Sala, Evis MD, PhD, FRCR*; Crawford, Robin MD, FRCS, FRCOG†; Senior, Emma MRCS, FRCR*; Shaw, Ashley MRCP, FRCR*; Simcock, Bryony FRANZCOG†; Vrotsou, Kalliopi BSc, MSc‡; Palmer, Christopher PhD‡; Rajan, Pauline FRCR*; Joubert, Ilse BSc*; Lomas, David FRCP, FRCR*
Endometrial carcinoma is the fourth most common cancer in females and the most common malignancy of the female reproductive tract.1 In 2007, 39,080 new cases and 7400 deaths are expected in the United States1 with the incidence rising due to longevity and increasing obesity. The prognosis of endometrial carcinoma depends on a number of factors, including stage, depth of myometrial invasion, lymphovascular invasion, nodal status, andhistologic grade2 with 5-year survival rates as high as 96% for stage I disease.1
Endometrial cancer primarily presents at stage I (80% of cases), and the standard treatment is total hysterectomy and bilateral salpingo-oophorectomy. The clinical challenge is to identify correctly patients at higher stages, who will benefit from additional treatment in the form of lymph node dissection and adjuvant radiotherapy while avoiding overtreating low-risk patients. For the patients with an early stage disease, there is increasing evidence that laparoscopic treatment techniques are equally effective as open operations and have significant benefits in reduced hospital stay, less patient morbidity, and earlier return-to-normal function.3,4
Surgical staging of endometrial carcinoma is intended as initial treatment and, at the same time, identifies patients who may require adjuvant therapy. The depth of myometrial invasion is probably the single most important morphologic prognostic factor because it correlates with tumor grade, tumor extension into the cervix, and the prevalence of lymph node metastases.5,6 Incidence of lymph node metastases (pelvic and/or para-aortic) increases from 3% with superficial myometrial invasion (stage IB) to 46% with deep myometrial invasion (stage IC).6-8 Evaluation of the extent of myometrial invasion by gross inspection at surgery or at frozen-section analysis remains inaccurate in a significant proportion of patients.9,10
There is considerable practice variation worldwide regarding the role of lymphadenectomy.11,12 The abstract from the UK MRC ASTEC study has shown that surgical staging by way of pelvic lymphadenectomy does not improve survival.13 In most UK centers, lymphadenectomy is no longer considered to be a routine part of the procedure unless there are clearly involved nodes. Lymphadenectomy carries a higher risk of complications, and high-risk patients should be carefully selected for specialist surgical referral to a gynecological oncology team.14
The role of magnetic resonance imaging (MRI) in gynecological oncology has evolved durin the last 2 decades, and although it is not formally incorporated into the FIGO staging system, it is widely used by clinicians where available to assess stage of disease in both uterine and cervical cancers. In our center, it is used on all patients with a histologic diagnosis of endometrial cancer to aid treatment planning. The overall staging accuracy of MRI has been reported to be between 85% and 95%.15-24 However, there is no consensus in the literature regarding the best MRI technique including the role of dynamic intravenous (IV) contrast enhancement.16,18,20,21,23,25,26
The aims of this study were (1) to evaluate the diagnostic performance of MRI in the identification of myometrial and cervical stromal invasion using state-of-the-art imaging sequences, (2) to assess the incremental value of multiphase dynamic contrast-enhanced MRI (DCE-MRI) compared with T2-weighted (T2W) images alone, and (3) to evaluate whether the MRI result influenced the surgical procedure undertaken.
MATERIALS AND METHODS
Research ethics approval was obtained in advance for this study. The informed consent requirement was waived because of the retrospective nature of this study. Between May 2005 and April 2007, 50 consecutive patients with a histologic diagnosis of primary endometrial carcinoma underwent preoperative assessment with MRI of the pelvis in a single gynecological cancer center. Patient characteristics are presented in Table 1.
