Patients with suspected or known focal liver lesions were enrolled in two of four non-randomized, intrapatient-controlled studies that evaluated predominantly the detection (studies 1 and 2) or morphological characterization (studies 3 and 4) of liver lesions. Studies 1 and 2 ("detection" studies) enrolled patients who were scheduled for liver surgery. MRI results were compared to a reference standard that consisted of surgical histopathology and the results from intra-operative ultrasound of the liver. The studies assessed the sensitivity of pre-contrast MRI and EOVIST-contrasted MRI for the detection of liver lesions, when each set of images was compared to the reference.
Studies 3 and 4 ("characterization" studies) enrolled patients with known or suspected focal liver lesions, including patients who were not scheduled for liver surgery. MRI results were compared to a reference standard that consisted of surgical histopathology and other prospectively defined criteria. The studies assessed the correctness of liver lesion characterization by pre-contrast MRI and EOVIST-contrasted MRI, when each set of images was compared to the reference. Lesions were characterized as one of the following choices: hepatocellular carcinoma, cholangiocarcinoma, metastasis, focal lymphoma, adenoma, focal nodular hyperplasia, hemangioma, abscess, focal liver fibrosis, regenerative nodule, focal fat, hydatid cyst, liver cyst, "not assessable", normal, no lesion or "other."
In all four studies, patients underwent a baseline, pre-contrast MRI followed by the administration of EOVIST at a dose of 0.025 mmol/kg, with MRI performed immediately (the "dynamic" phase) and at 10 to 20 minutes following EOVIST administration (the "hepatocyte" phase). Patients also underwent computerized tomography with contrast examinations of the liver. Pre-contrast MRI and EOVIST-contrasted MR images were evaluated in a systematic, randomized, paired and unpaired fashion by three radiologists who were blinded to clinical information. CT images were also evaluated by the radiologists in a separate reading session.
Diagnostic efficacy was determined in 621 patients. The average age was 57 years (range 19 to 84 years) and 54% were male. The ethnic representations were 90% Caucasian, 4% Black, 3% Hispanic, 2% Asian, and 1% of other ethnic groups.
The combination of non-contrasted and EOVIST-contrasted MR images had improved sensitivity for the detection and characterization of liver lesions, compared to pre-contrasted MR images (Tables 3 and 4). The improved sensitivity in detection of lesions was predominantly related to the detection of additional lesions among patients with multiple lesions on the pre-contrast MR images. The false positive rates for detection of lesions were similar for non-contrasted MR images and EOVIST-contrasted MR images (32% versus 34%, respectively). Liver lesion detection and characterization results were similar between CT and the combination of pre-contrasted and EOVIST-contrasted MR images.
Table 3 Sensitivity in Liver Lesion DetectionDiagnostic Procedure | Reader | Study 1 Sensitivity (%) n=129 | Study 2 Sensitivity (%) n=126 |
Pre-contrast MRI | | | |
Reader 1 | 76 | 77 |
Reader 2 | 76 | 73 |
Reader 3 | 71 | 72 |
Combined pre- and EOVIST-contrast MRI | | | |
Reader 1 | 81 | 82 |
Reader 2 | 78 | 76 |
Reader 3 | 74 | 78 |
Difference: combined pre + EOVIST-contrast MRI minus pre MRI (95% confidence interval) | | | |
Reader 1 | 5 (1, 9)* | 5 (1, 9)* |
Reader 2 | 2 (-1, 5) | 3 (-1, 7) |
Reader 3 | 3 (0, 6)* | 6 (0, 10)* |
* Statistically significant improvement
Table 4 Proportion of Correctly Characterized Lesions | | Study 3 | Study 4 |
Diagnostic Procedure | Reader | n | Proportion correct (%) ** | n | Proportion correct (%) ** |
Pre-contrast MRI | | | | | |
Reader 1 | 182 | 51 | 177 | 60 |
Reader 2 | 182 | 59 | 177 | 64 |
Reader 3 | 182 | 53 | 177 | 48 |
Combined pre- and EOVIST-contrast MRI | | | | | |
Reader 1 | 182 | 67 | 177 | 61 |
Reader 2 | 182 | 76 | 177 | 76 |
Reader 3 | 182 | 58 | 177 | 67 |
Difference: combined pre- and EOVIST-contrast MRI minus pre-contrast MRI (95% confidence interval) | | | | | |
Reader 1 | | 16 (7, 25)* | | 1 (-7, 10) |
Reader 2 | | 17 (9, 25)* | | 11 (5, 18)* |
Reader 3 | | 5 (-2, 12) | | 19 (11, 27)* |
* Statistically significant improvement
** Proportion of correctly characterized lesions with respect to the reference