Gadoxetate Disodium Enhanced MRI to Differentiate Dysplastic Nodules and Grade of Hepatocellular Carcinoma: Correlation With Histopathology

Transcription

1 Gastrointestinal Imaging Original Research Channual et al. Gadoxetate Disodium Enhanced MRI of DNs and HCCs Gastrointestinal Imaging Original Research Stephanie Channual 1 Nelly Tan 1 Surachate Siripongsakun 2 Charles Lassman 3 David S. Lu 1 Steven S. Raman 1 Channual S, Tan N, Siripongsakun S, Lassman C, Lu DS, Raman SS Keywords: dysplastic nodule, gadoxetate disodium, hepatocellular carcinoma DOI:1.2214/AJR S. Channual and N. Tan contributed equally to this study. Received February 11, 214; accepted after revision January 31, 215. D. S. Lu and S. S. Raman are consultants for Bayer HealthCare. 1 Department of Radiological Sciences, David Geffen School of Medicine at UCLA, Ronald Reagan UCLA Medical Center, 757 Westwood Plaza, Ste 1638, Los Angeles, CA 995. Address correspondence to S. Channual (schannua@gmail.com). 2 Department of Radiology, Chalubhorn Hospital, Bangkok, Thailand. 3 Department of Pathology and Laboratory Medicine, Anatomic Pathology & Clinical Pathology, David Geffen School of Medicine at UCLA, Los Angeles, CA. AJR 215; 25: X/15/ American Roentgen Ray Society Gadoxetate Disodium Enhanced MRI to Differentiate Dysplastic Nodules and Grade of Hepatocellular Carcinoma: Correlation With Histopathology OBJECTIVE. The objective of our study was to determine quantitative differences to differentiate low-grade from high-grade dysplastic nodules (DNs) and low-grade from highgrade hepatocellular carcinomas (HCCs) using gadoxetate disodium enhanced MRI. MATERIALS AND METHODS. A retrospective study of 149 hepatic nodules in 127 consecutive patients who underwent gadoxetic acid enhanced MRI was performed. MRI signal intensities (SIs) of the representative lesion ROI and of ROIs in liver parenchyma adjacent to the lesion were measured on unenhanced T1-weighted imaging and on dynamic contrast-enhanced MRI in the arterial, portal venous, delayed, and hepatobiliary phases. The relative SI of the lesion was calculated for each phase as the relative intensity ratio as follows: [mass SI / liver SI]. RESULTS. Of the 149 liver lesions, nine (6.%) were low-grade DNs, 21 (14.1%) were high-grade DNs, 83 (55.7%) were low-grade HCCs, and 36 (24.2%) were high-grade HCCs. The optimal cutoffs for differentiating low-grade DNs from high-grade DNs and HCCs were an unenhanced to arterial SI of or a relative SI on T2-weighted imaging of 1.5, with a positive predictive value (PPV) of 99.2% and accuracy of 88.6%. The optimal cutoffs for differentiating low-grade HCCs from high-grade HCCs were a relative hepatobiliary SI of.5 or a relative T2 SI of 1.5, with a PPV of 81.% and an accuracy of 6.5%. CONCLUSION. Gadoxetate disodium enhanced MRI allows quantitative differentiation of low-grade DNs from high-grade DNs and HCCs, but significant overlap was seen between low-grade HCCs and high-grade HCCs. H epatocellular carcinoma (HCC) is among the most prevalent solid organ cancers and most common causes of cancer-related mortality worldwide. Its incidence and prevalence are increasing in the United States due to viral hepatitis and nonalcoholic steatohepatitis. The carcinogenesis of HCC occurs in a stepwise fashion, comprising the following steps: dysplastic nodule (DN), low-grade HCC, and high-grade HCC [1, 2]. The histologic grade of HCC has been shown to be an important prognostic factor for patient outcome [3]. HCC is unique among solid organ cancers in that a variety of imaging-based criteria have been developed to diagnose HCC; among these criteria, the 21 American Association for the Study of Liver Diseases (AASLD) criteria are the most widely used. These AASLD criteria include the following: lesion size of 1 cm or larger, lesion showing early arterial enhancement greater than background, and lesion showing portal venous or delayed phase washout less than background [4]. However, a subgroup of both DNs and HCC lesions may maintain portal venous flow, making differentiation on dynamic phase imaging challenging [5, 6]. Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (gadolinium-eob- DTPA [gadoxetate disodium, Eovist, Bayer HealthCare]) is a unique hepatocyte-specific contrast agent that was approved for clinical use in the United States in 28. This contrast agent has dual properties functioning as a traditional blood pool agent in the first 5 minutes after injection but thereafter having a balanced excretion between liver (5%) and kidney (5%) due to active hepatocyte uptake and biliary excretion, which peaks and plateaus 2 minutes after injection. In multiple prior studies, gadoxetate disodium enhanced MRI has been shown to improve detection and characterization of liver lesions [7 9]. However, the ability of gadoxetate disodium enhanced MRI for differentiating DNs from HCCs and grading HCC is not well understood. In addition, the 546 AJR:25, September 215

2 Gadoxetate Disodium Enhanced MRI of DNs and HCCs results of prior studies evaluating gadoxetate disodium enhanced MRI for differentiating among the grades of HCCs and DNs have been equivocal [6, 1 19]. The purpose of this study was to determine the quantitative ability of gadoxetate disodium enhanced MRI to differentiate DNs from HCCs. Materials and Methods Patient Population An institutional review board approved HIPAAcompliant retrospective study was performed, and the requirement for informed consent was waived. We searched our imaging database to identify all patients who underwent gadoxetate disodium enhanced MRI from January 28 through September 212. From this group, we derived a study cohort of 127 consecutive patients with 149 histologically proven nodules (HCC or DN) within 6 months of MRI (Table 1). Each patient s Child-Pugh score was calculated by reviewing medical records for laboratory results (international normalized ratio, albumin, total bilirubin) and clinical findings (presence and severity of encephalopathy and ascites) as described previously [2, 21]. Imaging MRI was performed using a 1.5-T magnet (Sonata or Avanto, Siemens Healthcare) or a 3-T magnet (Trio or Verio, Siemens Healthcare) with a phased-array coil for signal reception. In general, sequences were similar across platforms but were optimized for each scanner. The following sequences were performed before IV injection of contrast material: dual-echo spoiled gradient-echo T1-weighted FLASH, single-shot T2-weighted HASTE, multishot 3D T2-weighted TSE sequence with a RESTORE pulse (Siemens Healthcare), and 3D gradient-echo volumetric interpolated breath-hold examination (VIBE) with fat saturation. Dynamic imaging was performed after the administration of gadoxetate disodium (.25 mmol/ml solution; Eovist, Bayer Health- Care) at a dose of.1 ml/kg of body weight followed by a 2-mL saline flush (.9%); both were administered at a 2 ml/s rate using a power injector. Dynamic 3D gradient-echo VIBE sequences were performed at the arterial, portal venous, delayed, and hepatobiliary phases after IV contrast injection. Fellowship-trained abdominal imagers with more than 1 years of experience interpreting abdominal MRI reviewed all studies. Quantitative Measurements We measured lesion diameter by determining the largest diameter of the enhancing portion of TABLE 1: Demographic and Descriptive Characteristics of Patients and Tumors Characteristic the mass (Fig. 1). The signal intensities (SIs) of the largest representative enhancing ROI in the lesion and of three adjacent liver parenchyma ROIs were also collected in the arterial, portal venous, delayed, and hepatobiliary phases. The SIs of the lesion ROI and adjacent liver parenchyma ROIs were also measured on the corresponding unenhanced T1-weighted imaging phase. The relative intensity ratio (RIR) was calculated by dividing the SI of the lesion by the SI of the liver. The imagers were blinded to pathology reports during these measurements. Qualitative Measurements Diffuse liver disease was suggested when any one or more of the following findings were present: nodular liver contour, widened liver fissures, relative enlargement of the caudate and left lateral Value Total no. of patients 127 Age (y) Median (range) 64.7 (21 88) Sex, no. (%) of patients M 89 (7) F 38 (3) Lesion size (cm) Median (IQR) 2.3 ( ) No. (%) of patients with cirrhosis 116 (91.3) Child-Pugh classification, no. (%) of tumors A 13 (87.2) B 13 (8.7) C 6 (4.) Total no. of tumors 149 No. of tumors per patient, no. (%) of patients 1 Tumor 19 (85.8) 2 Tumors 14 (11.) 