JOURNAL OF MAGNETIC RESONANCE IMAGING 39:1238–1245 (2014)
Original Research
Infiltrative Hepatocellular Carcinoma on Gadoxetic Acid-Enhanced and Diffusion-Weighted MRI at 3.0T SangHyeok Lim, MD, Young Kon Kim, MD, PhD,* Hyun Jeong Park, MD, PhD, Won Jae Lee, MD, PhD, Dongil Choi, MD, PhD, and Min Jung Park, MD, PhD Purpose: To determine imaging features of infiltrative hepatocellular carcinoma (HCC) on 3T magnetic resonance imaging (MRI) including gadoxetic acid-enhanced and diffusion-weighted imaging (DWI). Materials and Methods: Eighteen patients with infiltrative HCC underwent liver MRI that consisted of T1- and T2-weighted image (T2WI), gadoxetic acid-enhanced arterial, portal, 3-min late and 20-min hepatobiliary phase (HBP), and DWI. Two reviewers evaluated in consensus tumor characteristics and lesion conspicuity using a 4point scale. The tumor-to-liver contrast ratio was also measured. Results: Most of the tumors (n ¼ 16, 88.9%) were seen as irregular permeative masses (4.0–23.0 cm, mean 10.5 cm in diameter) and the remaining two as poorly defined amorphous infiltration among thrombosed portal veins. Internal reticulation within the tumor was characteristic and was most frequently observed on 3-min late phase (n ¼ 18), followed by HBP (n ¼ 15). Tumor conspicuity and tumor-to-liver contrast ratio was highest with b-800 DWI, which was significantly higher than those of other images (P < 0.05). Conclusion: DWI provides the highest conspicuity for infiltrative HCC compared to unenhanced T1- and T2WI and gadoxetic acid-enhanced MRI. The gadoxetic acidenhanced 3-min late image is useful in characterizing infiltrative HCC, as it clearly depicts internal reticulation in all tumors. Key Words: liver; magnetic resonance imaging; diffusion weighted imaging; Gd-EOB-DTPA; infiltrative hepatocellular carcinoma J. Magn. Reson. Imaging 2014;39:1238–1245. C 2013 Wiley Periodicals, Inc. V
Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea. *Address reprint requests to: Y.K.K., Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea. E-mail: [email protected] Received February 9, 2013; Accepted May 16, 2013. DOI 10.1002/jmri.24265 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
HEPATOCELLULAR CARCINOMA (HCC) is the most common primary liver cancer worldwide and is the third leading cause of cancer-related death (1). The incidence of HCC has markedly increased in recent decades, especially in Western countries (2). Morphologically, HCC can be classified as focal/nodular, massive, and diffuse/infiltrative (3–5). The majority comprise the single or multiple nodular type with complete or incomplete capsule. The infiltrative type accounts for 7% to 13% of HCCs (5,6). With active implementation of a surveillance program for highrisk HCC patients, HCC is being detected at an earlier stage and most HCCs encountered in the clinic setting are the small nodular type. Thus, most research on the imaging characteristics of HCC have dealt with the nodular type of HCC. Noninvasive imaging criteria proposed by study groups, such as the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD) allow diagnosis of nodular HCCs based on their vascular profile-intense arterial hyperenhancement, followed by early washout (7,8). Infiltrative HCC is not infrequently encountered in clinics, particularly in areas endemic for hepatitis B virus (9,10). It has peculiar morphological and clinical characteristics that differ from those of nodular HCC (9–11). Infiltrative HCC has a diffuse, permeative appearance, lacking a well-demarcated boundary on cross-sectional imaging, which is related to its poor prognosis due to difficulty in early detection and frequent portal vein invasion (6,11). Its hypovascular nature makes it difficult to distinguish from massforming cholangiocarcinoma (12,13). Despite its clinical importance, the imaging features of infiltrative HCC are poorly documented (6,12,14). Gadoxetic acid and diffusion-weighted imaging (DWI) has opened new horizons for liver magnetic resonance imaging (MRI) with promising results for liver lesion detection and characterization (15–18). Gadoxetic acid is a recently introduced dual-acting MR contrast agent and has become increasingly used for HCC workup due to its characteristics of an extracellular space agent (ECS) during the early vascularinterstitial phases and that of a liver-specific agent
1238
Infiltrative HCC on 3T MRI
during the delayed phase (15,16). As DWI is easy to perform and requires no contrast agent, it is routinely incorporated into liver MRI protocols. Several clinical trials have revealed additional benefits of DWI over conventional MRI or gadoxetic acid-enhanced MRI for diagnosing HCC, particularly HCC detection (16,19,20). To date, no study has documented imaging features of infiltrative HCC on gadoxetic acidenhanced MRI and DWI using 3.0T. Accordingly, we conducted this study to investigate the imaging features of infiltrative HCC on gadoxetic acid-enhanced MRI and DWI with a special emphasis on tumor conspicuity.