Magnetic Resonance Imaging
Magnetic resonance imaging examinations were performed on a 1.5-T whole-body machine (Excite; GEHT, Milwaukee, Wis) with an 8-channel cardiac array coil. All patients were imaged in supine position with a partially full bladder. Each examination included precontrast, high-resolution Fast Recovery Fast Spin Echo (FRFSE) sequences for optimal tumor localization and delineation using T2W sagittal, axial, and axial oblique planes, and T1W axial sequences of the pelvis and upper abdomen. Dynamic contrast-enhanced MRI after administration of 0.1 mmol of gadolinium per kilogram of body weight (pump injection) was performed by using a multiphase technique, which enabled acquisition of images at multiple phases (sagittal and axial oblique before contrast and at 25 seconds, 1 minute, and 2 minutes in the sagittal plane and at 4 minutes in the axial oblique plane after contrast). Dynamic imaging was performed by using a 3-dimensional (3D) gradient refocused echo T1W liver acquisition volume acceleration sequence. Imaging parameters were as follows: repetition time/echo time, 3.6:1.75 milliseconds; number of excitations, 0.75; matrix, 288 × 192; field of view, 36 cm; section thickness, 4 mm interpolated to 2 mm; and bandwidth, 80 kHz. There were no adverse effects to gadolinium contrast medium in our study.
Magnetic resonance images were analyzed by the gynecological oncology radiologist in our center who routinely reports most of the MRIs of the pelvis and is in charge of imaging review at the weekly cancer board meeting. The reader was blinded to the final histopathology results.
The unenhanced images (T2WI + T1WI axial) were evaluated initially, and local and overall staging was assigned for each case. An identical scoring system was used to evaluate the combination of DCE-MRI and unenhanced imaging (T2WI + DCE-MRI). On the basis of previously published criteria,25,27,28 imaging stage was categorized according to FIGO staging classification.29 An intact subendometrial enhancement (SEE) band on DCE-MRI indicated no myometrial invasion. In cases where an SEE band was not visible, a smooth tumor-to-myometrium interface indicated tumor confined to the endometrium. Superficial myometrial invasion was diagnosed when the SEE was interrupted or the interface between tumor and myometrium was irregular, with low signal intensity tumor extending to involve less than 50% of myometrial thickness. Extension of low signal intensity tumor beyond 50% of the myometrium thickness was indicative of deep myometrial invasion. The corresponding surgical histology was available in all cases and constituted the standard of reference.
The presence of potential pitfalls in the accurate MRI staging of depth of myometrial invasion20,21,25,27,30 was recorded for each case. These included leiomyoma, adenomyosis, loss of junctional zone (JZ) definition, polypoid tumor, poor tumor-to-myometrium contrast, and tumor extension into the uterine cornu.
The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall diagnostic accuracy for the T2W images alone and combined T2WI and DCE-MRI were calculated. The hypothesis of no disproportionate change between assessments (ie, correct, incorrect assessments) made using the T2WI alone and combined T2WI and DCE-MRI method was tested using the McNemar test. Associations between assessment of depth of myometrial invasion and possible pitfalls previously mentioned on T2W images were examined by implementing Fisher exact test. P < 0.05 was considered statistically significant. All analyses were performed using SPSS Version 14 and Stat-Exact 4.
Surgery and Histopathology
All patients underwent total hysterectomy and bilateral salpingo-oophorectomy, and 9 of 50 patients underwent lymph node sampling (Table 1) as MRI identified suspicious lymph nodes. Nineteen patients underwent open total abdominal hysterectomy (TAH), whereas in 31 patients, a laparoscopic approach was used. This decision was based on the shape and size of the uterus as seen on MRI and on the presence of extrauterine disease other than possible lymph node involvement. The aim of the surgery was to remove the uterus intact without morcellation or possible disruption due to a vaginal approach.
Specialized gynecological pathologists performed the histopathology staging of all tumors. The high proportion of high-grade endometrial pathology reflected the referral pattern to the cancer center. Patients at community hospitals with low-grade and presumed early stage were managed by their local gynecologists and not referred to the center, although their imaging was reviewed centrally. Depth of myometrial invasion, presence of cervical stromal invasion, and presence of metastasis within the sampled lymph nodes were confirmed microscopically. Presence of factors associated with potential inaccurate myometrial invasion, such as leiomyomas, adenomyosis, polypoid tumor, and tumor extension into the uterine cornu was also confirmed at histology.
Depth of Myometrial Invasion
The diagnostic performance of MRI in predicting depth of myometrial invasion is presented in Tables 2 and 3. Table 2 shows the percentage of correctly identified patients with myometrial invasion compared with surgicopathological findings. The depth of myometrial invasion (any depth) was correctly determined in 78% (39/50) of the cases on T2WI alone, whereas this percentage increased to 92% (46/50) with the addition of DCE-MRI. This increase in correctly identified cases was found to be statistically significant. The difference in proportions was 14% (95% confidence interval, 4.4%-23.6%, P = 0.016). Of note, the addition of DCE-MRI led to the correct detection of deep myometrial invasion in all cases. Furthermore, the addition of DCE-MRI to T2WI led to higher diagnostic accuracy, sensitivity, specificity, PPV, and NPV in the detection of depth of myometrial invasion (Table 3).
