| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Qilu Hospital of Shandong University, 107 Wenhuaxi
Rd., Jinan, Shandong, China 250012.
2 Department of Radiology, Weill Medical College of Cornell University, New
York, NY.
3 Department of Surgery, Qilu Hospital of Shandong University, Jinan, Shandong,
China.
Received May 3, 2007;
accepted after revision November 12, 2007.
Address correspondence to C. Li
(chuanfu.li{at}gmail.com).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Thirty-eight patients (23 men, 15 women) with clinically suspected rectal carcinoma underwent 3-T MRI. Coronal, axial, and sagittal T2-weighted sequences with and without fat suppression; axial T1-weighted spin-echo sequences; axial T1-weighted gradient-echo sequences with and without fat suppression; oblique 2D MR hydrography; and 3D fat-suppressed dynamic contrast-enhanced MRI were performed. Image quality with these sequences was evaluated by three radiologists experienced in body MRI. The significance of difference in results with the sequences was tested. The manner in which MRI staging and feasibility of sphincter-sparing surgery agreed with operative and pathologic findings was evaluated with kappa statistics.
RESULTS. Rectal carcinoma was identified on MRI and confirmed histologically in all 38 patients. MRI findings were correctly predictive of T category in 35 cases (accuracy, 92.1%). In 31 (96.9%) of 32 resectable cases,sphincter-sparing surgical approaches were accurately chosen on the basis of MRI findings. Among the 11 sequences, 3D fat-suppressed dynamic contrast-enhanced MRI best delineated tumor margins. Coronal and axial T2-weighted images also well depicted tumor margins with minimal artifact. T2-weighted images were superior to unenhanced T1-weighted images.
CONCLUSION. MRI of rectal cancer at 3 T is accurate for prediction of T category and the feasibility of sphincter-sparing surgery. The best images were obtained with coronal, sagittal, and axial T2-weighted sequences and 3D fat-suppressed dynamic contrast-enhanced MRI.
Keywords: dynamic contrast enhancement • MRI • rectal carcinoma
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Most studies of rectal MRI have been performed with a field strength of 1.5 T or lower [8–11] because susceptibility artifacts from bowel gas increase at higher field strength. This artifact, however, can be reduced with the use of spin-echo sequences and distention of the rectum with warm water. In a preliminary study, Chun et al. [12] found that 3-T high-field-strength MRI with only four MR sequences and without gadolinium enhancement or MR hydrography was almost as accurate as endorectal sonography for staging rectal carcinoma. Another study [13] of the use of 3-T MRI with a four-channel phased-array coil in the diagnosis of rectal cancer also showed promising results with T2-weighted fast spin-echo sequences in the axial and sagittal planes and 2D T1-weighted sequences with fat saturation before and after gadolinium enhancement. There remains uncertainty, however, about the optimal pulse sequences, and there are limited data on the diagnostic accuracy of rectal MRI at 3 T. The lack of a standard protocol causes inconsistent diagnostic accuracy among institutions. The purpose of this study was to evaluate the image quality of various 3-T MRI sequences for preoperative staging and planning of sphincter-sparing resection of rectal cancer.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Before MRI, bowel preparation included laxative cleansing (oral administration of magnesium sulfate at a dose of 20 g/100 mL followed by oral hydration the night before rectal MRI for inpatients or 20-mL glycerin enema 1 hour before rectal MRI for outpatients), luminal distention with warm sodium chloride enema, and 10 mg of anisodamine by intramuscular injection 10 minutes before the examination to reduce bowel peristalsis. Before anisodamine was administered, patients were interviewed to exclude contraindications to use of this agent, such as glaucoma, prostate gland hyperplasia, and cardiac disease.
MRI Technique
All patients were examined freely breathing in the supine position on a 3-T
MRI unit (Signa Excite, GE Healthcare). The body coil was used for signal
transmission, and an eight-channel phased-array surface coil was used for
signal reception. A landmark was made on the pubic symphysis.
