Gadoxetic Acid (Gd-EOB-DTPA)-Enhanced MRI versus Gadobenate Dimeglumine (Gd-BOPTA)-Enhanced MRI for Preoperatively Detecting Hepatocellular Carcinoma: an Initial Experience

Study Population

This is a retrospective study. Between January 2007 and January 2008, 65 consecutive patients who were suspected of having HCCs based on a sonographic evaluation and the presence of an elevated level of α-fetoprotein underwent gadobenate dimeglumine-enhanced MRI at our institution. Among these patients, 30 patients also underwent gadoxetic acid-enhanced MRI. For all the patients, gadoxetic acid-enhanced MRI was performed at an interval of five to 29 days after performing gadobenate dimeglumine-enhanced MRI. Of the 30 patients who also underwent gadoxetic acid-enhanced MRI, 18 patients (17 men and one woman, age range: 31-73 years, mean age: 53.3 years) underwent hepatic surgery with identifying 22 pathologically proven HCCs and these 18 patients were eventually enrolled in this study. Of these 18 patents, a portion of these patients, who were enrolled in a previous report (13), were included because this study's retrospective study design. All the hepatic surgeries were performed within five days from the last pre-operative gadoxetic acid enhanced MRI. Seven patients underwent segmentectomy, six patients underwent bisegmentectomy and five patients underwent right lobectomy. The 22 HCCs ranged from 0.5 cm to 10.0 cm in diameter (two HCCs were < 1 cm in diameter, six HCCs were 1 cm ≤ diameter < 2 cm, and 14 HCCS were ≥ 2 cm in diameter, mean diameter, 2.9 cm). The histological grades of the tumors included two well-differentiated HCCs, 17 moderately differentiated HCCs and three poorly-differentiated HCCs. Twelve patients (11 patients with hepatitis B virus and one with hepatitis C virus) had cirrhosis, five patients had chronic B virus related hepatitis and one patient had a normal liver. Written informed consent for the liver MRI examinations were obtained from all the patients. Approval by the Institutional Review Board was waived because this was a retrospective study.

MR Imaging

MRI examinations were performed with a 3.0 Tesla whole-body MR system (Intera Achieva 3.0T; Philips Medical Systems, Best, The Netherlands) with a 6-channel phased-array coil as the receiver coil. The MRI protocols and sequences were as follows: a respiratory triggered T1-weighted turbo-field-echo in-phase sequence (TR/TE: 10 ms/2.3 ms, flip angle: 15°, matrix: 288 × 230, bandwidth: 434.3 Hz per pixel), an out-of-phase sequence (TR/TE: 10 ms/3.45 ms, flip angle: 15°, matrix: 288 × 230, bandwidth: 434.3 Hz per pixel), a respiratory-triggered single-shot T2-weighted sequence with a reduction factor of two or four (TR/TE: 1,342 ms/80 ms, flip angle: 90°, matrix: 320 × 256, bandwidth: 506.4 Hz per pixel), a breath-hold multi-shot T2-weighted sequence with a reduction factor of two or four (TR/TE: 2,161 ms/70 ms, flip angle: 90°, matrix: 400 × 280, bandwidth: 235.2 Hz per pixel), a respiratory-triggered single-shot heavily T2-weighted sequence with a reduction factor of two or four (TR/TE: 1,573 ms/160 ms, flip angle: 90°, matrix: 320 × 256, bandwidth: 317.9 Hz per pixel) with a 5-7 mm section thickness, a 1-mm to 2-mm intersection gap and a field of view of 32-38 cm. A dose of 0.1 mL/kg (0.025 mmol/kg gadoxetic acid; 0.05 mmol/kg gadobenate dimeglumine) contrast agent was manually administered intravenously at a rate of 2 mL/sec, followed by a 20 mL saline solution flush. For the contrast-enhanced MRI, the unenhanced, arterial phase (20-35 sec), portal phase (70 sec), equilibrium phase (3 min) and hepatobiliary phase sequences were obtained using T1-weighted three-dimensional turbo-field-echo imaging (T1 high resolution isotropic volume examination; THRIVE, Philips Medical Systems, Best, The Netherlands) (TR/TE: 3.4 ms/1.8 ms, flip angle: 10°, matrix: 336 × 206, bandwidth: 995.7 Hz per pixel) with a 2 mm section thickness with no intersection gap and a field of view of 32-38 cm. The hepatobiliary phase images were obtained 20 minutes after the administration of gadoxetic acid. For gadobenate dimeglumine, the hepatobiliary phase images were obtained three hours after the administration to minimize its extracellular effect.