Cervical Stromal Invasion and Lymph Node Metastases
The diagnostic performance of MRI in predicting the depth of cervical stroma invasion is presented in Table 3. Only 8 of 50 patients in our study had cervical stromal invasion at histology. The addition of DCE-MRI to T2WI increased the sensitivity of detection of cervical stromal invasion by 12.5%; however, the confidence intervals were wide (34.9, 96.8, and 47.3, 99.7 for T2W and T2W + DCE T1W). The remaining diagnostic measures remained almost unchanged (Table 3).
The diagnostic performance of MRI in detecting lymph node metastasis was very high (100% sensitivity and 80% specificity for both T2W and T2W + DCE T1W). However, these calculations were based only on 9 patients who underwent lymph node sampling.
Factors Associated With Inaccurate Determination of Depth of Myometrial Invasion on T2WI
Table 4 shows the variables associated with incorrect determination of depth of myometrial invasion on T2WI alone. In this study, tumor extension into the uterine cornu (Fig. 1) was significantly associated (P = 0.014) with incorrect estimation of depth of myometrial invasion by endometrial carcinoma, although in the context of multiple testing, this should not be overinterpreted. Six of 11 patients with incorrect evaluation of depth of myometrial invasion had tumor extension into the uterine cornu, leading to an overestimation of depth of myometrial invasion in 4 patients and underestimation in 2 patients. One of these 2 patients had deep invasion on histology but was underestimated as superficial invasion on T2WI alone, whereas the second patient had superficial invasion but was categorized as no myometrial invasion on T2WI alone. Of 4 patients where T2WI alone overestimated the depth of myometrial invasion, 2 were thought to have deep invasion on T2WI but had superficial invasion on histopathology; and the other 2 patients were categorized as superficial invasion on T2WI but had no myometrial invasion at the final histologic finding.
Poor tumor-to-myometrium contrast (Fig. 2) on T2WI was seen in almost one third (27.3%) of patients who had incorrect evaluation of depth of myometrial invasion. This led to an overestimation of depth of myometrial invasion in 2 patients (superficial invasion on T2WI found to be deep invasion at histology) and underestimation in 1 patient (superficial invasion on T2WI but disease found to be confined within the endometrium at histology).
Presence of leiomyomas, adenomyosis, polypoid tumor, and loss of JZ definition was not associated with incorrect depth of myometrial invasion in this study.
Magnetic resonance imaging is now widely used for preoperative staging of patients with endometrial carcinoma, although there remains some debate concerning the best imaging protocol and the accuracy of prediction of metastases within normal-sized lymph nodes.16,18,20,21,23,25,26 This study attempts to evaluate the performance of MRI in the identification of myometrial and cervical stromal invasion using state-of-the-art imaging protocol during a relatively short period (2 years).
Accurate determination of the presence of deep myometrial invasion is a vital factor because it is directly related to the incidence of lymph node metastases, length of time to recurrence, and overall prognosis.6-8 These patients will benefit from referral to a specialist gynecological oncology team.14 Patients with no invasion or superficial myometrial invasion do not require a full staging procedure. In addition, preoperative MRI allows the clinician to select whether the patient is suitable for an open or laparoscopic procedure.3,4
In this study, the addition of multiphase 3D DCE-MRI to T2WI resulted in a significant improvement in the accuracy of assessment of the deep myometrial invasion. The diagnostic accuracy, specificity, and NPV increased from 88%, 80%, and 80% to 98%, 100%, and 93.7%, respectively. More importantly, the addition of DCE-MRI led to the correct identification of deep myometrial invasion in all cases. In the recent study by Rockall et al,23 dynamic enhancement did not improve diagnostic performance in 84 patients. However, this study reviewed the MRI examinations performed during a 10-year period, and one would expect the use of varying protocols as MRI technology improved during the decade. Chung et al22 found that administration of gadolinium did not distinguish myometrial invasion of tumor from normal myometrium in a subset of 24 patients in their study. However, a small number of patients were accrued for 9 years where changes in technique and practice may have had an effect. Our findings support previous literature where routine use of dynamic IV contrast enhancement has been reported to improve the accuracy of assessment of depth of myometrial invasion significantly (accuracy of 55%-77% for T2W images vs 85%-91% for dynamic contrast-enhanced images).16,18,20,21,25,26 In a recent prospective study, Manfredi et al16 stressed the importance of multiphase dynamic MR imaging because it allows detection of different enhancement times of the endometrial tumor compared with the adjacent normal myometrium.