3 Tumors 4 (3.1) Source of pathologic diagnosis, no. (%) of tumors Biopsy specimens 127 (85.2) Explants 11 (7.4) Resected tumors 11 (7.4) No. of tumors with histologic results 149 Histologic diagnosis, no. (%) of tumors Low-grade DN 9 (6.) High-grade DN 21 (14.1) Low-grade HCC 83 (55.7) High-grade HCC 36 (24.2) Note IQR = interquartile range, DN = dysplastic nodule, HCC = hepatocellular carcinoma. segment compared with the right lobe, or hepatic fatty infiltration. Cirrhosis was diagnosed if the liver contour was nodular or irregular. Pathology All the pathologically proven HCCs and DNs were reviewed by gastrointestinal pathologists with 8 and 15 years of experience, respectively. We subdivided the DNs into low-grade DNs and high-grade DNs and the HCCs into low-grade HCCs and high-grade HCCs. A low-grade HCC was defined as grade 1 or 2 or if the pathologist reported well differentiated or moderately differentiated on the pathology report. A highgrade HCC was defined by the presence of grade 3 or 4 diagnosis or if the pathologist noted poorly differentiated HCC on the pathology report. AJR:25, September

3 Channual et al. TABLE 2: Single and Multifocal Tumors With Corresponding Histopathologic Diagnoses Histopathologic confirmation of DNs and HCCs was made by core needle biopsy in 127 lesions (85.2%), liver explants in 11 (7.4%), and resected tumors in 11 (7.4%). Statistical Analysis Medians, SDs, and interquartile ranges (IQRs) are reported for continuous variables as appropriate. Proportions and percentages were calculated for categoric variables. The nonparametric Kruskal-Wallis test was used for comparisons of continuous variables, and posthoc pairwise comparisons were made using the Mann-Whitney U test. The Bonferroni correction was used to adjust for multiple testing. Analyses were performed using statistics software (Stata/ SE, version 12.1, StataCorp), and a p value of <.25 was considered statistically significant. The cutoff values to measure performance (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV], and accuracy) were determined visually at thresholds that yielded the best diagnostic value. Results Of the 127 patients, 89 (7%) were male and 38 (3%) were female (Table 1). The median age was 64.7 years (range, years). Cirrhosis was seen in 116 of the 127 patients (91.3%). When lesions are stratified by Child-Pugh score, 13 (87.2%) are class A; 13 (8.7%), class B; and six (4%), class C. There were a total of 149 liver lesions in the 127 patients. The median size of the 149 liver lesions was 2.3 cm, with an IQR of cm. There was no significant difference in size among the lowgrade DNs, high-grade DNs, low-grade HCCs, and high-grade HCCs (p >.25). Nine of the lesions were low-grade DNs (6.%), 21 were high-grade DNs (14.1%), 83 were low-grade HCCs (55.7%), and 36 were high-grade HCCs (24.2%). One hundred nine patients had a single lesion, 14 patients had two lesions, and four patients had three lesions (Table 2). Therefore, there were 18 patients with two or more lesions, accounting for a total of 4 tumors (Table 3). The mean values of lesion-toliver RIRs between low-grade DNs and high-grade DNs and between low-grade HCCs and high-grade HCCs are summarized in Table 4. TABLE 3: Histopathologic Diagnoses of Tumors for 18 Patients With Two or More Tumors a No. of Patients Tumors No. of Tumors Low-Grade DN High-Grade DN Low-Grade HCC High-Grade HCC Total Per Patient Note DN = dysplastic nodule, HCC = hepatocellular carcinoma. a Total number of tumors = 4. 1 Lesion Per Patient (19 Patients) 2 Lesions Per Patient (14 Patients) 3 Lesions Per Patient (4 Patients) Total (127 Patients) No. of tumors Histopathologic diagnosis, no. of lesions/total no. of lesions (%) Low-grade DN 8/19 (7.3) 1/28 (3.6) 9/149 (6.) High-grade DN 12/19 (11.) 7/28 (25.) 