/oncology and\nhad more than 20 years of experience, constituted the diagnostic standard for\ncomparison. The flow of the patients through the study is depicted in  Figure 1 .\nFigure 1 Flow of participants through the study, including the acquisition of\nthe MRI sets T2WI ( A ), DCE+T2WI ( B ), and\nDWI+T2WI ( C ).\nFlow of participants through the study, including the acquisition of\nthe MRI sets T2WI ( A ), DCE+T2WI ( B ), and\nDWI+T2WI ( C ).\nThe MRI studies were performed in a 1.5 T MRI scanner (Intera Pulsar; Philips\nMedical Systems, Best, The Netherlands) with an 8-channel phased-array body coil\nand saturation bands (anterior and superior). Patients were asked to fast for 4\nh before the examination. To reduce bowel motility and peristaltic artifacts,\nN-butylscopolamine bromide (20 mg) was administered via intramuscular injection\nbefore the MRI examination. During the examination, patients were placed in the\nsupine position.\nThe MRI protocol included axial imaging of the abdomen for evaluation of advanced\ndisease, with fast spin-echo T2WI (slice thickness, 6 mm; interslice gap, 1 mm;\nand breath-hold), and DWI with an echo-planar technique (slice thickness, 6 mm;\ninterslice gap, 1 mm; and b-values of 0, 500, and 1000 s/mm 2 ,\ntogether with the respective ADC maps). The pelvic evaluation included fast\nspin-echo T1-weighted imaging obtained in the axial plane (slice thickness, 4\nmm; and interslice gap, 0.4 mm) and fast spin-echo T2WI obtained in three\nplanes-the axial plane (slice thickness, 4 mm; and interslice gap, 0.4 mm), the\nsagittal plane (slice thickness, 4 mm; and interslice gap, 0.4 mm), and the\naxial oblique plane relative to the major axis of the uterine body (slice\nthickness, 4 mm; and interslice gap, 0.4 mm).\nFor DCE-MRI, a 3D fat-suppressed gradient echo T1-weighted sequence (slice\nthickness, 3 mm; and interslice gap, 0.5 mm) was acquired after intravenous\ninjection of gadopentetate dimeglumine (0.1 mmol/kg of body weight-Magnevist;\nBayer HealthCare AG, Leverkusen, Germany) at a rate of 2 mL/s. Images were\nobtained at 0, 25, 60, 120, and 150 s in the axial oblique plane perpendicular\nto the major axis of the uterine body and at 240 s in the axial plane. DWI with\nan echo-planar technique (slice thickness, 4 mm; interslice gap, 1 mm; and\nb-values of 0, 600, and 1000 s/mm 2 , with the respective ADC maps) was\nacquired in the axial plane. When there was suspicion of cervical invasion,\nadditional axial oblique T2WI was obtained perpendicular to the cervical canal\n(slice thickness, 4 mm; and interslice gap, 0.4 mm), to evaluate cervical and\nparametrial invasion, together with DCE-MRI, obtained at 0, 25, 60, 120, and 150\ns in the sagittal plane and at 240 s in the axial oblique plane perpendicular to\nthe cervical canal.\nTwo radiologists specializing in urogenital radiology, with 5 and 22 years of\nexperience, respectively, both of whom were blinded to the histopathological\nreports, evaluated the MRI scans and performed a consensus interpretation of the\ndepth of myometrial invasion and stage of the cancer based on the FIGO staging\nsystem, considering three sets of sequences: T2WI, DCE-MRI+T2WI, and DWI+T2WI.\nIn all patients, the observers started with the analysis of the T2WI sequence,\nevaluating the depth of myometrial invasion and the predictable stage of disease\n(considering cervical stromal invasion, vaginal/pelvic involvement, and lymph\nnode or distant metastases). The observers then analyzed the DCE-MRI+T2WI\nsequences and evaluated the same parameters. Finally, the observers analyzed the\nDWI+T2WI (cognitive fusion, rather than image fusion) sequences, again\nevaluating the same parameters. In each case, the T1-weighted imaging was\nanalyzed independently, to exclude potential pitfalls such as hemorrhage. The\nmyometrial invasion was classified as superficial if the tumor had invaded <\n50% of the myometrial thickness and deep if it had invaded ³ 50%.