Tumor extension into the uterine cornu was significantly associated with erroneous estimation of depth of myometrial invasion on T2WI alone in our study. In addition, poor tumor-to-myometrium contrast was potentially associated with an incorrect assessment of depth of myometrial invasion, although the numbers in this category were very small (6 patients of whom 3 had an incorrect assessment of depth of myometrial invasion). The addition of rapid, multiphase DCE-MRI provides the reader with superb contrast resolution between the avidly enhancing normal myometrium and relatively poor enhancement of the tumor. This is particularly important in cases where on T2WI, the signal intensity of the tumor is hardly differentiated from the normal myometrium. Similarly, in those cases where tumor extends into the uterine cornu, it can be difficult to evaluate accurately the depth of myometrial invasion owing to myometrial thinning. Our results demonstrate that, in these cases, the addition of DCE-MRI to T2WI can lead to a significant improvement in preoperative staging accuracy.
Other previously reported pitfalls in assessing the depth of myometrial invasion in endometrial carcinomas such as leiomyomas, adenomyosis, loss of JZ definition, and myometrial compression by polypoid tumor were not associated with an incorrect evaluation of depth of myometrial invasion in our study. The JZ has been suggested to be a useful landmark for determining the presence or absence of myometrial invasion. Unlike other authors,21,25,27,28 we found that JZ was present in most our patients (40/50). Paradoxically, Chung et al22 also found that the JZ was preserved in most of their patients regardless of menopausal status, but they did not find it useful in assessing deep myometrial invasion.
This study demonstrates that MRI is very accurate in predicting cervical stromal invasion. This is in keeping with the published literature where the accuracy of MRI in detecting cervical stromal invasion reaches 92% with sensitivities of 75% to 84% and specificities of 94% to 96%.16,31,32 Therefore, MRI findings can be used to guide the decision on the extent of surgery where a recent article suggested improved outcomes for stage II treated with radical hysterectomy.33
The following limitations of our study must be acknowledged. All the cases were discussed at the weekly tumor board meeting before surgery leading to the possibility of recall bias. A single reader evaluated all the MR images. Although this makes the evaluation of interobserver variability impossible, it matches our routine clinical practice where a single experienced reader specialized in gynecological oncology imaging reports most (>80%) of these studies and reviews all cases at the weekly tumor board meeting. Finally, lymph node dissection was performed in only 9 patients in our study, which is likely to have led to an overestimation of the accuracy of MRI in predicting lymph node metastases. However, during the study period, the institutional protocol did not include routine retroperitoneal lymphadenectomy and only patients where staging MRI suggested enlarged lymph nodes underwent the appropriate lymphadenectomy.
Our study differs from those previously reported because all 12MRI examinations were performed in modern MRI machines, using a state-of-the-art MRI technique that was constant during the study duration.
In conclusion, the addition of multiphase 3D DCE-MRI to T2WI can lead to a significant improvement in the accuracy of assessment of depth of myometrial and cervical stromal invasion, especially in the presence of certain pitfalls such as tumor extension into the uterine cornu and, potentially, poor tumor-to-myometrium contrast. Magnetic resonance imaging can effectively assess the depth of myometrial invasion in endometrial carcinoma and may be a useful tool to guide the surgical approach. This retrospective study will be followed up by a prospective study to confirm these findings and also evaluate the impact of MRI on treatment planning.
The authors would like to acknowledge the support of the NIHR Cambridge Biomedical Research Center, the Cambridge Experimental Cancer Medicine Center, and the Addenbrooke's Charitable trust.
