Does the hepatocellular phase of gadobenate dimeglumine help to differentiate hepatocellular carcinoma in cirrhotic patients according to histological grade?

Clinical Radiology 66 (2011) 845e852

Contents lists available at ScienceDirect

Clinical Radiology journal homepage: www.elsevierhealth.com/journals/crad

Original Paper

Does the hepatocellular phase of gadobenate dimeglumine help to differentiate hepatocellular carcinoma in cirrhotic patients according to histological grade? B. Tahir a, *, K. Sandrasegaran a, R. Ramaswamy a, K. Bertrand a, R. Mhapsekar a, F.M. Akisik a, R. Saxena b a b

Departments of Radiology & Imaging Sciences, Indiana Univeristy School of Medicine, Indianapolis, IN, USA Department of Pathology, Indiana Univeristy School of Medicine, Indianapolis, IN, USA

art icl e i nformat ion Article history: Received 12 February 2011 Accepted 23 March 2011

AIM: To assess the role of the hepatocellular phase on magnetic resonance imaging (MRI) following gadobenate in characterizing the grade of hepatocellular carcinoma (HCC) in cirrhotic patients. MATERIALS AND METHODS: A retrospective review of the MRI database from October 2004 to February 2009, performed for this Institutional Review Board-approved and Health Insurance Portability and Accountability Act (HIPAA)-complaint study, revealed 237 cirrhotic patients with focal liver lesions. Patients who had both a hepatocellular phase after gadobenate and pathological confirmation of HCC were included. Forty-six patients with 73 HCC were analysed independently by two reviewers for signal characteristics. Absolute contrast-to-noise ratio (CNR) and enhancement ratio (ER) were calculated. Univariate analysis, stepwise logistic regression analysis, and receiver operating characteristic curves (ROC) were performed. RESULTS: The mean age was 61.3 years (range 45 to 78 years). There were 11 females and 35 males, who had 22 well-differentiated (WD HCC), 35 moderately-differentiated (MD HCC), and 16 poorly-differentiated (PD HCC) hepatocellular carcinomas. On visual analysis of the hepatocellular phase, a hyperintense or isointense lesion had a sensitivity and specificity of 45% and 76%, respectively, for WD HCC. On quantitative analysis, the only significant predictor of the grade of HCC was the ER on the hepatocellular phase (p ¼ 0.019 and 0.001 for the two reviewers in logistic regression model). On ROC analysis, an ER of >13% was 47% sensitive and 89% specific in predicting WD HCC histology. CONCLUSION: Although the hepatocellular phase of gadobenate may help to differentiate some cases of WD HCC from the more aggressive grades, there is overlap between the different grades on qualitative and quantitative analysis. Ó 2011 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Introduction * Guarantor and correspondent: B. Tahir, 550 N University Blvd, UH 0279, Indianapolis, IN 46202, USA. Tel.: þ1 (317) 944 1837; fax: þ1 (317) 944 1848. E-mail address: [email protected] (B. Tahir).

Computed tomography (CT) and magnetic resonance imaging (MRI) are the primary investigative tools in the USA for diagnosing and characterizing liver masses. In cirrhotic

0009-9260/$ e see front matter Ó 2011 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.crad.2011.03.021

846

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

Table 1 Patients and hepatocellular carcinoma (HCC) lesions. HCC grade

No. of patients (M:F)

No. of lesions

Mean age, years (range)

Mean size, cm (range)

WD HCC MD HCC PD HCC MD & PD HCC All

11 27 8 35 46

22 35 16 51 73

61.6 62.1 58.1 61.2 61.3

2.6 3.4 2.9 3.2 3.1

(9:2) (20:7) (6:2) (26:9) (35:11)

(48e78) (45e74) (48e74) (45e75) (45e78)

(1.1e8.1) (0.9e13) (1.2e5) (0.9e13) (0.9e13)

WD HCC, well-differentiated HCC; MD HCC, moderately-differentiated HCC; PD HCC, poorly-differentiated HCC.

