The successful introduction of magnetic resonance imaging (MRI) for pelvic diseases has, in recent years, led to the gradual replacement of CT by MRI for local and regional rectal cancer staging. Initial MR studies were performed with a body coil. Because conventional body coil techniques showed a resolution that was still insufficient to differentiate the individual layers of the rectal wall, overall accuracies reported for MRI with a body coil have not been any better than those reported for CT, with values ranging from 59 to 88% [15, 30-35].
The introduction of endoluminal coils facilitated improved image resolution and made detailed evaluation of the layers of the rectal wall feasible. This was also reflected in improved and more consistent T staging, with accuracy ranging between 71 and 91% [36-43]. Endorectal MRI can be as accurate as endorectal US for staging of superficial tumours, as shown in studies comparing the two endoluminal techniques [36, 37]. However, some problems remain with endorectal MRI. Besides the limited availability and high cost, MRI with an endoluminal coil, especially when used in isolation, has a limited field of
Fig. 9. Magnetic resonance imaging. Sagittal fat sat T2-weighted turbo spin echo image shows a hyperintense lesion in the middle rectum view. Like endorectal US, the mesorectal fascia and surrounding pelvic structures are difficult to visualise owing to the sudden signal drop-off at a short distance from the coil . Furthermore, the positioning of an endoluminal device can be difficult or impossible in patients with high and/or stenosing tumours, and failed insertion rates of as high as 40% have been reported in patients with rectal cancer .
With the introduction of dedicated external coils, especially phased-array coils, improvement in MRI performance was expected [46-50]. The advantages of high spatial resolution with a large field of view make phased-array MRI suitable for staging of both superficial and advanced rectal tumours (Figs. 8 and 9). However, Authors of the first studies that used MR with the multiple surface coil technique reported an overall accuracy for T staging of only 55-65% and showed no benefit compared with the use of a body coil or even with CT [51,52]. The low performance of MRI in these studies could have been attributed to the low spatial resolution that was used with the early phased-array techniques. But even when a higher spatial resolution was applied with the new generation of phased-array coils, the accuracy for T staging was not as high as anticipated, with values varying between 65% and 86% [50,51], and was not as reproducible as expected, with considerable interobserver variability . One exception to the above was the study by Brown et al. , who reported 100% accuracy and complete agreement between two readers on the prediction of tumour stage with phased-array MRI results.
Most staging failures with MRI occur in the differentiation of T2-stage and borderline T3-stage lesions, with overstaging as the main cause of errors. Over-staging is often caused by desmoplastic reactions [43, 49,53], and it is difficult to distinguish on MR images between spiculation in the perirectal fat caused by fibrosis alone (stage pT2) and spiculation caused by fibrosis that contains tumour cells (stage pT3) .
The present T-staging system is sometimes used for clinical decision making. Post-operative combined chemotherapy and radiation therapy has been the standard in the United States for patients with T3- and/or Nl-stage tumours. There is now a growing tendency to give the adjuvant therapy pre-opera-tively and, therefore, a need for a good imaging method to select patients at high risk. In this respect, the present T-staging system does have its shortcomings: it does not discriminate between tumours with a wide circumferential resection margin (CRM) and tumours with a close or involved CRM. Although most of these tumours are classified as stage T3, they have a different risk for local recurrence. It has been repeatedly shown that the distance from the tumour to the circumferential mesorectal resection plane is a more powerful predictor for the local recurrence rate than is the T stage [54, 55]. It is therefore probably more important to use imaging to identify those tumours that will have a close or involved resection margin so that they can be selected for more extensive (neoadjuvant) treatment.
Rectal cancer has two main routes of lymphatic spread. For the upper portion of the rectum, the route is upward along the superior rectal vessels to the inferior mesenteric vessels. The lower portion of the rectum shows an additional lateral lymphatic route along the middle rectal vessels to the internal iliac vessels. Downward spread along the inferior rectal vessels to the groin is unusual except in very advanced cases and when the anal canal is involved.
