MR beats CT, ultrasound in staging advanced rectal cancer

MR is superior to CT and ultrasound for predicting local tumor invasion in patients with advanced rectal carcinoma, according to German researchers. They suggest that MR may replace CT as the primary staging modality.

Preoperative staging of advanced rectal tumors is important because most tumors that don't infiltrate the tunica muscularis can be operated on immediately, said lead researcher Dr. Jens C. Stollfuss of Munich, Germany.

"The presence or absence of muscularis involvement and the presence of positive nodes are the major determinants of local recurrence, and thus of the long-term survival of the patient," he said.

Stollfuss and colleagues evaluated 103 patients with tumors equal to or greater than T3 stage at initial presentation. MR's accuracy in predicting local tumor invasion was significantly higher than that of CT and ultrasound. Sensitivity and specificity for detecting mesorectal invasion were 84% and 68% for MR, 69% and 55% for CT, and 71% and 28% for ultrasound, respectively.

A separate comparison between T1- and T2-weighted MR images found no significant differences. Stollfuss suggested that it might not be necessary to give gadolinium to this patient population.

Responding to a question, Stollfuss said that multidetector CT may produce higher sensitivity than MR but will not improve specificity.

In another study, researchers from the Netherlands rated ultrasound's accuracy in assessing rectal cancer higher than that of MR and CT. However, an audience member pressed lead investigator Dr. Shandra Bipat to explain these findings, which are not in line with general experience. Bipat acknowledged that her research may have overestimated ultrasound because more ultrasound studies were included.

Bipat and colleagues reviewed the literature from 1985 to 2001 and found better sensitivity and specificity for endoluminal ultrasonography in assessing local rectal invasion, compared with CT and MR. Ultrasound's detection of lymph node involvement was comparable to CT and MR.

The selected literature had to fulfill the following criteria: minimal sample size of 20 patients, histopathology as a reference standard, sufficient data available to construct 2 x 2 tables, and original data.

Of 147 identified articles, 85 fulfilled all inclusion criteria. The summary estimates of sensitivity and specificity for endoluminal ultrasound in assessing local invasion were:

? 97% and 78% for T2 stage
? 93% and 81% for T3 stage
? 85% and 99% for T4 stage

CT and MRI each had lower values.

Summary sensitivity for lymph node involvement was between 59% and 69%; specificity was between 75% and 80%.

Although ultrasound received high marks, it has major limitations, Bipat said. It is suitable only for tumors within 10 cm of the anal canal, and is unsuitable for assessing stenotic tumors or detecting distant lymph nodes.

Data on helical CT are limited, she said, and these studies may increase CT's sensitivity and specificity. Additionally, new developments in MR might improve its numbers.