MATERIALS AND METHODS Patients This retrospective study was approved by the relevant Institutional Review Board and informed consent was waived. We retrospectively searched our hospital’s surgical database between October 2010 and August 2012 using the search term “infiltrative or diffuse hepatocellular carcinoma.” This search identified 125 consecutive patients who had undergone crosssectional liver imaging. The inclusion criteria were: 1) patients with pathologically proven infiltrative HCC; 2) patients who underwent gadoxetic acid-enhanced liver MRI and DWI according to our institutions’ routine protocol; and 3) patients who had not undergone treatment such as transarterial chemoembolization or local ablation treatment prior to MRI. Infiltrative HCC was defined as tumors showing an infiltrative growth without definitive evidence of expansive tumor growth pattern that is seen in nodular HCC on any portion of the tumor. Thus, we excluded patients with multinodular HCCs or nodular HCCs with a partially infiltrative growth pattern and patients with combined HCCICC. In all, 107 patients were excluded from our study for the following reasons: 106 patients had no histologic proof and one had a history of HCC treatment prior to MRI examination. Accordingly, we identified 18 patients (15 men and 3 women; age range, 35–66 years) with infiltrative HCCs fitting the inclusion criteria. The confirmation of tumors was based on surgical resection for three patients and on percutaneous biopsy for 15 patients. We reviewed the medical records to ensure that no patients included in the study had a possibility of hepatic tumors other than HCC. All patients were followed by computed tomography (CT) or MRI for more than 6 months after the index MRI. MR Examination All MR images were acquired using an Intera Achieva 3.0T whole-body MR system (Philips Healthcare, Best, The Netherlands) equipped with a dual-source parallel radiofrequency transmission system and a quadrature body coil. The baseline MR images included a T1weighted turbo field-echo in-phase and opposed sequence (TR/first echo TE, second echo TE, 10/2.3 [in-phase], 3.45 [opposed-phase]; flip angle, 15 ;
1239
matrix size, 256 194; bandwidth, 434.3 Hz/pixel), a breath-hold multishot T2-weighted sequence with an acceleration factor of 2 (1796/70; flip angle, 90 ; matrix size, 324 235; bandwidth, 258.4 Hz/pixel), a respiratory-triggered single-shot heavily T2-weighted sequence with an acceleration factor of 2 (1802/160; flip angle, 90 ; matrix size, 252 254; bandwidth, 420.9 Hz/pixel) with a 5 mm section thickness, and a field of view of 32–38 cm. For gadoxetic acid-enhanced imaging, unenhanced, arterial-phase (AP, 20–35 s), portal-phase (PVP, 60 s), 3-min late-phase, and 20-min hepatobiliary images (HBP) were obtained using a T1-weighted 3D turbofield-echo sequence (enhanced T1 high-resolution isotropic volume examination; eTHRIVE, Philips Health care) (3.1/1.5; flip angle, 10 ; matrix size, 256 256; bandwidth, 724.1 Hz/ pixel) with a 2-mm section thickness and a field of view of 32–38 cm. The measured voxel size was 1.5 1.5 4.0 mm and the reconstructed voxel size was 1.17 1.17 2.0 mm. The contrast agent was automatically administered intravenously at a rate of 1 mL/s for a dose of 0.025 mmol/kg body weight using a power injector, followed by a 20-mL saline flush. Diffusion-weighted single-shot echo planar imaging with the simultaneous use of respiratory triggering was performed using a TR/TE of 1,600 msec/70 msec; the TR was matched in each patient to the length of the respiratory cycle prior to the gadoxetic acidenhancement. The scanning parameters were as follows: a b-value of 0, 100, and 800 s/mm2; spectral presaturation with inversion recovery for fat suppression; matrix size, 100 100; acceleration factor of SENSE, 2.0; field of view, 35 35 cm; number of excitations, 4; slice thickness, 5 mm; slice gap, 1 mm; and 33 axial slices. Depending on the respiratory efficiency of each patient, the acquisition time for this sequence ranged from 3–4 minutes. The apparent diffusion coefficient (ADC) was calculated by a monoexponential function using b-values of 100 and 800 s/mm2. Image Analysis All MRIs were evaluated by the consensus of two gastrointestinal radiologists (S.L. and Y.K.K.) with 5 and 13 years of experience in liver MRI, respectively). All images were evaluated using a Picture Archiving and Communication System (PACS; Pathspeed, GE Medical Systems Integrated Imaging Solutions, Mt. Prospect, IL) with an optimal window setting adjustment in each case. Although there were minimal discrepancies in assigning lesion conspicuity between two readers, common consent was reached easily by consensus reading during an additional reading session. The MRI features evaluated were: 1) shape of tumors; 2) signal intensity the tumors relative to liver parenchyma on the precontrast T1-weighted images (T1WI), T2-weighted images (T2WI), gadoxetic acidenhanced images (AP, PVP, 3-min late, and HBP), and b-800 DWI; 3) presence of internal reticulation on tumors; and 4) presence of portal vein thrombosis. In addition, since most infiltrative HCC tends to accompany portal vein thrombosis with resultant arterial
1240
Lim et al.
Table 1 Clinical Characteristics of Patients With Infiltrative Hepatocellular Carcinoma Variables
Values
Age Sex Hepatitis B C AFP (ng/mL) PIVKA-II (mAU/mL) Child-Pugh Class (A/B/C) Treatment TACE TACE þ radiotherapy Systemic therapy (Sorafenib) Surgery None
Mean 57 (range; 35–66) 15 : 3 15 3 10.5–14500 592–64000 15/ 3/ 0 8 4 2 3 1
AFP: a-fetoprotein, PIVKA-II: protein induced by vitamin K absence II, TACE: transarterial chemoembolization.
hyperperfusion, lesion conspicuity with an emphasis on clear tumor margin at each of seven different images was assessed based on the following four grading scales. Grade 4 was defined as clear tumor margin showing more than of 90% of tumor circumference, Grade 3 as 50% < and 90%, Grade 2 as 10% < and 50%, and Grade 1 as 10%. Tumor extent was determined by the consent of two reviewers based on correlation between initial MRI and follow-up CT with/without MRI, which showed lipiodol uptake after TACE or tumor necrosis by radiotherapy with tumor shrinkage or progression. In addition, in tumors proven by biopsy, the biopsy report (sonography) was used to define the tumor extent: the area that showed same signal intensity to biopsied hepatic segment was considered tumor. Quantitative image analysis was performed by measuring the liver signal intensity and tumor signal intensity, using operator-defined regions of interest (ROIs) for each image by an abdominal radiologist who did not participate in the qualitative image analysis. For measurements within the lesion, the ROI was positioned manually as much as possible to avoid the necrotic foci. For signal intensities of the liver, ROIs were drawn in the same location as each sequence devoid of a large intrahepatic vessel. The standard deviation (SD) of background noise was measured along the phaseencoding direction outside the body just ventral to the right anterior abdominal wall and included respiratory or motion-related artifacts. However, since the introduction of a parallel acquisition technique that could interfere with estimating a correct background noise level, there has been no perfect maneuver for noise estimation, and this measurement has been commonly used in extensive studies. By applying the copy and paste function of the workstation, the position and size of the ROI were kept constant for all images as far as possible. The ROIs included at least 300 pixels (range 300–600 pixels). The tumor-to-liver contrast ratio for each sequence was calculated from the signal intensity of the liver and tumor as j(liver signal intensity tumor signal intensity)j / (liver signal intensity þ tumor signal
intensity); this formula yields a value between 0 and 1, with a larger value indicating greater relative tumorliver contrast (14). Statistical Analysis Subjective image ratings were compared with the Wilcoxon test corrected for multiple comparisons according to the Bonferroni adjustment. Paired t-tests were used for comparison of the tumor-to-liver contrast ratio at each sequence. Statistical analyses were performed using the SPSS software package, v. 20.0 (Chicago, IL). For all tests, P < 0.05 was considered significant.