\nAll the patients in the study sample underwent total hysterectomy, bilateral\nsalpingo-oophorectomy, pelvic lymphadenectomy, and peritoneal lavage. The\nhistopathological evaluation of the tumor included histological type, tumor\ngrade, and depth of myometrial invasion (superficial or deep). The presence of\ncervical stromal invasion, extension to the serosa or other organs (e.g.,\novaries, fallopian tubes, and peritoneum), and lymph node metastases were also\nevaluated.\nTo determine the depth of myometrial invasion, we considered two criteria. First,\nwe calculated the total thickness of the myometrium, considering the areas that\nwere not invaded, and deep invasion was defined as invasion of half or more than\nhalf of that thickness. Second, we determined whether tumor cells had reached\nthe vascular plexus that separates the two layers.\nAll statistical analyses were performed using R software, version 3.3.3 (a free\nsoftware environment available at https://www.r-project.org). Data are presented\nas absolute and relative values or as proportions and 95% confidence intervals\n(95% CI). The diagnostic accuracy, sensitivity, specificity, positive predictive\nvalue (PPV), and negative predictive value (NPV) were calculated for T2WI,\nDCE+T2WI, and DWI+T2WI. The accuracy of the different sets of sequences in the\nevaluation of deep myometrial invasion was compared with\n p -value adjustment by the Benjamini-Hochberg procedure for\ncontrolling the false discovery rate in multiple comparisons. A similar\nprocedure was performed to compare the accuracy in staging. Values of\n p  < 0.05 were considered statistically significant.\n\nThe demographic characteristics of the patients and the postoperative histological\nfindings are summarized in  Table 1 . Of the 44\npatients, 25 (57%) had superficial myometrial invasion, and 19 (43%) had deep\nmyometrial invasion. Cervical invasion was observed in two patients with superficial\nmyometrial invasion and in four patients with deep myometrial invasion, whereas\nserosal invasion was observed in only one patient (with deep myometrial\ninvasion).\nCharacteristics of and surgical findings in a sample of patients with\nendometrial cancer.\nAs can be seen in  Table 2 , the depth of\nmyometrial invasion was correctly determined with T2WI in 27 (61%) of the 44\npatients evaluated, with DCE-MRI+T2WI in 38 (86%), and with DWI+T2WI in 42 (95%). In\naddition, correct staging was achieved with T2WI in 22 patients (50%), with\nDCE-MRI+T2WI in 34 (77%), and with DWI+T2WI in 41 (93%). DWI+T2WI demonstrated\nhigher diagnostic accuracy, sensitivity, specificity, PPV, and NPV than did T2WI and\nDCE-MRI+T2WI in the assessment of the depth of myometrial invasion ( Table 3 ).\nCorrect identification of the degree of myometrial invasion, by MRI sequence\nset, together with a comparison between the MRI findings and pathologic\nfindings in terms of staging, in patients with endometrial cancer.\nDiagnostic performance of T2WI, DCE-MRI+T2WI, and DWI+T2WI in the assessment\nof the depth of myometrial invasion in patients with endometrial cancer.\nComparing each set of sequences ( Table 4 ), we\nfound that, for myometrial invasion and staging, DWI+T2WI and DCE-MRI+T2WI both\ndemonstrated diagnostic accuracy superior to that of T2WI alone, the differences\nbeing statistically significant ( p  < 0.05 for all). DWI+T2WI\nshowed greater diagnostic accuracy than did DCE-MRI+T2WI, for myometrial invasion\nand for staging, although the differences were not statistically significant\n( p  > 0.05 for both).\nComparison between the MRI sequence sets, in terms of their accuracy in\ndetermining the depth of myometrial invasion and the pathological stage, in\nthe preoperative assessment of endometrial cancer.\nAdjusted with the Benjamini-Hochberg procedure for controlling the false\ndiscovery rate in multiple comparisons.