patients, hepatic lesions larger than 1 cm showing hyperintensity on the arterial phase and washout on the venous phase are considered to be hepatocellular carcinomas (HCCs).1e5 However, subtyping HCC by histological grade is not possible using conventional MRI sequences, such as precontrast T1- and T2-weighted and dynamic postgadolinium sequences. Several papers suggest that the histological grade of HCC, in addition to the tumour size, predicts the outcome after surgical resection,6e11 liver transplantation,12e15 and radiofrequency ablation.16,17 Identifying HCC biology may not only help evaluate prognosis, but it may also affect management decisions. For instance, it may be useful to deliver bridging therapy (while awaiting orthotopic liver transplantation) in high-grade HCC patients, exclude poorlydifferentiated HCC patients from transplantation (even if the tumour is less than 5 cm in size),18 and include welldifferentiated histology for transplantation, even if it is outside the Milan criteria.19e22 A study using positronemission tomography (PET) found that fluorodeoxyglucose uptake correlated with tumour grade and 2-year recurrence-free survival.23 As MRI is used often for the diagnosis of HCC, it would be useful to investigate whether MRI techniques may help determine the histological grade of the tumour. The aim of the present study was to assess the role of the hepatocellular phase of gadobenate dimeglumine (MultiHance, Bracco Diagnostics, Princeton, NJ, USA) in differentiating the grade of HCC in cirrhotic patients.

Materials and methods Patients A retrospective study, which was Health Insurance Portability and Accountability Act (HIPAA) compliant, was performed. Institutional Review Board permission was

granted for gathering and analysing clinical and imaging data with a waiver of informed consent. A search of the departmental MRI database between October 2004 and February 2009 for reports of hypervascular liver lesions in patients with cirrhosis revealed 237 consecutive patients. Patients were included if they had histological proof (core biopsy or explant) of HCC and a hepatocellular phase gadobenate sequence (n ¼ 54). Patients were excluded if there was prior liver-directed treatment (e.g., embolization or ablation; n ¼ 5) or if there was portal venous thrombosis (n ¼ 3). Patients with only cytological proof of HCC were excluded as cytology may confirm the diagnosis, but it will not reliably indicate the histological grade of the cancer. The remaining 46 patients with 73 HCC were analysed (Table 1).

MRI examination MRI examinations were performed in a supine position using a 1.5 T MRI machine (Avanto, Siemens Medical Solutions, Erlangen, Germany) and a body phased-array coil. The standard imaging protocol consisted of in- and opposedphase gradient echo imaging, T2-weighted, half-Fourier axial single-shot fast spin-echo (HASTE), and threedimensional, T1-weighted, gradient-echo sequences. Details of the MRI sequences are listed in Table 2. Ten to 20 ml (0.1 mmol/kg body weight) of gadobenate dimeglumine was intravenously injected at a rate of 2 ml/s using a power injector (Medrad Spectris Solaris, Medrad, Indianola, PA, USA). An intravenous saline flush of 20 ml was given after gadolinium injection. Post-gadolinium images were obtained in the late arterial phase (25e40 s after the start of the contrast injection), venous (60e75 s), and hepatocellular phases (120 min).

Image review Two specialist abdominal radiologists with 13 and 12 years of experience in reading abdominal MRI images (K.S., F.A.) independently reviewed the MRI images, and were blinded to the histology of the lesion. The reviewers knew that there was at least one HCC lesion in each study. Regions of interest (ROIs) were placed on hepatocellular cancers and adjacent uninvolved liver in the unenhanced, arterial, and hepatocellular phases. The reviewers noted the image numbers, sizes, and liver segments of HCC lesions, so that measurements made on multiple lesions could be appropriately correlated. The ROIs were larger than 1 cm2. In tumours that were larger than 2 cm, multiple ROIs were placed and the mean signal intensity was recorded. An ROI

Table 2 Magnetic resonance imaging parameters. Sequence

TR/TE (ms)

Flip angle

ST/SG (mm)

NEX

RBW

Matrix

T1-weighted two-dimensional gradient-echo (FLASH) T2-weighted single-shot fast spin-echo (HASTE) Three-dimensional, fat-suppressed, T1-weighted gradient-echoa

123/2.2 (OP), 4.93 (IP) 1110/95 4.98/2.27

70 150 12

7/0.7 5/6 3/0

1 1 1

445 475 300

256  135 256  192 256  144

TR, repetition time; TE, echo time; IP, in-phase; OP, out-of-phase; ST, section thickness; SG, gap between sections; NEX, number of excitations; FLASH, fast lowangle shot; HASTE, half-Fourier turbo spin-echo; RBW, receiver bandwidth in hertz per pixel, the three post-contrast phases were obtained axially. a Used for post-gadobenate sequences.