RESULTS Table 1 shows the clinical features of study populations included in the study. This study was conducted in a viral hepatitis endemic Asian population. Thus, all patients had liver cirrhosis associated with viral hepatitis B (n ¼ 17) or viral hepatitis C (n ¼ 1). Based on the Child-Pugh classification, 15 patients were Class A and the remaining three were Class B. All three HCCs with surgical confirmation corresponded to Grade 2 according to Edmonson’s classification of HCC. Three surgically confirmed cases exhibited internal fibrous septum on histology. The MRI features of tumors are summarized in Table 2. Based on combined reading of all MRI sequences, all tumors were classified into one of two morphologic types as follows. Most tumors (n ¼ 16, 88.9%) were evident as a permeative mass with
Table 2 Summary of MRI Findings of 18 Infiltrative Hepatocellular Carcinomas Imaging characteristics of tumor Morphology Permeative mass Amorphous infiltration with thrombosed portal vein Portal vein thrombosis Main trunk with lobar branch Segmental branch Microvessel invasion Satellite nodules Signal intensity T1-weighted imaging (hypointensity) T2-weighted imaging (hyperintensity)/ internal bright foci Diffusion-weighted imaging (hyperintensity) Arterial phase (hypo/ iso/ high) Internal reticulation 3 min late phase 20 min hepatobiliary phase Portal venous phase Diffusion-weighted imaging Arterial phase T2-weighted imaging T1-weighted imaging
Number of tumor n 5 16 n52
n 5 16 n51 n51 n 5 12 n 5 18 n 5 18/10 n 5 18 n 5 9/ 6/ 3 n 5 18 n 5 15 n 5 14 n 5 11 n 5 10 n59 n50
Infiltrative HCC on 3T MRI
1241
Figure 1. A 50-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma. Large permeative liver mass with irregular border (arrows) is seen on T2weighted image (a), b-800 diffusionweighted image (b), unenhanced (c) and gadoxetic acid-enhanced arterial (d), portal (e), 3-min late (f), and 20-min hepatobiliary phase (g). Tumor margin is most clearly demonstrated on diffusion-weighted image (b). Reticulation within tumor is mostly accentuated on 3-min late phase (f). Note portal vein thrombosis (arrow in d) that is not strongly enhanced.
irregular outer contour, ranging in diameter from 4.0– 23.0 cm (mean 10.5 cm, Fig. 1). The remaining two were evident as poorly defined amorphous infiltration among thrombosed portal vein branches (8.0 cm and 13.0 cm in diameter, Fig. 2). The main feature of these two tumors was portal vein thrombosis rather than parenchymal mass. All but two tumors (88.9%) exhibited tumor thrombosis in the portal vein including right or left main trunk. One lesion (4 cm in size) accompanied tumor thrombosis in the segmental branch of portal vein. In the remaining one lesion (5 cm in size), portal vein thrombosis was not demonstrated upon MRI, but microscopic vascular invasion was exhibited on histology. The arterial hyperenhancement on hemiliver or hepatic segment was noted
in all 17 tumors with portal vein thrombosis. Twelve patients (66.7%) had satellite nodules. The signal intensities of most tumors were predominantly hypointense on the pre- and postcontrast T1WI, and moderately hyperintense on T2WI (Fig. 1). On AP imaging, most of the tumors (n ¼ 15, 83.3%) were considered hypovascular as they were hypo- (n ¼ 9) or isointense (n ¼ 6) with inhomogeneous areas of hyperenhancement, mostly at the periphery. The remaining three were heterogeneously hyperintense due to internal neovascularization. On b-800 DWI, both tumor and portal vein thrombosis were seen as hyperintense relative to background liver (Fig. 2). The internal reticulation was most frequently and clearly observed on 3-min late phase (n ¼ 18), followed by
1242
Lim et al.
Figure 2. A 58-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma. Poorly defined infiltrating tumors among thrombosed portal vein branches (arrows) is seen in right hepatic lobe on T2-weighted image (a), b-800 diffusion-weighted image (b), and unenhanced (c) and gadoxetic acidenhanced arterial (d), portal (e), 3min late (f), and 20-min hepatobiliary phase (g) of 3D T1-weighted image. Although tumor margin is not clear defined, the conspicuity of tumor is better with diffusionweighted image (b) than with other images.
20-min HBP (n ¼ 15), PVP (n ¼ 14), DWI (n ¼ 11), AP (n ¼ 10), T2WI (n ¼ 9), and T1WI (n ¼ 0) (Figs. 1, 3). Besides internal reticulation, most tumors showed homogenous signal intensity without obvious internal necrosis (Fig. 1). However, 10 tumors showed internal multifocal tiny bright foci on T2WI, indicating small necrotic foci or peliosis on histology. Table 3 summarizes the subjective assessments of lesion conspicuity as well as the tumor-to-liver contrast ratio for each of the sequences. The mean value for the subjective assignment of lesion conspicuity was highest in DWI (3.83 6 0.51), followed by T2WI (3.50 6 0.86), 20-min HBP (3.44 6 0.86), T1WI (2.94 6 1.00), 3-min late phase and PVP (2.89 6 1.02), and AP (2.61 6 0.85, Fig. 3). The differences for the mean
value were significant between DWI and other sequences (P < 0.05). All 16 tumors seen as permeative mass were assigned a rating score 4 on DWI. The remaining two infiltrating tumors among thrombosed portal veins were assigned a rating score 3 or 2 on DWI. For all tumors, no MR sequence achieved a higher score than DWI. As for the tumor-liver contrast ratio, it was the highest in DWI with 0.59 6 0.16, which was significantly higher than those of all other sequences (P < 0.0001), followed by T2WI (0.34 6 0.09), 20-min HBP (0.25 60.06), 3-min late phase (0.15 6 0.10), T1WI (0.14 6 0.06), PVP (0.14 6 0.09), and AP (0.11 6 0.07). The differences for the mean value were also significant for pairs of T1WI and T2WI, T2WI and AP,
Infiltrative HCC on 3T MRI
1243
Figure 3. A 51-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma surgically confirmed. Small permeative liver mass with irregular border (arrows) is seen on T2-weighted image (a), b800 diffusion-weighted image (b), unenhanced (c) and gadoxetic acidenhanced arterial (d), portal (e), 3min late (f), and 20-min hepatobiliary phase (g) of 3D T1-weighted image. Tumor conspicuity is better with diffusion-weighted image (b) than with other images. Reticulation within tumor is mostly accentuated on 3-min late phase (f).
T2WI and PVP, T2WI and 3-min late, AP and HBP, PVP and HBP, 3-min late and HBP (P < 0.0001), and for between T2WI and HBP (P ¼ 0.014).
DISCUSSION In our study, the imaging features of most infiltrative HCCs agreed with previous reports (6,12,14), as they were observed as irregular-shaped hypovascular permeative masses with portal vein thrombosis. On AP imaging, all but three tumors (83.3%) were hypo- (n ¼ 9) or isointense (n ¼ 6) with inhomogeneous areas of hyperenhancement, mostly at the periphery. Thus, it is difficult to clearly differentiate tumoral enhancement and parenchymal arterial hyperperfusion
derived from portal vein thrombosis. With early dynamic phases gadoxetic acid-enhanced MRI, defining tumor boundary or detection as well as tumor characterization is challenging. In fact, the current noninvasive diagnostic criteria for HCC (7,8) have been refuted in most study cases. Despite a large tumor size, most tumors tended to show relatively homogenous hypo- and hyperintensity on T1- and T2WI, respectively, without obvious internal necrosis that is commonly seen in large nodular HCC. The most noticeable finding was that the majority of the tumors exhibited internal reticulation. This feature was most frequently demonstrated on 3-min late phase (n ¼ 18), followed by 20-min HBP (n ¼ 15). This is line with previous reports (12,21). However, its prevalence is higher than the 16.4% previously
1244
Lim et al.