An initial three-plane localizer view covering the entire pelvis was obtained. Subsequent sequences included coronal, axial, and sagittal T2-weighted images; axial T2-weighted images with fat suppression; axial T1-weighted gradient-echo images; axial T1-weighted gradient-echo images with fat suppression; radial oblique 2D MR hydrographic single-shot fast spin-echo images; and axial 3D fat-suppressed dynamic gadolinium-enhanced MR images (Fig. 1A, 1B, 1C, 1D, 1E, 1F, 1G, 1H, 1I, 1J). An axial T1-weighted spin-echo sequence was added for the last 14 patients. Coronal and sagittal T2-weighted sequences with fat suppression were added if necessary as determined by the supervising radiologists. The imaging parameters for each sequence are listed in Table 1. Two-dimensional MR hydrography was performed with at least five slices prescribed radially centered on the tumor at the position of any luminal stenosis.
|
|
|
|
|
|
|
|
|
|
|
Three-dimensional dyn a mic gadolinium-enhanced MRI was performed with an axial multiphase spoiled gradient-echo se quence with fat suppression. The temporal resolu tion was 13–16 seconds per phase. Dynamic contrast-enhanced MRI acquisition was started simultaneously with IV injection of a standard dose (0.1 mmol/kg) of gadopentetate dimeglumine (Magnevist, Bayer HealthCare) and repeated continuously for 15–20 phases with a total data acquisition time of approximately 4 minutes. The gadolinium contrast agent was injected at a rate of 3 mL/s with a power injector (Spectris MR, Medrad) and followed by a bolus of 20 mL of normal saline solution at the same injection rate.
Image Analysis
MR images were evaluated independently by three radiologists with more than
8 years of experience in interpreting body MR images. The reviewers were
blinded to all clinical information. Images obtained with all sequences were
presented to each reviewer simultaneously without control of the order in
which the reviewers focused on the MR sequences. In this way the reviewers
could directly compare images obtained with the various sequences side by side
to assess relative image quality.
Overall image quality was graded on a four-point scale as follows: 0, nondiagnostic; 1, poor image quality but still diagnostic; 2, good image quality; 3, excellent image quality. Other factors evaluated were artifacts (0, artifact interfering with diagnosis; 1, moderate artifact not interfering with diagnosis; 2, mild artifact; 3, no artifact) and depiction of the border between rectal tumor and normal rectum (0, no distinct border visualized; 1, moderate blurring of rectal wall, rectal tumor border not clearly identified; 2, mild blurring of rectal wall, rectal tumor border identified; 3, well-demarcated tumor border).
Each rectal tumor was staged according to MRI features and later correlated with the operative and pathologic findings. T1 and T2 categories were not differentiated in this study. T1 and T2 lesions were differentiated from T3 lesions by identification of a smooth outer tumor border within the rectal wall with no invasion into fat surrounding the rectum. T3 lesions had irregular outer borders and invasion into fat surrounding the rectum with plaque, mass, or cordlike signal intensity projecting into perirectal fat. In T4 lesions, fat planes between rectal carcinoma and surrounding organs disappeared. Lymph nodes 1 cm or larger within the field of view were diagnosed on MRI as lymph node metastasis. Metastasis to other organs within the field of view also was identified. Metastatic disease outside the field of view of pelvic MRI was not assessed.
To predict the feasibility of sphincter-sparing surgery, the distance from the lower margin of a rectal tumor to the point at which the levator ani muscle attached to the rectum was measured. The surgeons preferred this landmark to the dentate line because it was considered easier to identify on MRI, particularly on coronal T2-weighted images, and because it is a more reliable marker of the superior sphincter margin. If the distance was 2 cm or greater, sphincter- and anus-sparing surgery was considered feasible. Surgical plans determined with MRI findings were correlated with the operative findings.