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

was placed in the air adjacent to the patient to obtain the signal of background noise. The contrast-to-noise ratio (CNR) and enhancement ratio (ER) in the arterial and hepatocellular phases were calculated according to established formula24e27: CNR (absolute) ¼ j(Tumor signal e Liver signal)j/Standard deviation of noise, and ER ¼ (Tumor signal postcontrast-Tumor signal precontrast)/Tumor signal precontrast. Following the quantitative assessment, qualitative assessment was performed in consensus by both reviewers. The HCC lesions were classified as hyperintense, isointense, or hypointense to liver, based on the predominant signal of the lesion. Some lesions showed no predominant signal and were classified as having a mixed signal.

Reference standard A specialist hepatopathologist with 16 years of experience classified the histological specimens as welldifferentiated (G1), moderately-differentiated (G2), and poorly-differentiated (G3) HCC according to the World Health Organization categorization.28 This categorization is based on the original EdmondsoneSteiner classification.29

Statistical analysis Categorical variables from the qualitative assessment were analysed with descriptive statistics and Fisher’s exact test. The quantitative measurements were checked for distribution prior to statistical analysis. The correlations between the measurements of the two reviewers were performed using ShrouteFleiss random effect intraclass correlation coefficient (ICC).30 Differences in variables between the three grades of HCC were performed using analysis of variance. Logistic regression analysis was performed to find the best predictor(s) of the grade of HCC. First, the parameters were assessed univariately as independent variables in the model where the dependent binary variable was low-grade HCC (well-differentiated HCC) or high-grade HCC (moderately- or poorly-differentiated HCC). Parameters with a pvalue less than or equal to 0.10 at univariate analysis were assessed with multivariate analysis. Stepwise forward analysis was performed, until all remaining variables had pvalues less than or equal to 0.05.

847

Receiver operating characteristic (ROC) curve analysis was used to evaluate the usefulness of CNR and ER in evaluating the grade of HCC (well-differentiated HCC versus higher grades of HCC). A two-tailed p-value less than 0.05 was used to indicate statistical significance. Statistical analysis was performed using MedCalc 11.1 (MedCalc Software, Mariakerke, Belgium; for ROC curves) and SPSS 17.0. (SPSS, Chicago, IL, USA).

Results Patients and lesions The epidemiology of the 46 patients is given in Table 1. There was no significant difference in age of patients (p ¼ 0.42) or size of HCC (p ¼ 0.49) in the three histological grades. The causes of cirrhosis were hepatitis C (n ¼ 24), alcohol (n ¼ 10), non-alcoholic steatohepatitis or cryptogenic (n ¼ 11), and miscellaneous (n ¼ 4). Three patients had both alcohol and hepatitis C contributing to cirrhosis development. The histological specimens for individual HCCs were obtained with 18 G core biopsy (n ¼ 33) or from the explanted liver (n ¼ 40). For statistical analysis, each HCC was treated separately. The mean size (standard deviation) of well-differentiated, moderately-differentiated and poorly-differentiated HCC were 2.7 (1.6), 3.4 (2.6), and 2.9 (1.3) cm, respectively. There was no difference in size between well-differentiated HCC and the other types of HCC (p ¼ 0.41).

Qualitative analysis Table 3 lists the results of the qualitative analysis. Hyperintensity on the arterial phase was seen in 77% (17 of 22) of well-differentiated, 60% (21 of 35) of moderatelydifferentiated, and 38% (6 of 16) of poorly-differentiated HCC. When well-differentiated HCC was compared to the more aggressive HCC (MD and PD HCC), the arterial hyperintensity had a sensitivity, specificity, positive predictive value, and negative predictive value of 38.6% (95% CI: 24.4e54.5%), 82.8% (95% CI: 64.2e94.2%), 77.3% (95% CI: 54.6e92.2%), and 47.1% (95% CI: 32.9e61.57%), respectively. The hyperintensity on the arterial phase was not highly associated with well-differentiated HCC (p ¼ 0.07).