Table 3 Results of Qualitative and Quantitative Image Analysis for Each MRI Sequence T1WI Rating for lesion conspicuity
1 2 3 4 Mean
P value Liver-tumor contrast ratio P value
T2WI
DWI
AP
PVP
3 min late
HBP
2 1 0 2 2 2 1 3 1 1 5 4 4 1 7 4 1 9 6 6 5 6 12 16 2 6 6 11 2.94 6 1.00 3.50 6 0.86 3.83 6 0.51 2.61 6 0.85 2.89 6 1.02 2.89 6 1.02 3.44 6 0.86 P < 0.0001 h, l, r, P ¼ 0.001 b, i, j, 0.002 a, m, n, t, u, 0.003f, 0.008o,0.014 c, g, 0.025 p,q, 0.317d, e, k, 1.000 s, 0.14 6 0.06 0.34 6 0.09 0.59 6 0.16 0.11 6 0.07 0.14 6 0.09 0.15 6 0.10 0.25 6 0.06 P < 0.0001 a,b, g, h, i, j, l, m, n, o, r, t, u, P ¼ 0.199 c, 0.819 d, 0.676 e, 0.210 f, 0.014 k, 0.306 p, 0.177 q, 0.506 s
Values are mean 6 SD. P values are T1WI and T2WI,a T1WI and DWI,b T1WI and AP,c T1WI and PVP,d T1WI and 3 min late,e T1WI and HBP,f T2WI and DWI,g T2WI and AP,h T2WI and PVP,i T2WI and 3 min late,j T2WI and HBP,k DWI and AP,l DWI and PVP,m DWI and 3 min late,n DWI and HBP,o AP and PVP,p AP and 3 min late,q AP and HBP,r PVP and 3 min,s PVP and HBP,t 3 min late and HBP.u
reported (21). The presence of a fibrous capsule or internal fibrous septum causing a mosaic appearance is a characteristic finding of HCC (22,23). Thus, we postulate that fibrosis could be responsible for internal reticulation, which was partially supported by three surgically proven cases. Abundant fibrous stroma in mass-forming cholangiocarcinoma is reportedly manifest as a wide area of central hyperenhancement on delayed phase gadoxetic acid-enhanced MRI (13). Considering the peculiar characteristic of gadoxetic acid including having relatively weak, short activity as an ECS agent and early, strong activity as a hepatocyte agent (24), the use of the 3-min late phase is prudent to depict fibrotic septum, as thin fibrosis could be strongly enhanced at this stage by an ECS agent characteristic of gadoxetic acid because it needs no more delayed time as in the broad fibrotic area of cholangiocarcinoma. Meanwhile, background hypointense tumor tissues are also accentuated by homogeneously enhanced adjacent liver that already accumulates gadoxetic acid into hepatocytes. Thus, the reticular appearance in gadoxetic acid-enhanced MRI could be a reliable feature for diagnosing infiltrative HCC and 3-min late phase image is mandatory for characterizing infiltrating HCC. In our study, the mean value for lesion conspicuity was highest with DWI, followed by T2WI and HBP. This was supported by quantitative results showing better tumor-to-liver contrast ratio with DWI and T2WI or HBP than with other sequences. This result agrees well with a recent report (14), in which T2WI and DWI exhibited greater tumor-liver contrast than an early conventional gadolinium-enhanced image. The greatest merit of gadoxetic acid is the clear depiction of HCC as hypointense on HBP (15,16). However, since most tumors present with portal vein thrombosis, peritumoral liver parenchyma could show decreased enhancement on HBP, causing decreased tumor-liver contrast (Fig. 1g). In particular, in two cases depicted as amorphous infiltration among thrombosed portal veins (Fig. 2g), it was difficult to differentiate tumor infiltration and liver parenchyma with decreased hepatic function on HBP. Nevertheless, in DWI all but one tumor were scored as 4 or 3. High tumor conspicuity on DWI could be explained by the recent advances in image quality due to improved MRI gradient performance, multichannel surface
receiver coils, parallel imaging techniques, improved fat suppression schemes, and application of a dualsource parallel radiofrequency transmission system (18). Based on DWI, HBP, and T2WI, infiltrative HCCs can be classified into two morphologic types. It is well known that infiltrative HCC can fail to have a discrete mass on imaging because this tumor blends into the background of the cirrhotic liver (11). However, 16 tumors (88.9%) displayed a discrete permeative mass with an irregular border, which is higher than the rate of 57.3% in a previous report (21). Meanwhile, as for two tumors seen as amorphous infiltration among thrombosed portal veins, portal vein thrombosis was their primary imaging feature rather than parenchymal mass. Given that portal vein thrombosis from infiltrative HCC might not be as highly enhanced as typical hypervascular HCC (Figs. 1, 2), caution needs to be taken in diagnosing this pattern of infiltrative HCC because even with DWI one of two tumors was assigned as poor conspicuity (Grade 2). In consideration of growing concern for the association between gadolinium-based contrast agents and nephrogenic systemic fibrosis (25), our results offer meaningful data because infiltrative HCC might not be readily detectable in other imaging modalities, while DWI, T1WI, and T2WI provide high tumor conspicuity. Furthermore, the sensitivity of DWI for detecting small nodular HCC is reportedly comparable to gadoxetic acid-enhanced MRI (16). In that sense, our observations could indicate that unenhanced MRI including DWI deserves to be introduced into HCC surveillance program. Further studies are needed. A major limitation of this study was that most tumors were not surgically confirmed, so the correlation between imaging and histological composition was not made. In particular, although this study focused on tumor conspicuity based on clear tumor margin at different MR sequences, a reference standard was not enough to determine exact tumor boundary. In addition, biopsied specimen could not represent whole tumor histology. Second, it was performed on a 3.0T system. Thus, our results might not necessarily be transferable to a 1.5T system. Third, due to the retrospective study design, we did not assess image acquired at variable delay timing after contrast administration. Fourth, most tumors were
Infiltrative HCC on 3T MRI
large, so whether these results might be exportable to small infiltrative HCC cannot be determined. Fifth, our study was also potentially limited by consensus review because we did not assess interobserver variability. Finally, this study included only infiltrative HCCs without control patients. Further investigation with hypovascular other hepatic tumors is needed. In conclusion, infiltrative HCCs showed two morphologic types: discrete permeative mass and indistinct infiltrations among thrombosed portal vein branches. DWI provided the highest tumor conspicuity compared to unenhanced T1WI and T2WI, and gadoxetic acid-enhanced dynamic and HBP as most tumors with portal vein thrombosis were clearly depicted as hyperintense on high-b value DWI. The gadoxetic acid-enhanced 3-min late image was useful in characterizing infiltrative HCC, as it clearly depicted internal reticulation in all tumors. POTENTIAL CONFLICT OF INTEREST We certify that all authors have had no relevant financial interests or personal affiliations in connection with the content of this article. REFERENCES 1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108. 2. El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999;340:745–750. 3. Baer HU, Gertsch P, Matthews JB, et al. Resectability of large focal liver lesions. Br J Surg 1989;76:1042–1044. 4. Okuda K, Noguchi T, Kubo Y, et al. A clinical and pathological study of diffuse type hepatocellular carcinoma. Liver 1981;1:280– 289. 5. Trevisani F, Caraceni P, Bernardi M, et al. Gross pathologic types of hepatocellular carcinoma in Italian patients. Relationship with demographic, environmental, and clinical factors. Cancer 1993; 72:1557–1563. 6. Kanematsu M, Semelka RC, Leonardou P, et al. Hepatocellular carcinoma of diffuse type: MR imaging findings and clinical manifestations. J Magn Reson Imaging 2003;18:189–195. 7. Bruix J, Sherman A, American Association for the Study of Liver Disease. Management of hepatocellular carcinoma: an update. Hepatology 2011;53:1020–1021. 8. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. J Hepatol 2001;35:421–430. 9. Myung SJ, Yoo JH, Kim KM, et al. Diffuse infiltrative hepatocellular carcinomas in a hepatitis B-endemic area: diagnostic and therapeutic impediments. Hepatogastroenterology 2006;53:266– 270.