\n\nOur study confirmed that MRI is a powerful tool for the preoperative evaluation of\nendometrial cancer, particularly for the assessment of the depth of myometrial\ninvasion, one of the most important prognostic factors associated with endometrial\ncancer. The interpretation of the diagnostic performance of DWI+T2WI and\nDCE-MRI+T2WI showed that both were superior to T2WI alone in the assessment of the\ndepth of myometrial invasion and staging ( Figures\n2  and  3 ). When comparing the\nfunctional sequences (DWI and DCE-MRI), we found that DWI combined with T2WI\nperformed slightly better than did DCE-MRI combined with T2WI, which implies\npotential advantages of the former combination, because DWI does not involve\nintravenous contrast administration and has shorter acquisition\ntimes ( 2 , 17 ) .\nFigure 2 MRI of a 78-year-old woman with endometrial cancer.  A:  Axial\noblique T2WI, perpendicular to the main axis of the uterus, showing an\nendometrial tumor that the observers judged to be invading the outer\nhalf of the myometrium (arrow).  B:  Axial oblique DCE-MRI\nsequence showing an endometrial tumor (arrow) with signal intensity that\nwas low in comparison with that of the myometrium, classified by the\nobservers as superficial myometrial invasion.  C,D:  Axial\nDWI sequence showing an endometrial tumor (arrows) with high signal\nintensity at a high (1000 s/mm2) b-value ( C ) and low signal\nintensity on the ADC map ( D ), classified by the observers\nas superficial invasion. The postoperative histological findings\nconfirmed the superficial myometrial invasion (stage IA).\nMRI of a 78-year-old woman with endometrial cancer.  A:  Axial\noblique T2WI, perpendicular to the main axis of the uterus, showing an\nendometrial tumor that the observers judged to be invading the outer\nhalf of the myometrium (arrow).  B:  Axial oblique DCE-MRI\nsequence showing an endometrial tumor (arrow) with signal intensity that\nwas low in comparison with that of the myometrium, classified by the\nobservers as superficial myometrial invasion.  C,D:  Axial\nDWI sequence showing an endometrial tumor (arrows) with high signal\nintensity at a high (1000 s/mm2) b-value ( C ) and low signal\nintensity on the ADC map ( D ), classified by the observers\nas superficial invasion. The postoperative histological findings\nconfirmed the superficial myometrial invasion (stage IA).\nFigure 3 MRI of a 66-year-old woman with endometrial cancer.  A:  Axial\noblique T2WI, perpendicular to the main axis of the uterus, showing a\nlarge endometrial tumor that the observers judged to present superficial\nmyometrial invasion (arrow).  B:  Axial oblique DCE-MRI\nsequence (at 120 s), perpendicular to the main axis of the uterus,\nshowing a large endometrial tumor (arrow) with a hypointense signal (in\ncontrast with the hyperintense signal of the myometrium), classified by\nthe observers as superficial myometrial invasion ( C ). On\naxial DWI, the observer consensus was that the tumor (arrows) had\ninvaded the outer half of the myometrium, well depicted at a b-value of\n1000 s/mm2 ( C ) and on the ADC map ( D ). The\npostoperative histological analysis confirmed the deep myometrial\ninvasion (stage IB).\nMRI of a 66-year-old woman with endometrial cancer.  A:  Axial\noblique T2WI, perpendicular to the main axis of the uterus, showing a\nlarge endometrial tumor that the observers judged to present superficial\nmyometrial invasion (arrow).  B:  Axial oblique DCE-MRI\nsequence (at 120 s), perpendicular to the main axis of the uterus,\nshowing a large endometrial tumor (arrow) with a hypointense signal (in\ncontrast with the hyperintense signal of the myometrium), classified by\nthe observers as superficial myometrial invasion ( C ). On\naxial DWI, the observer consensus was that the tumor (arrows) had\ninvaded the outer half of the myometrium, well depicted at a b-value of\n1000 s/mm2 ( C ) and on the ADC map ( D ). The\npostoperative histological analysis confirmed the deep myometrial\ninvasion (stage IB).