Table 3 Qualitative assessment of the three groups of hepatocellular carcinoma (HCC). Phase of gadobenate

Signal intensity

WD HCC

MD HCC

PD HCC

PD/MD HCC

All

Arterial phase

Hyper Iso Hypo Mixed

17 (77%) 0 0 5 (23%)

21 (60%) 7 (20%) 6 (17%) 1 (3%)

6 (38%) 6 (38%) 0 4 (25%)

27 (53%) 13 (25%) 6 (12%) 5 (10%)

44 (60%) 13 (18%) 6 (8%) 10 (14%)

Hepatocellular phase

Hyper Iso Hypo Mixed Total

2 (9%) 7 (32%) 10 (46%) 3 (14%) 22

1 (3%) 9 (26%) 25 (71%) 0 35

0 1 (6%) 15 (94%) 0 16

1 (2%) 10 (20%) 40 (78%) 0 51

3 (4%) 17 (23%) 50 (68%) 3 (4%) 73

WD HCC, well-differentiated HCC; MD HCC, moderately-differentiated HCC; PD HCC, poorly-differentiated HCC; MD/PD HCC, combined group of moderatelydifferentiated and poorly-differentiated HCC.

848

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

Figure 1 A 72-year-old man with a 2.4 cm well-differentiated HCC. (a) Axial arterial phase image showed hyperintense partiallyexophytic lesion (arrowhead). (b) On the hepatocellular phase, the lesion (arrowhead) was mildly hyperintense. This finding was more likely to be seen with well-differentiated histology.

On the hepatocellular phase, 41% (nine of 22) welldifferentiated HCC were hyperintense or isointense (Fig 1). Only 22% (11 of 51) of higher-grade HCC (MD and PD HCC) had these signal characteristics. Hyperintensity or isointensity on the hepatocellular phase had a sensitivity, specificity, positive predictive value, and negative predictive value of 45% (95% CI: 23.1e68.5%), 75.5% (95% CI: 61.7e86.2%), 40.9% (95% CI: 20.7e63.7%), and 78.4% (95% CI: 64.7e88.7%), respectively, for well-differentiated HCC, and was more common in well-differentiated HCC compared to higher grades of HCC (p ¼ 0.01). Conversely, hypointensity on the hepatocellular phase had a sensitivity, specificity, positive predictive value, and negative predictive value of 80% (95% CI: 66.3e90.0%), 52.2% (95% CI: 30.6e73.2%), 78.4% (95% CI: 64.7e88.7%), and 54.6% (95% CI: 32.2e75.6%), respectively, for higher grades of HCC. Hypointensity on the hepatocellular phase was significantly more common with higher grades of HCC compared to well-differentiated HCC (p ¼ 0.01; Fig 2). Lesions that were predominantly hypointense or isointense in the arterial phase and hypointense in the

Figure 2 A 64-year-old man with a 1.2 cm moderately-differentiated HCC. (a) Axial arterial phase image showed barely visible hyperintense lesion (arrowhead). On the venous phase (not shown), the lesion was isointense. (b) On the hepatocellular phase, the lesion was hypointense. Hypointensity on this phase was more likely to indicate a higher grade HCC.

hepatocellular phase (n ¼ 15) were moderatelydifferentiated (n ¼ 9) or poorly-differentiated (n ¼ 6) HCC (Fig 3). Four lesions that were difficult to see (minimally hyperintense or isointense) on arterial and venous phases were clearly hypointense on the hepatocellular phase (Figs 2 and 4). These lesions were of poorly-differentiated (n ¼ 2) or moderately-differentiated (n ¼ 2) histology.

Quantitative analysis The intraclass correlation coefficients of the quantitative measurements by the two reviewers varied. The correlation coefficient of the signal intensities of lesions before contrast was 0.85 (95% CI: 0.75e0.94), in the arterial phase was 0.44 (95% CI: 0.25e0.66) and in the hepatocellular phase was 0.56 (95% CI: 0.41e0.65). These correspond to moderate to excellent correlation.

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

849

Figure 4 A 61-year-old man with poorly-differentiated HCC. (a) On the axial arterial phase image, the lesion (arrowhead) was minimally hypointense and barely visible. On the venous phase (not shown), the lesion was isointense and remained difficult to visualize. (b) On the hepatocellular phase, the lesion (arrowhead) was clearly hypointense and is larger than was apparent on the earlier phases. Among contrast-enhanced phases, 5% (four of 73) of lesions in the present cohort were best appreciated in the hepatocellular phase.