1245 10. Benvegnu L, Noventa F, Bernardinello E, et al. Evidence for an association between the aetiology of cirrhosis and pattern of hepatocellular carcinoma development. Gut 2001;48:110–115. 11. Demirjian A, Peng P, Geschwind JF, et al. Infiltrating hepatocellular carcinoma: seeing the tree through the forest. J Gastrointest Surg 2011;15:2089–2097. 12. Kim YK, Han YM, Kim CS. Comparison of diffusion hepatocellular carcinoma and intrahepatic cholangiocarcinoma using sequentially acquired gadolinium-enhanced and Resovist-enhanced MRI. Eur J Radiol 2009;70:94–100. 13. Chong YS, Kim YK, Lee MW, et al. Differentiating mass-forming intrahepatic cholangiocarcinoma from atypical hepatocellular carcinoma using gadoxetic acid-enhanced MRI. Clin Radiol 2012; 67:766–773. 14. Rosenkrantz AB, Lee L, Matza BW, Kim S. Infiltrative hepatocellular carcinoma: comparison of MRI sequences for lesion conspicuity. Clin Radiol 2012;67:e105–111. 15. Haradome H, Grazioli L, Tinti R, et al. Additional value of gadoxetic acid-DTPA-enhanced hepatobiliary phase MR imaging in the diagnosis of early-stage hepatocellular carcinoma: comparison with dynamic triple-phase multidetector CT imaging. J Magn Reson Imaging 2011;34:69–78. 16. Park MJ, Kim YK, Lee WJ, et al. Small hepatocellular carcinomas: improved sensitivity by combining gadoxetic acid-enhanced and diffusion-weighted MR imaging patterns. Radiology 2012;264: 761–770. 17. Parikh T, Drew SJ, Lee VS, et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008; 246:812–822. 18. Taouli B, Koh DM. Diffusion-weighted MR imaging of the liver. Radiology 2010;254:47–66. 19. Piana G, Trinquart L, Meskine N, Barrau V, Beers BV, Vilgrain V. New MR imaging criteria with a diffusion-weighted sequence for the diagnosis of hepatocellular carcinoma in chronic liver diseases. J Hepatol 2011;55:126–132. 20. Holzapfel K, Eiber MJ, Fingerie AA, et al. Detection, classification, and characterization of focal liver lesions: value of diffusionweighted MR imaging, gadoxetic acid-enhanced MR imaging and the combination of both methods. Abdom Imaging 2012;37:74– 82. 21. Kneuertz PJ, Demirjian A, Firoozmand A, et al. Diffuse infiltrative hepatocellular carcinoma: assessment of presentation, treatment, and outcomes. Ann Surg Oncol 2012;19:2897–2907. 22. Stevens WR, Gulino SP, Batts KP, Stephens DH, Johnson CD. Mosaic pattern of hepatocellular carcinoma: histologic basis for a characteristic CT appearance. J Comput Assist Tomogr 1996;20: 337–342. 23. Ishiazaki M, Ashida K, Higashi T, et al. The formation of capsule and septum in human hepatocellular carcinoma. Virchows Arch 2001;438:574–580. 24. Tamada T, Ito K, Sone T, et al. Dynamic contrast-enhanced magnetic resonance imaging of abdominal solid organ and major vessel: comparison of enhancement effect between Gd-EOB-DTPA and Gd-DTPA. J Magn Reson Imaging. 2009;29:636–640. 25. Sadowski EA, Bennett LK, Chan MR, et al. Nephrogenic systemic fibrosis: risk factors and incidence estimation. Radiology 2007; 243:148–157.