Statistical Method
The significance of differences among sequences was tested with ridit
analysis (a statistical method used to describe differences between groups on
an ordered categoric basis) and ordinal logistic regression for image quality,
artifacts, and depiction of rectal tumor border with normal rectal wall. The
kappa statistic was used to evaluate the manner in which MRI staging and
judgment about feasibility of sphincter-sparing surgery for rectal cancer
agreed with the operative and pathologic findings. A value of p <
0.05 in a two-tailed test was considered statistically significant. All
analyses were performed with SAS software (version 9.1, SAS Institute).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Rectal carcinoma was identified on MRI in all 38 patients and was confirmed at histologic examination of surgical (n = 32) and biopsy (n = 6) specimens. The mean tumor size was 5.2 cm (range, 2.5–10 cm). Histologic classification of the tumors showed seven well-differentiated adenocarcinomas (Fig. 2), 14 moderately well-differentiated adenocarcinomas (Fig. 3), 13 poorly differentiated adenocarcinomas, and four signetring cell carcinomas. Histopathologic staging revealed two lesions in category T1, 11 in category T2, 21 in category T3 (Fig. 4), and four in category T4 (Fig. 5). Lymph node metastasis was confirmed in 17 patients. Two patients were found to have distal metastasis (liver, lung), which was identified with separate abdominal MRI and chest CT examinations performed on a later date. Twenty-five patients underwent sphincter-sparing resection of the rectum, and seven underwent abdominoperineal excision. In four patients, the tumor was not completely resected owing to local tumor extent. Instead, diverting colostomy was performed to relieve intestinal obstruction. The two patients with distal metastasis were treated with chemotherapy without undergoing surgery. One of these two patients also underwent radiation therapy.
|
|
|
|
MR Image Quality
The mean ridits of image quality scores for each 3-T MRI sequence are shown
in Table 2 with a statistically
significant difference (p < 0.0001). According to the mean ridit
of each sequence, among the 11 sequences, 3D fat-suppressed dynamic
gadolinium-enhanced MRI was considered to provide the best delineation of
tumor margins. Coronal T2-weighted and axial T2-weighted images also depicted
tumor margins well and had the highest image quality score with the least
artifact. In general, T2-weighted images were superior to T1-weighted images
(p < 0.0001) in ordinal logistic regression. For T1-weighted
images, the fast spin-echo sequence was better than the spoiled
gradient-recalled echo sequence (p < 0.05) in ordinal logistic
regression, although the acquisition time was longer. The lesions were
depicted more clearly on sequences without fat suppression than on those with
fat suppression because the high signal intensity of fatty tissue surrounding
the rectum gave better contrast to the tumor with lower signal intensity.
Radially oblique MR hydrography depicted the rectal lumen similarly to barium
enema, showing endoluminal features such as filling defects and ulcerations,
but did not show mural features.
|
Diagnosis, Staging, and Planning of Sphincter-Saving Surgery
With MRI, T category was correctly estimated for 35 of the 38 patients
(accuracy, 92.1%; 
= 0.86, p < 0.0001)
(Table 3). In the evaluation of
tumor invasion into fat tissue surrounding the rectum to differentiate T3 or
higher from T2 or lower, surgical findings confirmed 25 cases with invasion
and 13 without invasion; 35 of 38 cases were depicted accurately with MRI
(accuracy, 92.1%; = 0.83; p < 0.0001). MRI showed four of
four rectal carcinomas with invasion to adjacent organs, which differentiated
category T4 from category T3 with an accuracy of 100%. With MRI, 11 of 17
patients with lymph node metastasis and 19 of 21 patients without nodal
metastasis were correctly identified (accuracy, 79.0%; = 0.56;
p < 0.001) (Table
4).
|
|
For the 32 resectable tumors, with a 2 cm or greater distance between the
lower margin of rectal cancer to the point at which the levator ani muscle
attached to the rectum as the criterion for predicting the feasibility of
sphincter-sparing surgery, MRI was accurate for determining the surgical
approach in 31 cases (accuracy, 96.9%; = 0.9; p < 0.0001).