Figure 3 A 59-year-old woman with poorly-differentiated HCC. (a) Axial arterial phase image showed predominately hypointense lesion (arrowhead). (b) On the venous phase, the lesion (arrowhead) remained hypointense with more internal enhancement. This is an atypical appearance for untreated HCC. (c) On the hepatocellular phase, the lesion was predominantly hypointense. Hypointensity on the arterial and hepatocellular phases was seen only in higher grades of HCC.

The results of univariate analysis of the quantitative parameters are given in Table 4. The only parameter that showed significant differences between the three HCC groups was the ER in the hepatocellular phase (p ¼ 0.005 and 0.001 for the two reviewers). The results of multivariate analysis

were similar to that of univariate analysis. On stepwise logistic regression, the only variables to significantly predict the HCC grade were the ER measurements in the hepatocellular phase by both reviewers (p ¼ 0.019 and 0.001). The other parameters were eliminated from the model. The ROC curves for the quantitative parameters are shown in Fig 5. Table 5 provides the area-under-curve (AUC) for these parameters. Apart from ER in the hepatocellular phase, the parameters had an AUC close to 0.5. In a HCC lesion, a hepatocellular phase ER value of more than 13% had a sensitivity of 47% and specificity of 88.6% in predicting a well-differentiated histology.

Discussion Gadobenate dimeglumine is a lipophilic gadolinium agent. Using the organic anionic transporter mechanism,

850

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

Table 4 Univariate analysis. Parameter

Reviewer

WD HCC

MD HCC

PD HCC

All

CNR arterial

1 2 1 2 1 2 1 2

10 12.2 3.6 0.7 43.9 33.6 57.6 49.2

10.3 13.8 2.9 0.9 46.5 38.3 38.7 21.6

2.3 2.8 0.3 0.6 24.7 35.4 27.1 6.2

8.4 10.1 2.5 0.6 41.0 36.3 41.9 26.6

CNR hepatocellular ER arterial ER hepatocellular

(5.6e14.3) (10.2e14.0) (0e7.4) (0e1.6) (18.0e69.7) (16.9e50.3) (39.9e75.3) (35.2e67.3)

(6.6e14.0) (12.4e15.2) (0.9e4.8) (0.4e1.5) (29.7e63.3) (19.9e56.7) (26.5e50.9) (11.5e31.8)

(0.5e4.1) (0e5.8) (0e0.9) (0.2e1.0) (11.9e37.5) (15.0e55.1) (8.7e45.5) (0e27.0)

Significance (6.2e10.7) (7.1e14.8) (1.1e4.0) (0.3e1.0) (30.0e46.4) (25.7e46.9) (33.0e50.7) (17.8e35.4)

0.43 0.74 0.65 0.88 0.33 0.14 0.005 0.001

Mean values of the parameters (95% confidence intervals) are given. The significance of the differences in mean values between the three groups of HCC was calculated using ANOVA, with TukeyeKramer correction. WD HCC, well-differentiated HCC; MD HCC, moderately-differentiated HCC; PD HCC, poorlydifferentiated HCC; CNR, absolute contrast-to-noise ratio; ER, enhancement ratio.

a small portion of the injected dose (3e5%) is excreted by hepatocytes into bile.27,31 Despite the relatively low hepatobiliary excretion compared to other agents, such as gadoxetate disodium, the high relaxivity of gadobenate results in intense enhancement of functioning hepatic tissue 1e3 h after intravenous injection.32 The hepatocellular phase of gadobenate has been reported to accurately differentiate lesions with functioning hepatocytes, such as focal nodular hyperplasia, from those without, such as hepatic adenoma.33 Previous studies have investigated the usefulness of MRI contrast agents in differentiating the histological grades of HCC. Reticuloendothelial agents, which are also no longer available in USA, have been reported to be aid the characterization of HCC grade.34,35 Studies using mangafodipir trisodium, which is a hepatocellular agent that is no longer marketed in USA, indicate that well-differentiated HCC have

Figure 5 ROC curve plotting sensitivity (y-axis) and 1especificity (xaxis) of quantitative parameters. The dark grey solid curve indicates the contrast-to-noise ratio (CNR) of the arterial phase. The black dashed curve indicates the CNR of the hepatocellular phase. The light grey dotted curve indicates the enhancement ratio (ER) of the arterial phase. The dotedash curve indicates the ER of the hepatocellular phase, which has a substantially larger AUC compared to the other parameters. Table 5 lists the AUC for these curves. The grey diagonal line represents AUC of 0.50.