Original Research
Infiltrative Hepatocellular Carcinoma on Gadoxetic Acid-Enhanced and Diffusion-Weighted MRI at 3.0T SangHyeok Lim, MD, Young Kon Kim, MD, PhD,* Hyun Jeong Park, MD, PhD, Won Jae Lee, MD, PhD, Dongil Choi, MD, PhD, and Min Jung Park, MD, PhD Purpose: To determine imaging features of infiltrative hepatocellular carcinoma (HCC) on 3T magnetic resonance imaging (MRI) including gadoxetic acid-enhanced and diffusion-weighted imaging (DWI). Materials and Methods: Eighteen patients with infiltrative HCC underwent liver MRI that consisted of T1- and T2-weighted image (T2WI), gadoxetic acid-enhanced arterial, portal, 3-min late and 20-min hepatobiliary phase (HBP), and DWI. Two reviewers evaluated in consensus tumor characteristics and lesion conspicuity using a 4point scale. The tumor-to-liver contrast ratio was also measured. Results: Most of the tumors (n ¼ 16, 88.9%) were seen as irregular permeative masses (4.0–23.0 cm, mean 10.5 cm in diameter) and the remaining two as poorly defined amorphous infiltration among thrombosed portal veins. Internal reticulation within the tumor was characteristic and was most frequently observed on 3-min late phase (n ¼ 18), followed by HBP (n ¼ 15). Tumor conspicuity and tumor-to-liver contrast ratio was highest with b-800 DWI, which was significantly higher than those of other images (P < 0.05). Conclusion: DWI provides the highest conspicuity for infiltrative HCC compared to unenhanced T1- and T2WI and gadoxetic acid-enhanced MRI. The gadoxetic acidenhanced 3-min late image is useful in characterizing infiltrative HCC, as it clearly depicts internal reticulation in all tumors. Key Words: liver; magnetic resonance imaging; diffusion weighted imaging; Gd-EOB-DTPA; infiltrative hepatocellular carcinoma J. Magn. Reson. Imaging 2014;39:1238–1245. C 2013 Wiley Periodicals, Inc. V
Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea. *Address reprint requests to: Y.K.K., Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea. E-mail: [email protected] Received February 9, 2013; Accepted May 16, 2013. DOI 10.1002/jmri.24265 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
HEPATOCELLULAR CARCINOMA (HCC) is the most common primary liver cancer worldwide and is the third leading cause of cancer-related death (1). The incidence of HCC has markedly increased in recent decades, especially in Western countries (2). Morphologically, HCC can be classified as focal/nodular, massive, and diffuse/infiltrative (3–5). The majority comprise the single or multiple nodular type with complete or incomplete capsule. The infiltrative type accounts for 7% to 13% of HCCs (5,6). With active implementation of a surveillance program for highrisk HCC patients, HCC is being detected at an earlier stage and most HCCs encountered in the clinic setting are the small nodular type. Thus, most research on the imaging characteristics of HCC have dealt with the nodular type of HCC. Noninvasive imaging criteria proposed by study groups, such as the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD) allow diagnosis of nodular HCCs based on their vascular profile-intense arterial hyperenhancement, followed by early washout (7,8). Infiltrative HCC is not infrequently encountered in clinics, particularly in areas endemic for hepatitis B virus (9,10). It has peculiar morphological and clinical characteristics that differ from those of nodular HCC (9–11). Infiltrative HCC has a diffuse, permeative appearance, lacking a well-demarcated boundary on cross-sectional imaging, which is related to its poor prognosis due to difficulty in early detection and frequent portal vein invasion (6,11). Its hypovascular nature makes it difficult to distinguish from massforming cholangiocarcinoma (12,13). Despite its clinical importance, the imaging features of infiltrative HCC are poorly documented (6,12,14). Gadoxetic acid and diffusion-weighted imaging (DWI) has opened new horizons for liver magnetic resonance imaging (MRI) with promising results for liver lesion detection and characterization (15–18). Gadoxetic acid is a recently introduced dual-acting MR contrast agent and has become increasingly used for HCC workup due to its characteristics of an extracellular space agent (ECS) during the early vascularinterstitial phases and that of a liver-specific agent
1238
Infiltrative HCC on 3T MRI
during the delayed phase (15,16). As DWI is easy to perform and requires no contrast agent, it is routinely incorporated into liver MRI protocols. Several clinical trials have revealed additional benefits of DWI over conventional MRI or gadoxetic acid-enhanced MRI for diagnosing HCC, particularly HCC detection (16,19,20). To date, no study has documented imaging features of infiltrative HCC on gadoxetic acidenhanced MRI and DWI using 3.0T. Accordingly, we conducted this study to investigate the imaging features of infiltrative HCC on gadoxetic acid-enhanced MRI and DWI with a special emphasis on tumor conspicuity.
MATERIALS AND METHODS Patients This retrospective study was approved by the relevant Institutional Review Board and informed consent was waived. We retrospectively searched our hospital’s surgical database between October 2010 and August 2012 using the search term “infiltrative or diffuse hepatocellular carcinoma.” This search identified 125 consecutive patients who had undergone crosssectional liver imaging. The inclusion criteria were: 1) patients with pathologically proven infiltrative HCC; 2) patients who underwent gadoxetic acid-enhanced liver MRI and DWI according to our institutions’ routine protocol; and 3) patients who had not undergone treatment such as transarterial chemoembolization or local ablation treatment prior to MRI. Infiltrative HCC was defined as tumors showing an infiltrative growth without definitive evidence of expansive tumor growth pattern that is seen in nodular HCC on any portion of the tumor. Thus, we excluded patients with multinodular HCCs or nodular HCCs with a partially infiltrative growth pattern and patients with combined HCCICC. In all, 107 patients were excluded from our study for the following reasons: 106 patients had no histologic proof and one had a history of HCC treatment prior to MRI examination. Accordingly, we identified 18 patients (15 men and 3 women; age range, 35–66 years) with infiltrative HCCs fitting the inclusion criteria. The confirmation of tumors was based on surgical resection for three patients and on percutaneous biopsy for 15 patients. We reviewed the medical records to ensure that no patients included in the study had a possibility of hepatic tumors other than HCC. All patients were followed by computed tomography (CT) or MRI for more than 6 months after the index MRI. MR Examination All MR images were acquired using an Intera Achieva 3.0T whole-body MR system (Philips Healthcare, Best, The Netherlands) equipped with a dual-source parallel radiofrequency transmission system and a quadrature body coil. The baseline MR images included a T1weighted turbo field-echo in-phase and opposed sequence (TR/first echo TE, second echo TE, 10/2.3 [in-phase], 3.45 [opposed-phase]; flip angle, 15 ;
1239
matrix size, 256 194; bandwidth, 434.3 Hz/pixel), a breath-hold multishot T2-weighted sequence with an acceleration factor of 2 (1796/70; flip angle, 90 ; matrix size, 324 235; bandwidth, 258.4 Hz/pixel), a respiratory-triggered single-shot heavily T2-weighted sequence with an acceleration factor of 2 (1802/160; flip angle, 90 ; matrix size, 252 254; bandwidth, 420.9 Hz/pixel) with a 5 mm section thickness, and a field of view of 32–38 cm. For gadoxetic acid-enhanced imaging, unenhanced, arterial-phase (AP, 20–35 s), portal-phase (PVP, 60 s), 3-min late-phase, and 20-min hepatobiliary images (HBP) were obtained using a T1-weighted 3D turbofield-echo sequence (enhanced T1 high-resolution isotropic volume examination; eTHRIVE, Philips Health care) (3.1/1.5; flip angle, 10 ; matrix size, 256 256; bandwidth, 724.1 Hz/ pixel) with a 2-mm section thickness and a field of view of 32–38 cm. The measured voxel size was 1.5 1.5 4.0 mm and the reconstructed voxel size was 1.17 1.17 2.0 mm. The contrast agent was automatically administered intravenously at a rate of 1 mL/s for a dose of 0.025 mmol/kg body weight using a power injector, followed by a 20-mL saline flush. Diffusion-weighted single-shot echo planar imaging with the simultaneous use of respiratory triggering was performed using a TR/TE of 1,600 msec/70 msec; the TR was matched in each patient to the length of the respiratory cycle prior to the gadoxetic acidenhancement. The scanning parameters were as follows: a b-value of 0, 100, and 800 s/mm2; spectral presaturation with inversion recovery for fat suppression; matrix size, 100 100; acceleration factor of SENSE, 2.0; field of view, 35 35 cm; number of excitations, 4; slice thickness, 5 mm; slice gap, 1 mm; and 33 axial slices. Depending on the respiratory efficiency of each patient, the acquisition time for this sequence ranged from 3–4 minutes. The apparent diffusion coefficient (ADC) was calculated by a monoexponential function using b-values of 100 and 800 s/mm2. Image Analysis All MRIs were evaluated by the consensus of two gastrointestinal radiologists (S.L. and Y.K.K.) with 5 and 13 years of experience in liver MRI, respectively). All images were evaluated using a Picture Archiving and Communication System (PACS; Pathspeed, GE Medical Systems Integrated Imaging Solutions, Mt. Prospect, IL) with an optimal window setting adjustment in each case. Although there were minimal discrepancies in assigning lesion conspicuity between two readers, common consent was reached easily by consensus reading during an additional reading session. The MRI features evaluated were: 1) shape of tumors; 2) signal intensity the tumors relative to liver parenchyma on the precontrast T1-weighted images (T1WI), T2-weighted images (T2WI), gadoxetic acidenhanced images (AP, PVP, 3-min late, and HBP), and b-800 DWI; 3) presence of internal reticulation on tumors; and 4) presence of portal vein thrombosis. In addition, since most infiltrative HCC tends to accompany portal vein thrombosis with resultant arterial
1240
Lim et al.