The only patient with less than 2 cm between the tumor and the levator ani
muscle had T2 well-differentiated rectal adenocarcinoma. This patient
underwent internal sphincter resection with prolapsing technique to save at
least the external sphincter and anus, although the distance was only 1.5 cm
on MRI.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Various MRI sequences have been used to image the rectum. The image-quality data in this study confirmed that for 3-T units, images obtained with T2-weighted spin-echo sequences are superior to those obtained with T1-weighted sequences, as has been shown for 1.5-T MRI [23]. On T2-weighted images, tumors were slightly better evaluated without fat suppression because on T2-weighted images perirectal fat with its high signal intensity has excellent contrast to the tumor. In some cases, however, high-signal-intensity perirectal inflammation was better visualized on T2-weighted images with fat suppression. For the two T1-weighted sequences studied, image quality was better with axial T1-weighted fast spin-echo sequences, albeit with a longer acquisition time than for the axial T1-weighted gradient-echo sequence, but was still substantially inferior to that of the T2-weighted images. Coronal and sagittal images were especially useful for showing the relations among the tumor, levator ani muscle, and sphincter.
All three reviewers identified dynamic contrast-enhanced MRI as the best sequence for depicting tumor margins even through respiratory motion artifact was present. All reviewers considered this sequence essential in the rectal MRI protocol. This finding was consistent with those of a study by Wallengren et al. [24]. In the detection of rectal cancer, those investigators reported 100% sensitivity and 70% specificity of MRI performed with contrast enhancement by superparamagnetic ferristene enema. This double-contrast method also was clinically valuable for staging and determining the depth of tumor invasion into the rectal wall, which was not possible in this study because of our goal of assessing MR hydrography performed with saline enema.
MR hydrography has been used successfully to visualize the lumens of biliary ducts, pancreatic ducts, the ureters, and the bladder through depiction of static fluid. In this study, sodium chloride was introduced into the rectum to show the rectal lumen distended with water in a manner similar to that used for 2D single-shot fast spin-echo MR hydrography performed radially to view the endoluminal features from different angles. Depiction of the rectal lumen was similar to that on images from barium enemas.
Bowel preparation is crucial to avoid interference from feces and to reduce susceptibility artifact from air–tissue borders. In this study, a simplified 20-mL glycerin enema 1 hour before the examination was adequate for patients with rectal cancer in the middle and lower sections who had not yet undergone magnesium sulfate bowel cleansing for surgery. Distention of the rectum is useful but often is not possible owing to encasement by tumor; at least one group of authors [25] has suggested that bowel preparation and distention are unnecessary. Lauenstein et al. [26] suggested oral administration of multiple doses of gadolinium or barium sulfate beginning 3 days before MRI examination to label feces, entirely avoiding enemas. Many contrast materials have been used to distend the rectum, including air [27], water [28], dilute gadolinium, and other paramagnetic agents [24, 29]. In our study, warm sodium chloride was used mainly because of its low cost, absence of toxicity, and acceptance by patients.
In addition to facilitating the diagnosis and staging of rectal tumors, MRI contributes to surgical planning by showing the relations among the tumor, the sphincter, and the levator ani muscle. Sphincter invasion is identified with an accuracy of 87% [30, 31]. Complete tumor resection and sphincter sparing are important goals of rectal surgery to improve quality of life and have fewer complications than abdominoperineal excision. A rectal cancer distal resection margin greater than 2 cm is considered optimal for avoiding recurrence [32]. Thus the length of normal rectum above the levator ani muscle is the key to determining whether sphincter-sparing surgery can be performed. In this study, the distance from the lower margin of rectal cancer to the upper margin of the external sphincter (the point at which the levator ani muscle attaches to the rectum) was measured on good-quality coronal and sagittal images to assess the feasibility of sphincter-sparing surgery with adequate tumor margins. Data on 32 patients with surgical confirmation showed that findings on 3-T MRI were accurate predictors of the feasibility of sphincter-sparing surgery in 31 (96.9%) of the patients. In one patient with a 1.5-cm distance between the tumor and the levator ani muscle, MRI helped the surgeons to plan a modified procedure that spared the external sphincter and anus.
The data on the diagnostic performance of 3-T MRI compare favorably with data on sonography, which is reported to have 69–97% accuracy, and on CT, which has a reported accuracy of 52–87% for TNM T categorization of rectal cancer [4–6, 33, 34]. Endoluminal sonography is reliable in assessing depth of penetration by tumor into the rectal wall, but this technique is invasive and operator dependent. Endoluminal sonography cannot be used to assess tumor close to the sigmoid colon or tumor extension into adjacent organs. CT cannot be used to assess depth of invasion into the rectal wall, but it has the benefits of being fast, having a large field of view, and enabling simultaneous evaluation for distant metastasis.