higher signal intensity after contrast medium administration compared to poorly-differentiated HCC.36,37 A study concluded that the hepatocellular phase of gadoxetate disodium (Eovist, Bayer HealthCare, Wayne, NJ, USA) was not useful in differentiating well-differentiated HCC from the more aggressive grades.38 This study examined 25 HCC lesions, with only five having well-differentiated histology. To the authors’ knowledge, there are only two reports on the usefulness of the hepatocellular phase of gadobenate in distinguishing the histological grades of HCC.27,39 The first study was performed using a 1.5 T MRI system using twodimensional gradient-echo sequences.39 This study did not find a difference in the CNR between 19 welldifferentiated HCC and nine poorly-differentiated HCC. The second study was performed on a 0.5 T magnet using two-dimensional T1-weighted spin-echo and gradientecho sequences.27 Seven patients had histological grading of HCC, and there was a significantly higher ER of welldifferentiated HCC compared to moderately- or poorlydifferentiated HCC on the hepatocellular phase. The results of the present study are mostly in agreement with these findings; it was found that the enhancement in the hepatocellular phase was of some use in distinguishing well-differentiated HCC from higher grades of HCC on qualitative and quantitative analysis. The hyperintensity or isointensity on this phase had a specificity of 76% for welldifferentiated HCC. Combining the appearance on the dynamic and hepatocellular phases helps to characterize some lesions. For instance, all lesions that were hypointense or isointense on the arterial phase and hypointense on the hepatocellular phase (n ¼ 15) were of higher histological grades. Occasionally, the hepatocellular phase was useful to detect a lesion that was difficult to visualize on the dynamic phases (four of 73; 5%). On quantitative analysis, there was Table 5 Receiver operating characteristic (ROC) curve analysis. Parameter

AUC (95% CI)

CNR in arterial phase CNR in hepatocellular phase ER in arterial phase ER in hepatocellular phase

0.50 0.52 0.51 0.70

(0.42e0.58) (0.44e0.61) (0.43e0.60) (0.62e0.77)

AUC, area-under-curve; 95% CI, 95% confidence interval of AUC; CNR, absolute contrast-to-noise ratio; ER, enhancement ratio.

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

a significant difference in ER in the hepatocellular phase between well-differentiated HCC from other histological types. An ER of more than 13% was highly specific (89%) for well-differentiated HCC. These findings may indicate that the enhancement ratio in the hepatocellular phase may be considered to be a marker for the percentage of tumour containing functioning hepatocytes. It should be noted that many factors including vascularity, intratumoral fibrosis, fatty metaplasia, peliosis, and necrosis may affect the initial dynamic enhancement of HCC lesions.40 These factors are less likely to be important on the hepatocellular phase. CNR was not significantly predictive of histological grade. CNR is determined not only by the enhancement of the tumour, but also that of the surrounding liver, which may confound the results. The degree of parenchymal enhancement is dependent on the overall hepatic functional status.40 The results of the present study indicate that the findings of the hepatocellular phase may, in the future, help in management decisions. The treatment of HCC may be customized not only using anatomical information, such as the number and size of lesions, but also functional factors. A lesion that is hyperintense in the hepatocellular phase may be expected to have better-than-average prognosis and be considered for more aggressive therapy, such as liver transplantation, even if the lesion size is greater than 5 cm (as per the Milan criteria) or 6.5 cm (as per the University of California, San Francisco, criteria). There are some limitations of the present study including its retrospective nature. Although histopathology was used as the reference standard, the interobserver variation of pathological grading and heterogeneity of tumour grades may reduce the accuracy of this reference standard. The number of poorly-differentiated HCC was small. In conclusion, subjective analysis of the hepatocellular phase suggests that lesions that are hyperintense or isointense tend to be well-differentiated HCC (specificity of 76%). However, there was overlap in the visual appearances of the different grades of HCC on the hepatocellular phase. Lesions that were hypointense or isointense on the arterial phase and hypointense on the hepatocellular phase were all of higher histological grades. The enhancement ratio in the hepatocellular phase was the most useful quantitative parameter, with a ratio of more than 13% having a specificity of 89% for well-differentiated HCC.