Table 1 Clinical Characteristics of Patients With Infiltrative Hepatocellular Carcinoma Variables
Values
Age Sex Hepatitis B C AFP (ng/mL) PIVKA-II (mAU/mL) Child-Pugh Class (A/B/C) Treatment TACE TACE þ radiotherapy Systemic therapy (Sorafenib) Surgery None
Mean 57 (range; 35–66) 15 : 3 15 3 10.5–14500 592–64000 15/ 3/ 0 8 4 2 3 1
AFP: a-fetoprotein, PIVKA-II: protein induced by vitamin K absence II, TACE: transarterial chemoembolization.
hyperperfusion, lesion conspicuity with an emphasis on clear tumor margin at each of seven different images was assessed based on the following four grading scales. Grade 4 was defined as clear tumor margin showing more than of 90% of tumor circumference, Grade 3 as 50% < and 90%, Grade 2 as 10% < and 50%, and Grade 1 as 10%. Tumor extent was determined by the consent of two reviewers based on correlation between initial MRI and follow-up CT with/without MRI, which showed lipiodol uptake after TACE or tumor necrosis by radiotherapy with tumor shrinkage or progression. In addition, in tumors proven by biopsy, the biopsy report (sonography) was used to define the tumor extent: the area that showed same signal intensity to biopsied hepatic segment was considered tumor. Quantitative image analysis was performed by measuring the liver signal intensity and tumor signal intensity, using operator-defined regions of interest (ROIs) for each image by an abdominal radiologist who did not participate in the qualitative image analysis. For measurements within the lesion, the ROI was positioned manually as much as possible to avoid the necrotic foci. For signal intensities of the liver, ROIs were drawn in the same location as each sequence devoid of a large intrahepatic vessel. The standard deviation (SD) of background noise was measured along the phaseencoding direction outside the body just ventral to the right anterior abdominal wall and included respiratory or motion-related artifacts. However, since the introduction of a parallel acquisition technique that could interfere with estimating a correct background noise level, there has been no perfect maneuver for noise estimation, and this measurement has been commonly used in extensive studies. By applying the copy and paste function of the workstation, the position and size of the ROI were kept constant for all images as far as possible. The ROIs included at least 300 pixels (range 300–600 pixels). The tumor-to-liver contrast ratio for each sequence was calculated from the signal intensity of the liver and tumor as j(liver signal intensity tumor signal intensity)j / (liver signal intensity þ tumor signal
intensity); this formula yields a value between 0 and 1, with a larger value indicating greater relative tumorliver contrast (14). Statistical Analysis Subjective image ratings were compared with the Wilcoxon test corrected for multiple comparisons according to the Bonferroni adjustment. Paired t-tests were used for comparison of the tumor-to-liver contrast ratio at each sequence. Statistical analyses were performed using the SPSS software package, v. 20.0 (Chicago, IL). For all tests, P < 0.05 was considered significant.
RESULTS Table 1 shows the clinical features of study populations included in the study. This study was conducted in a viral hepatitis endemic Asian population. Thus, all patients had liver cirrhosis associated with viral hepatitis B (n ¼ 17) or viral hepatitis C (n ¼ 1). Based on the Child-Pugh classification, 15 patients were Class A and the remaining three were Class B. All three HCCs with surgical confirmation corresponded to Grade 2 according to Edmonson’s classification of HCC. Three surgically confirmed cases exhibited internal fibrous septum on histology. The MRI features of tumors are summarized in Table 2. Based on combined reading of all MRI sequences, all tumors were classified into one of two morphologic types as follows. Most tumors (n ¼ 16, 88.9%) were evident as a permeative mass with
Table 2 Summary of MRI Findings of 18 Infiltrative Hepatocellular Carcinomas Imaging characteristics of tumor Morphology Permeative mass Amorphous infiltration with thrombosed portal vein Portal vein thrombosis Main trunk with lobar branch Segmental branch Microvessel invasion Satellite nodules Signal intensity T1-weighted imaging (hypointensity) T2-weighted imaging (hyperintensity)/ internal bright foci Diffusion-weighted imaging (hyperintensity) Arterial phase (hypo/ iso/ high) Internal reticulation 3 min late phase 20 min hepatobiliary phase Portal venous phase Diffusion-weighted imaging Arterial phase T2-weighted imaging T1-weighted imaging
Number of tumor n 5 16 n52
n 5 16 n51 n51 n 5 12 n 5 18 n 5 18/10 n 5 18 n 5 9/ 6/ 3 n 5 18 n 5 15 n 5 14 n 5 11 n 5 10 n59 n50
Infiltrative HCC on 3T MRI
1241
Figure 1. A 50-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma. Large permeative liver mass with irregular border (arrows) is seen on T2weighted image (a), b-800 diffusionweighted image (b), unenhanced (c) and gadoxetic acid-enhanced arterial (d), portal (e), 3-min late (f), and 20-min hepatobiliary phase (g). Tumor margin is most clearly demonstrated on diffusion-weighted image (b). Reticulation within tumor is mostly accentuated on 3-min late phase (f). Note portal vein thrombosis (arrow in d) that is not strongly enhanced.