1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2007. CA Cancer J Clin
2. Larson DM, Connor GP, Broste SK, et al. Prognostic significance of gross myometrial invasion with endometrial cancer. Obstet Gynecol
3. Eltabbakh GH, Shamonki MI, Moody JM, et al. Laparoscopy as the primary modality for the treatment of women with endometrial carcinoma. Cancer
4. Wong CK, Wong YH, Lo LS, et al. Laparoscopy compared with laparotomy for the surgical staging of endometrial carcinoma. J Obstet Gynaecol Res
5. Berman ML, Ballon SC, Lagasse LD, et al. Prognosis and treatment of endometrial cancer. Am J Obstet Gynecol
6. Boronow RC, Morrow CP, Creasman WT, et al. Surgical staging in endometrial cancer: clinical-pathologic findings of a prospective study. Obstet Gynecol
7. Piver MS, Lele SB, Barlow JJ, et al. Paraaortic lymph node evaluation in stage I endometrial carcinoma. Obstet Gynecol
8. Sironi S, De Cobelli F, Scarfone G, et al. Carcinoma of the cervix: value of plain and gadolinium-enhanced MR imaging in assessing degree of invasiveness. Radiology
9. Goff BA, Rice LW. Assessment of depth of myometrial invasion in endometrial adenocarcinoma. Gynecol Oncol
10. Quinlivan JA, Petersen RW, Nicklin JL. Accuracy of frozen section for the operative management of endometrial cancer. BJOG
11. Creutzberg CL. Lymphadenectomy in apparent early-stage endometrial carcinoma: do numbers count? J Clin Oncol
12. Maggino T, Romagnolo C, Landoni F, et al. An analysis of approaches to the management of endometrial cancer in North America: a CTF study. Gynecol Oncol
13. Kitchener H, Redman CW, Swart AW, et al. A study in the treatment of endometrial carcinoma: a randomised trial of lymphadenectomy in the treatment of endometrial carcinoma. Gynecol Oncol
14. Frei KA, Kinkel K, Bonel HM, et al. Prediction of deep myometrial invasion in patients with endometrial cancer: clinical utility of contrast-enhanced MR imaging-a meta-analysis and Bayesian analysis. Radiology
15. Kim SH, Kim HD, Song YS, et al. Detection of deep myometrial invasion in endometrial carcinoma: comparison of transvaginal ultrasound, CT, and MRI. J Comput Assist Tomogr
16. Manfredi R, Mirk P, Maresca G, et al. Local-regional staging of endometrial carcinoma: role of MR imaging in surgical planning. Radiology
17. Hricak H, Rubinstein LV, Gherman GM, et al. MR imaging evaluation of endometrial carcinoma: results of an NCI cooperative study. Radiology
18. Ito K, Matsumoto T, Nakada T, et al. Assessing myometrial invasion by endometrial carcinoma with dynamic MRI. J Comput Assist Tomogr
19. Joja I, Asakawa M, Asakawa T, et al. Endometrial carcinoma: dynamic MRI with turbo-FLASH technique. J Comput Assist Tomogr
20. Seki H, Kimura M, Sakai K. Myometrial invasion of endometrial carcinoma: assessment with dynamic MR and contrast-enhanced T1-weighted images. Clin Radiol
21. Yamashita Y, Harada M, Sawada T, et al. Normal uterus and FIGO stage I endometrial carcinoma: dynamic gadolinium-enhanced MR imaging. Radiology
22. Chung HH, Kang SB, Cho JY, et al. Accuracy of MR imaging for the prediction of myometrial invasion of endometrial carcinoma. Gynecol Oncol
23. Rockall AG, Meroni R, Sohaib SA, et al. Evaluation of endometrial carcinoma on magnetic resonance imaging. Int J Gynecol Cancer
24. Vasconcelos C, Felix A, Cunha TM. Preoperative assessment of deep myometrial and cervical invasion in endometrial carcinoma: comparison of magnetic resonance imaging and histopathologic evaluation. J Obstet Gynaecol
25. Sironi S, Colombo E, Villa G, et al. Myometrial invasion by endometrial carcinoma: assessment with plain and gadolinium-enhanced MR imaging. Radiology
26. Nakao Y, Yokoyama M, Hara K, et al. MR imaging in endometrial carcinoma as a diagnostic tool for the absence of myometrial invasion. Gynecol Oncol
27. Hricak H, Stern JL, Fisher MR, et al. Endometrial carcinoma staging by MR imaging. Radiology
28. Lien HH, Blomlie V, Trope C, et al. Cancer of the endometrium: value of MR imaging in determining depth of invasion into the myometrium. AJR Am J Roentgenol
29. Shepherd JH. Revised FIGO staging for gynaecological cancer. Br J Obstet Gynaecol
30. Ascher SM, Reinhold C. Imaging of cancer of the endometrium. Radiol Clin North Am
31. Seki H, Takano T, Sakai K. Value of dynamic MR imaging in assessing endometrial carcinoma involvement of the cervix. AJR Am J Roentgenol
32. Nagar H, Dobbs S, McClelland HR, et al. The diagnostic accuracy of magnetic resonance imaging in detecting cervical involvement in endometrial cancer. Gynecol Oncol
33. Cohn DE, Woeste EM, Cacchio S, et al. Clinical and pathologic correlates in surgical stage II endometrial carcinoma. Obstet Gynecol