\nRecent advances in DWI hardware and acquisition contributed to the minimization of\nartifacts associated with arterial pulsation, peristalsis, and susceptibility\neffects ( 12 ) . Nevertheless, DWI is a motion-sensitive sequence\nand is susceptible to local field heterogeneity, high b-value images typically\nhaving a low degree of anatomical detail ( 6 ) . Therefore, DWI should be interpreted in\nconjunction with anatomical sequences for spatial reference and to avoid confusion\nwith other structures and pathologies that show high signal intensity on DWI,\nincluding bowel loops, lymph nodes, endometriosis, and hemorrhagic\ncysts ( 4 , 18 ) . In our study, DWI in conjunction with morphologic\nT2WI demonstrated greater diagnostic accuracy, sensibility, specificity, PPV, and\nNPV in the assessment of the depth of myometrial invasion than did DCE-MRI+T2WI or\nT2WI alone. These results are consistent with those of the prospective study\nconducted by Rechichi et al. ( 4 ) , and those of the retrospective study conducted by\nBeddy et al. ( 11 ) , both of whom concluded that the combination of DWI\nand T2WI is highly accurate in assessing the depth of myometrial invasion and might\nbe able to replace DCE-MRI in the preoperative evaluation of endometrial cancer. Our\nfindings are also supported by those of the study conducted by Bonatti et\nal. ( 19 ) ,\nwho drew comparisons between DWI+T2WI and contrast-enhanced T1-weighted images, also\nshowing that the former was superior. In the present study, DWI+T2WI demonstrated\nhigh sensitivity and specificity for detecting deep myometrial invasion, with an NPV\nof 96%, suggesting that a negative DWI result can reliably rule out deep myometrial\ninvasion, which is in keeping with the findings of the meta-analysis performed by\nDas et al. ( 20 ) . Nevertheless, when we compared the accuracy of\nDWI+T2WI with that of DCE-MRI+T2WI, the differences were not statistically\nsignificant, which is consistent with the results obtained in the meta-analyses\nperformed by Andreano et al. ( 2 )  and Deng et al. ( 21 ) .\nMorphologic evaluation with T2WI provided a high degree of anatomical detail to\nassess the uterus and the pelvis but was found to be limited in the assessment of\nmyometrial invasion. On T2WI, endometrial cancer in the endometrial cavity usually\nhas a signal intensity higher than that of the junctional zone, although confounding\nfactors, such as an unclear junctional zone (common in postmenopausal women) and\npoor tumor-to-myometrium contrast, as well as myometrial compression by polypoid\ntumors, leiomyomas, or adenomyosis, reduce the accuracy of the\ntechnique ( 4 , 11 , 20 ) . In our study, T2WI correctly assessed the depth of\nmyometrial invasion in 27 (64%) of the 44 patients evaluated, demonstrating lower\nspecificity than in previous studies ( 4 , 19 ) . The anatomical detail provided by T2WI is also\nimportant for the assessment of lymph node metastases ( 3 , 20 ) . Because of their high\nsignal intensity at high b-values, lymph nodes are easier to identify on DWI than on\nT2WI. However, DWI is still limited in predicting lymph node metastases and the DWI\nfindings should be considered together with the classical morphological\ncriteria ( 6 ) : short axis diameter > 8 mm for pelvic lymph\nnodes and > 10 mm for abdominal lymph nodes; irregular contours; necrosis; and\nclusters of lymph nodes.\nOur results suggest that DWI+T2WI is superior to DCE-MRI+T2WI in the assessment of\nthe depth of myometrial invasion, DWI+T2WI having a diagnostic accuracy of 95%,\ncompared with only 86% for DCE-MRI+T2WI. On DCE-MRI, endometrial tumors enhance at ?