As in other cancers, detecting lymph node metastasis is the most challenging aspect of MRI diagnosis of rectal cancer. Kim et al. [34] found that the accuracy rates of MRI, CT, and endoluminal sonography for local lymph node metastasis of rectal cancer were 63%, 56.5%, and 63.5%, respectively. The accuracy of imaging is low mainly because the diagnosis of metastasis is made only on the basis of the size and shape of lymph nodes, and thus micrometastasis is missed. Because lymph nodes enlarge in both inflammatory and neoplastic processes, which are difficult to differentiate morphologically, false-positive and false-negative results occur. In numerous studies, lymph nodes larger than 1 cm have been considered metastasis [35]; in other studies, cutoffs of 8 mm [36] and 6 mm [37] have been used. With greater than 6 mm as the criterion for the diagnosis of lymph node metastasis around rectal wall and surrounding fat, the sensitivity, specificity, and accuracy were only 57%, 88%, and 76%, respectively [37]. Brown et al. [38] analyzed 437 lymph nodes and concluded that benign and malignant lymph nodes were similar in size. Those authors believed that accuracy could be increased by evaluating lymph node borders and signal intensity. They proposed that using irregular border and mixed signal intensity as the criteria for metastatic lymph nodes would improve sensitivity to 85–95% and specificity to 95–97%. Heterogeneous contrast enhancement is another indicator of lymph node metastasis [39]. New contrast agents have been developed for detection of metastatic lymph nodes [40, 41], but they were not used in this study.
There were additional limitations to this study. With the small number of patients, there were only two T1 lesions, and insufficient number for assessment of the utility of 3-T MRI in differentiating T1 and T2 disease. Therefore, T1 and T2 lesions were combined as T1 and T2 for statistical analyses. Imaging with an endorectal coil to depict the layers of the rectal wall and differentiate T1 from T2 lesions has a reported accuracy of 92% [42, 43]. However, the technique has a limited field of view and requires expertise for coil insertion to avoid rectal injury. The failure rate can be as high as 40% [44]. Endorectal coils can produce artifacts and are of limited use in imaging of tumors close to the sigmoid colon or of obstructing lesions. Phased-array surface coils are more acceptable because of their simplicity and larger, more homogeneous field of view [15, 33], but tumor invasion into the submucosa and muscularis propria of the rectal wall has not been well evaluated.
Despite the larger field of view of the eight-channel phased-array coil compared with endorectal coils, this study was limited to the pelvis. Separate examinations of the abdomen and chest were performed to assess the presence of hepatic and pulmonary metastasis. Newer body-array coils and use of total image matrix technique for greater anatomic coverage may make it possible to cover the abdomen and pelvis in a single examination.
To save time, three of the sequences—axial T1-weighted spin echo and coronal and sagittal T2-weighted with fat suppression— were not performed for all of the patients. This omission caused bias in comparison of image-quality scores. The added value of 3 T compared with 1.5 T was not directly assessed because the patients underwent only 3-T imaging. It is our impression, however, that the signal-to-noise ratio at 3 T is noticeably greater than that at 1.5 T.
We conclude that MRI of rectal cancer at 3 T is accurate for prediction of tumor stage and the feasibility of sphincter-sparing surgery. In clinical practice, a basic imaging protocol can be simplified to four sequences: coronal, sagittal, and axial T2-weighted sequences and 3D fat-suppressed dynamic gadolinium-enhanced MRI. MR hydrographic images and axial T1-weighted fast spin-echo and axial T2-weighted images with fat suppression add only slightly more information.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Kim, J. S. Lim, J. Y. Choi, J. Park, Y. E. Chung, M.-J. Kim, E. Choi, N. K. Kim, and K. W. Kim Rectal Cancer: Comparison of Accuracy of Local-Regional Staging with Two- and Three-dimensional Preoperative 3-T MR Imaging Radiology, February 1, 2010; 254(2): 485 - 492. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