References 1. Marrero JA, Hussain HK, Nghiem HV, et al. Improving the prediction of hepatocellular carcinoma in cirrhotic patients with an arteriallyenhancing liver mass. Liver Transpl 2005;11:281e9. 2. Monzawa S, Ichikawa T, Nakajima H, et al. Dynamic CT for detecting small hepatocellular carcinoma: usefulness of delayed phase imaging. AJR Am J Roentgenol 2007;188:147e53. 3. Willatt JM, Hussain HK, Adusumilli S, et al. MR imaging of hepatocellular carcinoma in the cirrhotic liver: challenges and controversies. Radiology 2008;247:311e30. 4. Lutz AM, Willmann JK, Goepfert K, et al. Hepatocellular carcinoma in cirrhosis: enhancement patterns at dynamic gadolinium- and superparamagnetic iron oxide-enhanced T1-weighted MR imaging. Radiology 2005;237:520e8.

851

5. Murakami T, Kim T, Takamura M, et al. Hypervascular hepatocellular carcinoma: detection with double arterial phase multi-detector row helical CT. Radiology 2001;218:763e7. 6. Kim BK, Han KH, Park YN, et al. Prediction of microvascular invasion before curative resection of hepatocellular carcinoma. J Surg Oncol 2008;97:246e52. 7. Zhou L, Rui JA, Ye DX, et al. EdmondsoneSteiner grading increases the predictive efficiency of TNM staging for long-term survival of patients with hepatocellular carcinoma after curative resection. World J Surg 2008;32:1748e56. 8. Pawlik TM, Delman KA, Vauthey JN, et al. Tumor size predicts vascular invasion and histologic grade: implications for selection of surgical treatment for hepatocellular carcinoma. Liver Transpl 2005;11:1086e92. 9. Ariizumi S, Takasaki K, Yamamoto M, et al. Histopathologic differentiation of the main nodule determines outcome after hepatic resection for synchronous multicentric hepatocellular carcinomas. Hepatogastroenterology 2004;51:500e4. 10. Nathan H, Schulick RD, Choti MA, et al. Predictors of survival after resection of early hepatocellular carcinoma. Ann Surg 2009;249:799e805. 11. Zhou L, Rui JA, Wang SB, et al. Factors predictive for long-term survival of male patients with hepatocellular carcinoma after curative resection. J Surg Oncol 2007;95:298e303. 12. Parfitt JR, Marotta P, Alghamdi M, et al. Recurrent hepatocellular carcinoma after transplantation: use of a pathological score on explanted livers to predict recurrence. Liver Transpl 2007;13:543e51. 13. Poon RT, Fan ST, Lo CM, et al. Difference in tumor invasiveness in cirrhotic patients with hepatocellular carcinoma fulfilling the Milan criteria treated by resection and transplantation: impact on long-term survival. Ann Surg 2007;245:51e8. 14. Zimmerman MA, Trotter JF, Wachs M, et al. Predictors of long-term outcome following liver transplantation for hepatocellular carcinoma: a single-center experience. Transpl Int 2007;20:747e53. 15. Dudek K, Kornasiewicz O, Remiszewski P, et al. Impact of tumor characteristic on the outcome of liver transplantation in patients with hepatocellular carcinoma. Transplant Proc 2009;41:3135e7. 16. Schwarz RE, Smith DD. Trends in local therapy for hepatocellular carcinoma and survival outcomes in the US population. Am J Surg 2008;195:829e36. 17. Kim SH, Lim HK, Choi D, et al. Percutaneous radiofrequency ablation of hepatocellular carcinoma: effect of histologic grade on therapeutic results. AJR Am J Roentgenol 2006;186:S327e33. 18. Cillo U, Vitale A, Bassanello M, et al. Liver transplantation for the treatment of moderately or well-differentiated hepatocellular carcinoma. Ann Surg 2004;239:150e9. 19. Cillo U, Vitale A, Grigoletto F, et al. Intention-to-treat analysis of liver transplantation in selected, aggressively treated HCC patients exceeding the Milan criteria. Am J Transplant 2007;7:972e81. 20. Schwartz ME, D’Amico F, Vitale A, et al. Liver transplantation for hepatocellular carcinoma: are the Milan criteria still valid? Eur J Surg Oncol 2008;34:256e62. 21. Jonas S, Bechstein WO, Steinmuller T, et al. Vascular invasion and histopathologic grading determine outcome after liver transplantation for hepatocellular carcinoma in cirrhosis. Hepatology 2001;33:1080e6. 22. Suh KS, Cho EH, Lee HW, et al. Liver transplantation for hepatocellular carcinoma in patients who do not meet the Milan criteria. Dig Dis 2007;25:329e33. 23. Yang SH, Suh KS, Lee HW, et al. The role of (18)F-FDG-PET imaging for the selection of liver transplantation candidates among hepatocellular carcinoma patients. Liver Transpl 2006;12:1655e60. 24. Kim JI, Lee JM, Choi JY, et al. The value of gadobenate dimeglumineenhanced delayed phase MR imaging for characterization of hepatocellular nodules in the cirrhotic liver. Invest Radiol 2008;43:202e10. 25. Kim YK, Lee JM, Kim CS. Gadobenate dimeglumine-enhanced liver MR imaging: value of dynamic and delayed imaging for the characterization and detection of focal liver lesions. Eur Radiol 2004;14:5e13. 26. Kuwatsuru R, Kadoya M, Ohtomo K, et al. Comparison of gadobenate dimeglumine with gadopentetate dimeglumine for magnetic resonance imaging of liver tumors. Invest Radiol 2001;36:632e41. 27. Manfredi R, Maresca G, Baron RL, et al. Delayed MR imaging of hepatocellular carcinoma enhanced by gadobenate dimeglumine (GdBOPTA). J Magn Reson Imaging 1999;9:704e10.