\n30 s (in the arterial phase, earlier than does normal endometrium), which allows the\ndetection of small tumors confined to the endometrial cavity, as does the fact that\nmost tumors are hypovascular compared with normal myometrium ( 22 ) . Maximum contrast\nbetween the high signal intensity of normal myometrium and the low signal intensity\nof endometrial cancer occurs 120-180 s after contrast administration, in the\nequilibrium phase ( 17 ) . However, some endometrial tumors are either\nisovascular or hypervascular in comparison with the myometrium, which hinders their\nevaluation ( 23 ) . In addition, adenomyosis, tumor extension to one\nor both of the cornua, loss of the junctional zone, and peritumoral inflammatory\nenhancement have been shown to reduce the accuracy of DCE-MRI ( 11 , 19 ) . Our study also showed\nthat T2WI and DCE-MRI+T2WI are both less accurate than is DWI+T2WI in the staging of\nendometrial cancer. The proportion of correctly staged patients increased from 77%\n(n = 34) with DCE-MRI+T2WI to 93% (n = 41) with DWI+T2WI. However, our sample\nincluded a small number of patients with advanced-stage disease (only six of the\npatients had cervical stromal invasion and none had abnormal lymph nodes), which\nlimits the comparative evaluation of the staging. On DWI, restricted diffusion\ndisrupting the cervical stroma is consistent with cervical stroma invasion, which is\nalso associated with lymph node metastases and poor survival ( 10 , 24 ) . MRI is also helpful\nin diagnosing advanced disease involving the adnexa and the peritoneum, which\nusually contraindicates laparoscopic and robotic surgery ( 25 ) . Our results are\nconsistent with those presented by Beddy et al. ( 11 ) , differing only in\nthat, in our study, more tumors were understaged than were overstaged, with\nDCE-MRI-T2WI (7 vs. 3) and with DWI-T2WI (2 vs. 1). However, because of the\nretrospective nature of our study, it was not possible to assess the real effects\nthat understaging and overstaging had on patient care. Nevertheless, the high\naccuracy of DWI+T2WI in the assessment of FIGO staging, as demonstrated in the\npresent study, could contribute to better preoperative selection of patients for\nappropriate therapy.\nWe recognize that our study has certain limitations. First, because it was a\nretrospective study, we could not assess exposures or outcomes. However, our dataset\nfeatured accurate recordkeeping and a satisfactory temporal relationship between\npreoperative MRI assessment, surgery, and histopathological reporting. Second, our\nsample size was small (n = 44), which precluded the investigation of other potential\ncauses of heterogeneity, such as the real impact of pitfalls and confounding factors\nrelated to accurate estimation of the depth of myometrial invasion. Another\nimportant limitation is the fact that the analysis of the different sets of\nsequences was not randomized-all T2WI sequences being interpreted first, followed by\nthe DCE-MRI+T2WI sets and finally by the DWI+T2WI sets-which could have introduced a\nlearning bias. In addition, the level of interobserver agreement was not assessed\nand all of the studies were evaluated by two radiologists specialized in urogenital\nradiology, which could limit the generalizability of the results. Finally, the MRI\nstudies were conducted during a period when DWI was performed only in the axial\nplane, whereas DCE-MRI was performed in two planes, including one axial oblique\nplane perpendicular to the major axis of the uterine body. Nevertheless, our results\nshow that DWI+T2WI was the superior protocol.\n\nOur study confirmed the high diagnostic performance of MRI in the preoperative\nassessment of endometrial cancer. The combination of DWI and morphological T2WI\ndemonstrated superior diagnostic accuracy in the assessment of the depth of\nmyometrial invasion when compared with that of DCE-MRI and T2WI, indicating that\nDWI+T2WI is a potential replacement for DCE-MRI in the preoperative staging of\nendometrial cancer, especially for patients in whom contrast agents are\ncontraindicated.","source_license":"CC-BY-4.0","license_restricted":false}