2/12 (16.7) 21/149 (14.1) Low-grade HCC 56/19 (51.4) 17/28 (6.7) 1/12 (83.3) 83/149 (55.7) High-grade HCC 33/19 (3.3) 3/28 (1.7) 36/149 (24.2) Note DN = dysplastic nodule, HCC = hepatocellular carcinoma. Low-Grade Dysplastic Nodule Versus High-Grade Dysplastic Nodule and Hepatocellular Carcinoma On T2-weighted imaging, low-grade DNs had an RIR similar to high-grade DNs, whereas the RIR of low-grade DNs was lower than the RIRs of low-grade HCCs and high-grade HCCs (p <.1; Table 4). In addition, on unenhanced T1-weighted imaging, low-grade DNs had a higher RIR than high-grade DNs and HCCs, although the difference was not statistically significant (p =.3). During the hepatobiliary phase, low-grade DNs also had a higher RIR than high-grade DNs and HCCs, but this difference was also not statistically significant (p =.1). During the arterial, portal venous, and delayed phases, the RIRs for low-grade DNs, high-grade DNs, and HCCs were also similar (Table 4). The low-grade DNs had a difference in unenhanced to arterial SI of.6, which was significantly lower than high-grade DNs and HCCs (p <.1). A relative SI on T2-weighted imaging of 1.5 was the optimal cutoff for differentiating lowgrade DNs from high-grade DNs and HCCs (Fig. 2). An unenhanced to arterial SI of was the optimal cutoff for differentiating lowgrade DNs from high-grade DNs and HCCs (Fig. 2). An unenhanced to arterial SI of or relative T2 SI of 1.5 differentiated lowgrade DNs from high-grade DNs and HCCs with a PPV of 99.2% and accuracy of 88.6% (Table 5 and Fig. 3). Low-Grade Hepatocellular Carcinoma Versus High-Grade Hepatocellular Carcinoma Low-grade HCCs had lower RIRs on T2- weighted imaging than high-grade HCCs (p =.2) and higher RIRs in the hepatobiliary phase than HCCs (p =.1; Table 4 and Fig. 2). When relative SI on T2-weighted imaging and on hepatobiliary phase imaging were stratified by low-grade HCCs and high-grade HCCs, there was a trend for the two groups to separate, but there was still significant overlap 548 AJR:25, September 215

4 Gadoxetate Disodium Enhanced MRI of DNs and HCCs TABLE 4: Relative Signal Intensity (SI) Defined as [Mass SI / Liver SI] at Different Sequences Relative SI and Difference in Relative SI Low-Grade DN High-Grade DN Low-Grade HCC High-Grade HCC p between the two groups (Fig. 4). The optimal cutoffs for differentiating low-grade HCC from high-grade HCC were a relative hepatobiliary SI of.5 or a relative T2 SI of 1.5 (Fig. 4). The performance of T2 or hepatobiliary SI in differentiating low-grade HCC from high-grade HCC is shown in Table 5. Relative Intensity Ratio Over Time After Injection of Gadoxetate Disodium There was significant overlap in the SDs of the RIR means when relative mean SI change between unenhanced and contrast-enhanced T1-weighted imaging was plotted over time for each tumor grade (Fig. 5). Low-grade DNs did not appear to enhance on the arterial phase when compared with high-grade DNs and HCCs (Fig. 5). Low-grade HCCs and high-grade HCCs showed faster enhancement from the unenhanced phase to the arterial phase of T1-weighted imaging when compared with high-grade DNs. Higher-grade lesions showed earlier deenhancement from the arterial phase to the portal venous phase. Discussion In this study, we found that low-grade DNs show lower RIRs on T2-weighted imaging and higher unenhanced to arterial SIs when compared with high-grade DNs and p for p for Low-Grade DN Low-Grade HCC vs High-Grade vs High-Grade DN and HCC HCC Relative SI, median (IQR) T2-weighted imaging 1. ( ) 1. ( ) 1.33 ( ) 1.62 ( ) <.1 a <.1 a.2 a T1-weighted imaging Unenhanced phase 1.1 ( ).98 ( ).94 ( ).84 (.71.99).1 a.3.3 Arterial phase.98 ( ) 1.1 ( ) 1.16 ( ) 1.1 ( ) Portal venous phase.86 (.78.96).92 ( ).91 ( ).85 ( ) Delayed phase.73 (.69.89).87 (.79.94).82 (.74.95).75 (.69.95) Hepatobiliary phase.86 (.75.94).75 (.72.84).79 (.64.9).62 (.53.79).1 a.1.1 a Difference in relative SI, median (IQR) Between unenhanced and.6 (.12 to.3).5 (.14 to.24).21 (.9.41).28 (.6.5) <.1 a <.1 a.38 arterial phases b Between arterial and.24 (.29 to.13).19 (.45 to.5).33 (.52 to.14).34 (.53 to.17) a portal venous phases c Between delayed and.4 (.5 to.9).13 (.17 to.3).6 (.15 to.1).12 (.22 to.3) hepatobiliary phases d Note The nonparametric Kruskal-Wallis one-way ANOVA was performed followed by Mann-Whitney U pairwise comparison for posthoc analysis with Bonferroni correction for multiple testing. DN = dysplastic nodule, HCC = hepatocellular carcinoma, IQR = interquartile range. a Indicates statistical significance after p value adjustment with the Bonferroni correction. A p value of <.25 was considered to be statistically significant. b Enhancement. c Deenhancement. d Referred to as Eovist dropout. (Eovist [gadoxetate disodium] is manufactured by Bayer HealthCare.) TABLE 5: Highlights of Results Purpose Helpful Parameter Recommended Cutoffs To differentiate low-grade DNs from high-grade DNs and HCCs To differentiate low-grade HCCs from high-grade HCCs Difference in relative SI between unenhanced and arterial phases and relative SI on T2-weighted imaging Relative SI on T2-weighted imaging and relative SI in hepatobiliary phase Unenhanced to arterial SI or relative T2 SI of 1.5 Relative T2 SI 1.5 or relative hepatobiliary SI.5 Performance, No. of Lesions/Total No. of Lesions (%) Sensitivity Specificity PPV NPV Accuracy 124/14 (88.6) 8/9 (88.9) 124/125 (99.2) 8/24 (33.3) 132/149 (88.6) 25/36 (69.4) 47/83 (56.6) 47/58 (81.) 25/61 (41.) 72/119 (6.5) Note PPV = positive predictive value, NPV = negative predictive value, DN = dysplastic nodule, HCC = hepatocellular carcinoma, SI = signal intensity. HCCs. Moreover, high-grade HCCs had lower RIRs on hepatobiliary phase imaging and higher RIRs on T2-weighted imaging compared with low-grade HCCs. Together, our results suggest that quantitative parameters may help distinguish the grade of DN and HCC. However, quantitative parameters were found to be less reliable for differentiating low-grade HCC from high-grade HCC. RIR on T2-weighted imaging and unenhanced to arterial SI were the two parameters that best differentiated low-grade DNs from high-grade DNs and HCCs. As shown in pathologic studies, the development of neovascularity seen with HCC is one of the AJR:25, September

5 Channual et al. biologic features that separate HCCs from benign DNs [2, 22]. Others have shown qualitative relative arterial enhancement as a key defining feature using extracellular gadolinium agents [14, 15]. Our findings using a hepatocyte-specific agent are similar to theirs. Lowgrade DNs did not appear to enhance during the arterial phase compared with high-grade DNs and HCCs, and a cutoff of in unenhanced to arterial SI allows differentiating low-grade DNs from high-grade DNs and HCCs. We concurrently evaluated the performance of T2 SI and found that, similar to other studies [13 15], T2 hyperintensity predicted high-grade DNs and HCCs as being more likely than low-grade DNs. RIR on T2-weighted imaging and relative arterial enhancement showed a high PPV (99.2%) for differentiating low-grade DNs from highgrade DNs and HCCs. Distinguishing between high-grade DNs and low-grade DNs is important because high-grade DNs are considered to be premalignant, whereas lowgrade DNs are considered to be benign [23]. Distinguishing between DNs and lowgrade HCCs on imaging may be challenging because low-grade HCCs are more likely to maintain portal venous flow compared with high-grade HCCs [5, 6, 24 26]. In addition, both DNs and low-grade HCCs frequently show hyperintensity on unenhanced T1-weighted imaging [13, 27]. In the current study, there were no differences in the portal venous and delayed phase RIRs between DNs and HCCs. However, unenhanced T1 RIRs appeared to be higher for low-grade lesions compared with high-grade lesions, which is consistent with previous studies [16, 26, 27]. In addition, similar to the results of previous studies [6, 11], the results of our study showed that HCC grade correlated with a difference in relative SI in the hepatobiliary phase. High-grade HCC had a lower RIR compared with low-grade HCC in the hepatobiliary phase. This difference in RIR may be because of the change in OATP1B3 receptor expression with carcinogenesis. The OATP1B3 receptor is expressed in normal hepatocytes; as suggested previously, with dedifferentiation of HCC cancer cells, there is reduced expression of the OATP1B3 receptor and, thus, reduced uptake of gadoxetate disodium [28]. We speculate that high-grade HCC lesions are less likely to express the OATP1B3 receptor than low-grade HCC lesions. However, when relative T2 SI and relative hepatobiliary SI were stratified by low-grade HCC and high-grade HCC, we found a significant overlap between the two HCC grades. Relative T2 SI and hepatobiliary SI had moderate accuracy (6.5%) for differentiating lowgrade HCC from high-grade HCC. Therefore, quantitative parameters cannot be used to reliably distinguish between low-grade HCC and high-grade HCC on T2-weighted imaging and in hepatobiliary phase. Our findings are similar to those of Choi et al. [29]: They also found the degree of tumor enhancement or contrast enhancement ratio cannot be used for determination of tumor grade even though most HCCs are predominantly hypointense relative to surrounding liver parenchyma on hepatobiliary phase imaging. In addition, although highgrade DNs and HCCs overall had lower RIRs in the hepatobiliary phase compared with lowgrade DNs, this difference was not found to be significant (p =.1). Similar RIRs in the hepatobiliary phase were also found for highgrade DNs and low-grade HCCs (p >.25), which suggests that there may be a significant overlap in the OATP1B3 receptor expression in high-grade DNs and low-grade HCCs. Nevertheless, there was a significant difference in the RIRs between low-grade HCCs and highgrade HCCs in the hepatobiliary phase (p =.1), as stated earlier. There are some limitations to our study that extend beyond the limitations inherent to a retrospective study. First, most of the pathologic confirmations were based on biopsy specimens, so sampling error could be a limitation. The preference to treat potential transplant patients with HCC with thermal ablation at our institution precludes our ability to collect a surgical specimen. Another limitation of our study is the small sample of DNs in our cohort. However, we have a comparable number of DNs compared with the studies in the literature. We also did not assess for the presence of a capsule in our study, which is quite specific for HCC. In addition, we did not evaluate for fat within lesions, which has been shown to be important in differentiating DNs from early-stage HCCs [15]. We also included patients with Child-Pugh class B and C disease, which may affect the hepatic function to the enhancement of hepatic parenchyma. However, the differences between the Child-Pugh class and nodules were not significant. Finally, the study may not reflect the true performance of MRI because lesions with normal or negative biopsy findings were excluded. Despite these limitations, to our knowledge, we are the first to describe the combination of T2 characteristics and contrast enhancement nature of HCCs and DNs. In addition, we used objective quantitative values of SI to differentiate DNs from HCCs that are reproducible rather than qualitative interpretation criteria, which may be subject to bias and radiologists experience. Conclusions Gadoxetate disodium enhanced MRI allows quantitative differentiation of lowgrade DNs from high-grade DNs and HCCs, whereas significant overlap was seen between low-grade HCCs and high-grade HCCs. A relative change in SI between the unenhanced and arterial phases was the most helpful parameter in differentiating lowgrade DNs from high-grade DNs and HCCs, because low-grade DNs do not appear to enhance in the arterial phase. Acknowledgments We thank Bita Naini for help in acquiring the data and reviewing the pathology slides. We also thank Khobkhoon Ajwichai for help in study conception, study design, and data collection. References 1. Hussain SM, Zondervan PE, IJzermans JN, Schalm SW, de Man RA, Krestin GP. Benign versus malignant hepatic nodules: MR imaging findings with pathologic correlation. RadioGraphics 22; 22: van den Bos IC, Hussain SM, Terkivatan T, Zondervan PE, de Man RA. Stepwise carcinogenesis of hepatocellular carcinoma in the cirrhotic liver: demonstration on serial MR imaging. J Magn Reson Imaging 26; 24: Takayama T, Makuuchi M, Hirohashi S, et al. Early hepatocellular carcinoma as an entity with a high rate of surgical cure. Hepatology 1998; 28: Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. Hepatology 211; 53: Honda H, Tajima T, Taguchi K, et al. Recent developments in imaging diagnostics for HCC: CT arteriography and CT arterioportography evaluation of vascular changes in premalignant and malignant hepatic nodules. J Hepatobiliary Pancreat Surg 2; 7: Saito K, Kotake F, Ito N, et al. Gd-EOB-DTPA enhanced MRI for hepatocellular carcinoma: quantitative evaluation of tumor enhancement in hepatobiliary phase. Magn Reson Med Sci 25; 4: Hammerstingl R, Huppertz A, Breuer J, et al. Diagnostic efficacy of gadoxetic acid (Primovist)- enhanced MRI and spiral CT for a therapeutic strategy: comparison with intraoperative and histopathologic findings in focal liver lesions. Eur Radiol 55 AJR:25, September 215