Results of early anatomic studies [56-60] showed that over half of the metastatic nodes were within 3 cm of the primary tumour and were smaller than 5 mm in size. With standard total mesorectal excision (TME), the perirectal nodes are removed with the primary tumour but the internal iliac nodes are left in situ. In lower rectal cancer, therefore, there is a risk that involved internal iliac nodes will be left behind, with the chance for local recurrence. The magnitude of this risk was illustrated by Moriya et al. , who showed that as many as 28% of lymph node-positive distal rectal cancers have involvement of lateral nodes, and in 6% of cases those lateral nodes were the only lymph nodes involved. This means that disease in 6% of patients is incorrectly staged as node-negative at TME. The fact that nodal disease is a prognostic indicator not only for distant metastases but also for local recurrence has been confirmed in the large Dutch TME trial , where patients with stage III (TxN1) disease had a 10-fold higher risk for local recurrence than did those with stage I (T1-2N0 stage) disease and a threefold higher risk than did those with stage II (T3N0 stage) disease.
When the treatment strategy is post-operative chemotherapy and radiation therapy for patients with T3N1 disease, there is little need to identify the lymph node status pre-operatively. When the emphasis is on pre-operative radiation therapy, with or without chemotherapy, and one wants to select patients at high risk, determination of lymph node status becomes essential.
Some surgeons, mainly from Japan, claim improved local control by adding extended pelvic lym-phadenectomy to resection of the rectum. This approach is not favoured by most surgeons because of the additional urologic and sexual morbidity, while the benefit is unclear. Again, selection of those patients with the highest risk for lateral lymph node metastases could be useful for centres where pelvic lymphadenectomy is practised.
Identification of nodal disease is still a diagnostic problem for the radiologist. Despite the identification of lymph nodes as small as 2-3 mm on high-spatial-resolution images, reliable detection of nodal metastases is presently not possible. The radiologic assess ment of nodal involvement generally relies on morphologic criteria such as the size and shape of the node [63-65]. The problem with morphologic imaging, however, is that with enlarged nodes it is difficult to distinguish between reactive and metastatic nodes, and with small nodes micrometastases are easily missed. An additional problem in rectal cancer, as compared with other pelvic tumours, is the high frequency of micrometastases in normal-sized nodes. Large variations in accuracy (62-83%) for nodal detection can be found for endorectal US [67, 68], as well as for CT (22-73%) [29, 68, 69]. Despite the superior soft-tissue contrast, it has not been possible with unenhanced MRI to accurately distinguish between inflammatory and metastatic nodes on the basis of signal intensity criteria, nor has the use of non-specific MR contrast agents improved detection accuracy. Accuracy rates for nodal detection with unenhanced MR imaging vary between 39 and 95% [70-72].
An alternative method would be metabolic imaging by fluorodeoxyglucose positron emission tomography (PET). Fluorodeoxyglucose PET scanning has been shown to be useful in the management of recurrent rectal cancer [73-78]. For primary rectal cancer, there may be some benefit in terms of the detection of distant metastases, but to our knowledge there has been only one study  that focused on nodal staging, and a disappointingly low sensitivity of 29% was reported in that study. The reason for the low sensitivity may well be that the proximity of the primary tumour to the urinary bladder obscures small nodal metastases.
Recently, MR imaging with the use of ultrasmall superparamagnetic iron oxide (USPIO) contrast agents has shown promising results for staging nodal metastases. USPIO is a contrast agent that undergoes phagocytosis by the reticuloendothelial system (macrophages in normal lymph nodes). The use of USPIO results in shortening of the T2 relaxation time and in a decrease in signal intensity on gradient-echo images of normal lymph nodes owing to increased susceptibility artefacts. These MR properties are used to aid in the detection of micrometastases in small lymph nodes. In metastatic nodes, the reticuloen-dothelial system is displaced by tumour deposits and shows deficits in the uptake of USPIO. In patients with head and neck cancer and urologic pelvic tumours, sensitivities for detection have been reported to be good [80, 81]. At present, the value of MR imaging with USPIO in the detection of nodal metastases in rectal cancer patients is not clear and warrants further evaluation.
Was this article helpful?