Image Analysis

Three radiologists who each had at least two years experience with interpreting MR images reviewed all of the MR images on a 2,000 × 2,000 picture archiving and communications system (PACS; PathSpeed workstation; GE Healthcare, Milwaukee, WI) workstation in a random order and in a blinded fashion. They were unaware which contrast agent was used when reading the images. The interval between the reviews of the two MRI examinations was at least one month. Each observer recorded the presence and segment location of the HCC lesions by using a four-point scale to assign a confidence level to each lesion. The scores were 1: probably not an HCC lesion, 2: a possible HCC lesion, 3: a probable HCC lesion and 4: a definite HCC lesion. An HCC was defined as showing typical arterial enhancement and venous washout irrespective of its size, based on previous studies (14-19). Nodules larger than 2 cm with predominant hypointensity, as seen on the contrast-enhanced portal and equilibrium phases with no definite arterial enhancement, were also regarded as HCCs (19, 20). In addition to these features, a hypointense nodule seen on the hepatobiliary phase MR images was considered as an HCC based on the findings of a previous report (21). A hypervascular nodule seen on the contrast-enhanced arterial phase with a washout pattern was considered as an HCC, although the nodule was seen with isointensity or hyperintensity relative to the surrounding liver parenchyma on the hepatobiliary phase (21, 22). To achieve accurate correlation between the findings of the scored lesions and the reference standard, each observer recorded the individual image number, the locations of all of the lesions and the diameter of each lesion. For patients with multiple lesions located in the same segment, the observers added further descriptions regarding the size and location of the lesion within each segment in order to avoid confusion during the data analysis. The sensitivities were calculated using the number of lesions assigned a confidence level of 3 or 4. All false-positive findings with confidence levels of 3 or 4 and that were confirmed as benign lesions, or all false-negative findings with confidence levels of 1 or 2 and that were confirmed as an HCC, were assessed by two study coordinators who worked in consensus.

For the quantitative analysis, a radiologist chose the regions of interest on the MR images. For heterogeneous lesions, the regions of interest were chosen in the more homogeneous areas. Areas of intratumoral necrosis or hemorrhage were excluded from the regions of interest. Measurements of tumors and the background along the phase-encoding axis were obtained from the same areas on the six images. The contrast-to-noise ratios (CNRs) were calculated using (SIL-SIBL)/SDN, where SIL is the signal intensity (SI) of the lesion, SDN is the standard deviation of the background noise and SIBL is the mean SI of the background liver. For measurement purposes, 17 tumors that were 1 cm or more in diameter and visible on the images obtained with all the pulse sequences were chosen for the analysis in order to reduce any measurement inaccuracies resulting from partial volume effects.

Statistical Analysis

On the basis of the three observers' reviews, an alternative free-response receiver operating characteristic (ROC) curve was generated on a tumor-by-tumor basis. The diagnostic performance of each contrast agent for each observer was assessed by measuring the area under the ROC curve (Az), in accordance with a previous report (23), and the average Az values for the two contrast agents were compared as previously described by McNeil et al. (24).

The sensitivity for each observer and technique was calculated and the differences of the sensitivities were assessed by the McNemar test. Statistical analysis for the overall sensitivities of the two techniques for all observers was assessed by the generalized estimating equation. Statistical analyses for the differences of the calculated positive predictive values for each observer and technique were based on a previous study (25). A p value less than 0.05 was considered to indicate a statistically significant difference. An analysis of all false-positive and false-negative observations was also undertaken. Kappa statistics were used to assess the interobserver agreement for the detection of HCCs with the use of each technique. The degree of agreement was categorized as follows: kappa values of 0.00-0.20 were considered to indicate poor agreement, kappa values of 0.21-0.40 indicated fair agreement, kappa values of 0.41-0.60 indicated moderate agreement, kappa values of 0.61-0.80 indicated good agreement and kappa values of 0.81-1.00 indicated excellent agreement (26).

In the quantitative image analysis, the tumor to liver CNR of each phase was statistically compared using the Wilcoxon signed ranks test.