6 Gadoxetate Disodium Enhanced MRI of DNs and HCCs 28; 18: Huppertz A, Haraida S, Kraus A, et al. Enhancement of focal liver lesions at gadoxetic acidenhanced MR imaging: correlation with histopathologic findings and spiral CT initial observations. Radiology 25; 234: Cruite I, Schroeder M, Merkle EM, Sirlin CB. Gadoxetate disodium-enhanced MRI of the liver. Part 2. Protocol optimization and lesion appearance in the cirrhotic liver. AJR 21; 195: Fujita M, Yamamoto R, Takahashi M, et al. Paradoxic uptake of Gd-EOB-DTPA by hepatocellular carcinoma in mice: quantitative image analysis. J Magn Reson Imaging 1997; 7: Ni Y, Marchal G, Yu J, Mühler A, Lukito G, Baert AL. Prolonged positive contrast enhancement with Gd-EOB-DTPA in experimental liver tumors: potential value in tissue characterization. J Magn Reson Imaging 1994; 4: Vogl TJ, Stupavsky A, Pegios W, et al. Hepatocellular carcinoma: evaluation with dynamic and static gadobenate dimeglumine-enhanced MR imaging and histopathologic correlation. Radiology 1997; 25: Matsui O, Kadoya M, Kameyama T, et al. Adenomatous hyperplastic nodules in the cirrhotic liver: differentiation from hepatocellular carcinoma with MR imaging. Radiology 1989; 173: Winter TC 3rd, Takayasu K, Muramatsu Y, et al. Early advanced hepatocellular carcinoma: evaluation of CT and MR appearance with pathologic correlation. Radiology 1994; 192: Sano K, Ichikawa T, Motosugi U, et al. Imaging study of early hepatocellular carcinoma: usefulness of gadoxetic acid-enhanced MR imaging. Radiology 211; 261: Efremidis SC, Hytiroglou P, Matsui O. Enhancement patterns and signal-intensity characteristics of small hepatocellular carcinoma in cirrhosis: pathologic basis and diagnostic challenges. Eur Radiol 27; 17: Desmet VJ. East-West pathology agreement on precancerous liver lesions and early hepatocellular carcinoma. Hepatology 29; 49: Kojiro M. Diagnostic discrepancy of early hepatocellular carcinoma between Japan and West. Hepatol Res 27; 37(suppl 2):S121 S Marin D, Di Martino M, Guerrisi A, et al. Hepatocellular carcinoma in patients with cirrhosis: qualitative comparison of gadobenate dimeglumine-enhanced MR imaging and multiphasic 64-section CT. Radiology 29; 251: Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973; 6: Lucey MR, Brown KA, Everson GT, et al. Minimal criteria for placement of adults on the liver transplant waiting list: a report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Liver Transpl Surg 1997; 3: International Consensus Group for Hepatocellular Neoplasia. Pathologic diagnosis of early hepatocellular carcinoma: a report of the international consensus group for hepatocellular neoplasia. Hepatology 29; 49: Borzio M, Fargion S, Borzio F, et al. Impact of large regenerative, low grade and high grade dysplastic nodules in hepatocellular carcinoma development. J Hepatol 23; 39: Tanaka Y, Sasaki Y, Katayama K, et al. Probability of hepatocellular carcinoma of small hepatocellular nodules undetectable by computed tomography during arterial portography. Hepatology 2; 31: Tajima T, Honda H, Taguchi K, et al. Sequential hemodynamic change in hepatocellular carcinoma and dysplastic nodules: CT angiography and pathologic correlation. AJR 22; 178: Muramatsu Y, Nawano S, Takayasu K, et al. Early hepatocellular carcinoma: MR imaging. Radiology 1991; 181: Li CS, Chen RC, Lii JM, et al. Magnetic resonance imaging appearance of well-differentiated hepatocellular carcinoma. J Comput Assist Tomogr 26; 3: Narita M, Hatano E, Arizono S, et al. Expression of OATP1B3 determines uptake of Gd-EOB- DTPA in hepatocellular carcinoma. J Gastroenterol 29; 44: Choi JY, Kim MJ, Park YN, et al. Gadoxetate disodium enhanced hepatobiliary phase MRI of hepatocellular carcinoma: correlation with histological characteristics. AJR 211; 197: A B C D Fig. 1 T1-weighted imaging findings for dysplastic nodules (DNs), low-grade hepatocellular carcinomas (HCCs), and high-grade HCCs. A D, 82-year-old woman with hepatitis C. Unenhanced (A), arterial phase (B), portal venous phase (C), and hepatobiliary phase (D) images show low-grade dysplastic nodule (DN) in segment VII of liver. DN appears hyperintense in all phases. (Fig. 1 continues on next page) AJR:25, September

