MRI with a lymph-node-specific contrast agent as an alternative to CT scan and lymph-node dissection in patients with prostate cancer:: a prospective multicohort study

Background In patients with prostate cancer who are deemed to be at intermediate or high risk of having nodal metastases, invasive diagnostic pelvic lymph-node dissection (PLND) is the gold standard for the detection of nodal disease. However, a new lymph-node-specific MR-contrast agent ferumoxtran-10 can detect metastases in normal-sized nodes (ie, <8 mm in size) by use of MR lymphoangiography (MRL). In this prospective, multicentre cohort study, we aimed to compare the diagnostic accuracy of MRL with up-to-date multidetector CT (MDCT), and test the hypothesis that a negative MRL finding obviates the need for a PLND. Methods We included consecutive patients with prostate cancer who had an intermediate or high risk (risk of >5% according to routinely used nomograms) of having lymph-node metastases. All patients were assessed by MDCT and MRL, and underwent PLND or fine-needle aspiration biopsy. Imaging results were correlated with histopathology. The primary outcomes were sensitivity, specificity, accuracy, NPV, and PPV of MRL and MDCT. This study is registered with ClinicalTrials.gov, number NCT00185029. Findings The study was done in 11 hospitals in the Netherlands between April 8, 2003, and April 19, 2005. 375 consecutive patients were included. 61 of 375 (16%) patients had lymph-node metastases. Sensitivity was 34% (21 of 61; 95% Cl 23-48) for MDCT and 82% (50 of 61; 70-90) for MRL (McNemar's test p<0.05). Specificity was 97% (303 of 314; 94-98) for MDCT and 93% (291 of 314; 89-95) for MRL. Positive predictive value (PPV) was 66% (21 of 32; 47-81.) for MDCT and 69% (50 of 73; 56-79) for MRL. Negative predictive value (NPV) was 88% (303 of 343; 84-91) for MDCT and 96% (291 of 302; 93-98) for MRL (McNemar's test p<0.05). Of the 61 patients with lymph-node metastases, 50 were detected by MRL, of which 40 (80%) had metastases in normal-sized lymph nodes. The high sensitivity and NPV of MRL imply that in patients with a negative MRL, the chance of positive lymph nodes is less than 11/302 (4%). Interpretation MRL had significantly higher sensitivity and NPV than MDCT for patients with prostate cancer who had intermediate or high risk of having lymph-node metastases. In such patients, after a negative MRL, the post-test probability of having lymph-node metastases is low enough to omit a PLND.