852

B. Tahir et al. / Clinical Radiology 66 (2011) 845e852

28. Ishak KG, Anthony PP, Solin LH. Histological typing of tumors of the liver. 2nd ed. Berlin, Germany: Springer; 1994; P. 20. 29. Edmodson HA, Steiner PE. Primary carcinoma of the liver: a study of 100 cases among 48,900 necropsies. Cancer 1954;7:462e503. 30. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:420e8. 31. Spinazzi A, Lorusso V, Pirovano G, et al. Safety, tolerance, biodistribution, and MR imaging enhancement of the liver with gadobenate dimeglumine: results of clinical pharmacologic and pilot imaging studies in nonpatient and patient volunteers. Acad Radiol 1999;6:282e91. 32. Caudana R, Morana G, Pirovano GP, et al. Focal malignant hepatic lesions: MR imaging enhanced with gadolinium benzyloxypropionictetra-acetate (BOPTA)epreliminary results of phase II clinical application. Radiology 1996;199:513e20. 33. Grazioli L, Morana G, Kirchin MA, et al. Accurate differentiation of focal nodular hyperplasia from hepatic adenoma at gadobenate dimeglumineenhanced MR imaging: prospective study. Radiology 2005;236:166e77. 34. Imai Y, Murakami T, Yoshida S, et al. Superparamagnetic iron oxideenhanced magnetic resonance images of hepatocellular carcinoma: correlation with histological grading. Hepatology 2000;32:205e12.

35. Takeshita K, Nagashima I, Frui S, et al. Effect of superparamagnetic iron oxide-enhanced MRI of the liver with hepatocellular carcinoma and hyperplastic nodule. J Comput Assist Tomogr 2002;26:451e5. 36. Kim JH, Kim MJ, Park YN, et al. Mangafodipir trisodium-enhanced MRI of hepatocellular carcinoma: correlation with histological characteristics. Clin Radiol 2008;63:1195e204. 37. Murakami T, Baron RL, Peterson MS, et al. Hepatocellular carcinoma: MR imaging with mangafodipir trisodium (Mn-DPDP). Radiology 1996;200: 69e77. 38. Frericks BB, Loddenkemper C, Huppertz A, et al. Qualitative and quantitative evaluation of hepatocellular carcinoma and cirrhotic liver enhancement using Gd-EOB-DTPA. AJR Am J Roentgenol 2009;193: 1053e60. 39. 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;205: 721e8. 40. Grazioli L, Morana G, Caudana R, et al. Hepatocellular carcinoma: correlation between gadobenate dimeglumine-enhanced MRI and pathologic findings. Invest Radiol 2000;35:25e34.