An audit of MRI's accuracy in staging uterine cancer shows the modality is good at assessing cervical extension and lymph node spread, but is not so useful for ascertaining myometrial invasion, a group from the U.K. Royal College of Radiologists (RCR) reports.
Clinical staging of uterine cancer is unreliable, so the International Federation of Gynecology and Obstetrics (FIGO) recommended surgical staging in 1988. The most important prognostic factor for endometrial cancer is the degree of myometrial invasion. But in the absence of deep myometrial invasion, the reported lymph node metastatic rate is only 3% of cases.
Conversely, when there is deep myometrial invasion, lymph node metastatic rates have been reported in as many as 46% of cases, according to Dr. Karen Duncan from the Clinical Radiology Audit Committee, RCR, London (Clinical Radiology, June 2012, Vol. 67:6, pp. 523-530).
If the extent of the disease is more accurately predicted, this may affect the type of surgery planned. The FIGO guidelines from the 1980s include lymphadenectomy as a component of the surgical staging system, but in 2000 the organization suggested it only be performed in high-risk cases.
However, many U.K. centers no longer perform lymphadenectomy due to the associated morbidity and uncertainty as to its benefit. As a result, they've restricted the procedure to patients with clearly involved nodes. Due in part to this differing surgical practice, indications for MRI of the uterus are not firmly established. Imaging's place remains controversial and practice varies.
"The aim of the present national audit, undertaken by the RCR Clinical Radiology Audit Committee (CRAC), was to assess the use and accuracy of MRI in women with uterine cancer as currently performed in the U.K. for benchmarking purposes and, by distribution of results, hopefully promotion of good practice," the authors wrote. "No attempt was made to assess its efficacy in terms of disease management, but instead elected to determine whether the depth of myometrial invasion was being assessed accurately with MRI, as this directly affects patient management."
There is no nationally agreed upon guideline for which grade or tumor stage should be assessed with MRI, no agreed upon sequences to be used, nor are there defined expected accuracy rates. As such, the researchers' standards were based on a review of the current literature in 2007.
"It was hoped that this audit would determine whether all centers could achieve the standards suggested by the [American College of Radiology (ACR)], namely that MRI with contrast enhancement can be between 85% and 91% accurate in assessing the depth of myometrial invasion, with accuracy of cervical extension ranging from 86 to 95%," the authors wrote. "As the ACR did not discuss the accuracy of lymphadenopathy assessment, work by Rockall et al and Manfredi et al was used to set the lymphadenopathy standard; they reported accuracy rates between 72% and 90%."
All U.K. radiology departments were invited to participate using a Web-based tool for submitting anonymized data for a 12-month period. MRI staging was compared with histopathological staging using target accuracies of 85%, 86%, and 70%. Only 37 out of 87 departments participated.
Targets for MRI staging were achieved for two of the three standards nationally, with diagnostic accuracy for depth of myometrial invasion, 82%; for cervical extension, 90%; and for pelvic nodal involvement, 94%; the latter two being well above the targets. However, only 13/37 (35%) of individual centers met the target for assessing depth of myometrial invasion, 31/36 (86%) for cervical extension, and 31/34 (91%) for pelvic nodal involvement.
Statistical analysis demonstrated no significant difference for the use of intravenous contrast medium, but did show some evidence of increasing accuracy in assessment of depth of myometrial invasion with increasing caseload, the researchers wrote.
"The margins are small, but for departments concerned about their accuracy in assessing the depth of myometrial invasion, it may be appropriate for them to consider increasing their exposure to this examination type by double reporting within the center or with another center," the authors said.
Smaller centers reported that their patients were often transferred to a larger center for further management and the MRI studies were re-reported there. When looking for other reasons besides lack of exposure to myometrial invasion for varying performance, one of the main variations in practice between departments was the use of intravenous contrast medium.
Opinion in the literature is mixed regarding intravenous contrast medium; there is no agreed ideal set of imaging sequences for evaluating endometrial cancer, and as a result, departments have devised their own protocols depending on experience, equipment, and personal preferences, the authors added.
"From the data collected for this audit, it would appear that 57% of departments always use intravenous contrast medium, and 32% of departments never use intravenous contrast medium in endometrial cancer staging examinations," Duncan wrote. "When collecting data for this audit standard post contrast and dynamic contrast imaging were not differentiated. However, analysis of the collected results showed only a borderline improvement in sensitivity with intravenous contrast medium usage in pelvic nodal assessment and no benefit in the assessment of cervical stromal or deep myometrial invasion."
Therefore, besides small caseloads, no other clear factors to improve performance were identified; however consideration of nationally rationalizing MRI protocols and reporting templates, e.g. those to be proposed by the RCR/National Cancer Intelligence Network project may be appropriate, the authors added.
"The responses to the departmental questionnaire do highlight the wide variation in practice with regard to the routine or selective use of MRI in staging of endometrial cancer," Duncan wrote. "This may reflect its greater usefulness in certain cases, such as high-grade disease, and in a resource-limited healthcare system, this too is worth consideration when new guidelines are being developed."
For future audits the minimum achievable standards of MRI accuracy could be set at 80% for predicting myometrial involvement; 90% for determining cervical extension; and 85% for predicting lymph node involvement, they added.