7 Channual et al. Relative SI on T2-Weighted Imaging E I LG DN HG DN LG HCC HG HCC F J Fig. 1 (continued) T1-weighted imaging findings for dysplastic nodules (DNs), low-grade hepatocellular carcinomas (HCCs), and high-grade HCCs. E H, 62-year-old woman who presented with ascites and was found to have hepatitis B and low-grade HCC in segment VI on workup. HCC lesion appears isointense on unenhanced T1-weighted image (E), enhanced on arterial phase image (F), isointense on portal venous phase image (G), and hypointense on hepatobiliary phase image (H). I L, 71-year-old woman with hepatitis C cirrhosis who was found to have high-grade HCC in segment VIII of liver. In setting of hepatic steatosis, HCC lesion is hypointense on unenhanced T1-weighted image (I), appears to enhance on arterial phase image (J), appears to lose signal intensity on portal venous phase image (K), and shows more pronounced signal intensity on hepatobiliary phase image (L). A Relative SI from Unenhanced to Arterial Phase LG DN HG DN LG HCC HG HCC Fig. 2 Relative signal intensity (SI) and difference in relative SI of low-grade (LG) and high-grade (HG) dysplastic nodules (DNs) and hepatocellular carcinomas (HCCs) on MRI. A D, Box plots show relative SI on T2-weighted imaging (A), relative SI difference between unenhanced and arterial phases (B), relative SI on unenhanced T1-weighted imaging (C), and relative SI in hepatobiliary phase (D) for low-grade and high-grade DNs and HCCs. Relative SI was calculated as follows: [mass SI / liver SI]. Middle lines in boxes show median, upper and lower limits of boxes show 25th and 75th percentiles, and whiskers show minimum and maximum limit. = outliers. (Fig. 2 continues on next page) G K H L B 552 AJR:25, September 215

8 2.5 Gadoxetate Disodium Enhanced MRI 1.5 of DNs and HCCs Relative SI on Unenhanced T1-Weighted Imaging LG DN HG DN LG HCC HG HCC C Relative SI in Hepatobiliary Phase 1.5 LG DN HG DN LG HCC HG HCC Fig. 2 (continued) Relative signal intensity (SI) and difference in relative SI of low-grade and high-grade dysplastic nodules (DNs) and hepatocellular carcinomas (HCCs) on MRI. A D, Box plots show relative SI on T2-weighted imaging (A), relative SI difference between unenhanced and arterial phases (B), relative SI on unenhanced T1-weighted imaging (C), and relative SI in hepatobiliary phase (D) for low-grade and high-grade DNs and HCCs. Relative SI was calculated as follows: [mass SI / liver SI]. Middle lines in boxes show median, upper and lower limits of boxes show 25th and 75th percentiles, and whiskers show minimum and maximum limit. = outliers. Difference in Relative SI Between Unenhanced and Arterial Phases Relative SI on T2-Weighted Imaging Low-Grade DN High-Grade DN High-Grade HCC Low-Grade HCC Fig. 3 Scatterplot shows relative signal intensity (SI) on T2-weighted imaging (x-axis) and difference in relative SI between unenhanced and arterial phases of T1-weighted imaging (y-axis) stratified by histologic subtype (color-coded). Shaded blue regions represent visual enhancement cutoff of (horizontal dotted line) and relative T2 SI cutoff of 1.5 (vertical dotted line) that can be used to differentiate low-grade dysplastic nodules (DNs) from high-grade DNs and hepatocellular carcinomas (HCCs). Relative SI was calculated as follows: [mass SI / liver SI]. Mean Relative SI Change Time After Contrast Administration (min) Relative SI in Hepatobiliary Phase Relative SI on T2-Weighted Imaging High-Grade HCC Low-Grade HCC Fig. 4 Scatterplot shows relative signal intensity (SI) on T2-weighted imaging (x-axis) and in hepatobiliary phase (y-axis) stratified by grade of hepatocellular carcinoma (HCC). Relative SI was calculated as follows: [mass SI / liver SI]. Shaded regions represents visual relative SI cutoff in hepatobiliary phase of.5 (horizontal dotted line) and relative T2 SI cutoff of 1.5 (vertical dotted line) that can be used to differentiate low-grade HCC from high-grade HCC. Low-Grade DN High-Grade DN Low-Grade HCC High-Grade HCC Fig. 5 Line graph of mean relative signal intensity (SI) change on unenhanced and contrast-enhanced T1-weighted imaging (y-axis) over time after gadoxetate disodium administration (x-axis) stratified by histologic subtypes. Error bars represent ± 1 SD of mean. Relative SI was calculated as follows: [mass SI / liver SI]. D AJR:25, September