irregular outer contour, ranging in diameter from 4.0– 23.0 cm (mean 10.5 cm, Fig. 1). The remaining two were evident as poorly defined amorphous infiltration among thrombosed portal vein branches (8.0 cm and 13.0 cm in diameter, Fig. 2). The main feature of these two tumors was portal vein thrombosis rather than parenchymal mass. All but two tumors (88.9%) exhibited tumor thrombosis in the portal vein including right or left main trunk. One lesion (4 cm in size) accompanied tumor thrombosis in the segmental branch of portal vein. In the remaining one lesion (5 cm in size), portal vein thrombosis was not demonstrated upon MRI, but microscopic vascular invasion was exhibited on histology. The arterial hyperenhancement on hemiliver or hepatic segment was noted
in all 17 tumors with portal vein thrombosis. Twelve patients (66.7%) had satellite nodules. The signal intensities of most tumors were predominantly hypointense on the pre- and postcontrast T1WI, and moderately hyperintense on T2WI (Fig. 1). On AP imaging, most of the tumors (n ¼ 15, 83.3%) were considered hypovascular as they were hypo- (n ¼ 9) or isointense (n ¼ 6) with inhomogeneous areas of hyperenhancement, mostly at the periphery. The remaining three were heterogeneously hyperintense due to internal neovascularization. On b-800 DWI, both tumor and portal vein thrombosis were seen as hyperintense relative to background liver (Fig. 2). The internal reticulation was most frequently and clearly observed on 3-min late phase (n ¼ 18), followed by
1242
Lim et al.
Figure 2. A 58-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma. Poorly defined infiltrating tumors among thrombosed portal vein branches (arrows) is seen in right hepatic lobe on T2-weighted image (a), b-800 diffusion-weighted image (b), and unenhanced (c) and gadoxetic acidenhanced arterial (d), portal (e), 3min late (f), and 20-min hepatobiliary phase (g) of 3D T1-weighted image. Although tumor margin is not clear defined, the conspicuity of tumor is better with diffusionweighted image (b) than with other images.
20-min HBP (n ¼ 15), PVP (n ¼ 14), DWI (n ¼ 11), AP (n ¼ 10), T2WI (n ¼ 9), and T1WI (n ¼ 0) (Figs. 1, 3). Besides internal reticulation, most tumors showed homogenous signal intensity without obvious internal necrosis (Fig. 1). However, 10 tumors showed internal multifocal tiny bright foci on T2WI, indicating small necrotic foci or peliosis on histology. Table 3 summarizes the subjective assessments of lesion conspicuity as well as the tumor-to-liver contrast ratio for each of the sequences. The mean value for the subjective assignment of lesion conspicuity was highest in DWI (3.83 6 0.51), followed by T2WI (3.50 6 0.86), 20-min HBP (3.44 6 0.86), T1WI (2.94 6 1.00), 3-min late phase and PVP (2.89 6 1.02), and AP (2.61 6 0.85, Fig. 3). The differences for the mean
value were significant between DWI and other sequences (P < 0.05). All 16 tumors seen as permeative mass were assigned a rating score 4 on DWI. The remaining two infiltrating tumors among thrombosed portal veins were assigned a rating score 3 or 2 on DWI. For all tumors, no MR sequence achieved a higher score than DWI. As for the tumor-liver contrast ratio, it was the highest in DWI with 0.59 6 0.16, which was significantly higher than those of all other sequences (P < 0.0001), followed by T2WI (0.34 6 0.09), 20-min HBP (0.25 60.06), 3-min late phase (0.15 6 0.10), T1WI (0.14 6 0.06), PVP (0.14 6 0.09), and AP (0.11 6 0.07). The differences for the mean value were also significant for pairs of T1WI and T2WI, T2WI and AP,
Infiltrative HCC on 3T MRI
1243
Figure 3. A 51-year-old man with chronic hepatitis B and infiltrative hepatocellular carcinoma surgically confirmed. Small permeative liver mass with irregular border (arrows) is seen on T2-weighted image (a), b800 diffusion-weighted image (b), unenhanced (c) and gadoxetic acidenhanced arterial (d), portal (e), 3min late (f), and 20-min hepatobiliary phase (g) of 3D T1-weighted image. Tumor conspicuity is better with diffusion-weighted image (b) than with other images. Reticulation within tumor is mostly accentuated on 3-min late phase (f).
T2WI and PVP, T2WI and 3-min late, AP and HBP, PVP and HBP, 3-min late and HBP (P < 0.0001), and for between T2WI and HBP (P ¼ 0.014).
DISCUSSION In our study, the imaging features of most infiltrative HCCs agreed with previous reports (6,12,14), as they were observed as irregular-shaped hypovascular permeative masses with portal vein thrombosis. On AP imaging, all but three tumors (83.3%) were hypo- (n ¼ 9) or isointense (n ¼ 6) with inhomogeneous areas of hyperenhancement, mostly at the periphery. Thus, it is difficult to clearly differentiate tumoral enhancement and parenchymal arterial hyperperfusion
derived from portal vein thrombosis. With early dynamic phases gadoxetic acid-enhanced MRI, defining tumor boundary or detection as well as tumor characterization is challenging. In fact, the current noninvasive diagnostic criteria for HCC (7,8) have been refuted in most study cases. Despite a large tumor size, most tumors tended to show relatively homogenous hypo- and hyperintensity on T1- and T2WI, respectively, without obvious internal necrosis that is commonly seen in large nodular HCC. The most noticeable finding was that the majority of the tumors exhibited internal reticulation. This feature was most frequently demonstrated on 3-min late phase (n ¼ 18), followed by 20-min HBP (n ¼ 15). This is line with previous reports (12,21). However, its prevalence is higher than the 16.4% previously
1244
Lim et al.
Table 3 Results of Qualitative and Quantitative Image Analysis for Each MRI Sequence T1WI Rating for lesion conspicuity
1 2 3 4 Mean
P value Liver-tumor contrast ratio P value
T2WI
DWI
AP
PVP
3 min late
HBP
2 1 0 2 2 2 1 3 1 1 5 4 4 1 7 4 1 9 6 6 5 6 12 16 2 6 6 11 2.94 6 1.00 3.50 6 0.86 3.83 6 0.51 2.61 6 0.85 2.89 6 1.02 2.89 6 1.02 3.44 6 0.86 P < 0.0001 h, l, r, P ¼ 0.001 b, i, j, 0.002 a, m, n, t, u, 0.003f, 0.008o,0.014 c, g, 0.025 p,q, 0.317d, e, k, 1.000 s, 0.14 6 0.06 0.34 6 0.09 0.59 6 0.16 0.11 6 0.07 0.14 6 0.09 0.15 6 0.10 0.25 6 0.06 P < 0.0001 a,b, g, h, i, j, l, m, n, o, r, t, u, P ¼ 0.199 c, 0.819 d, 0.676 e, 0.210 f, 0.014 k, 0.306 p, 0.177 q, 0.506 s
Values are mean 6 SD. P values are T1WI and T2WI,a T1WI and DWI,b T1WI and AP,c T1WI and PVP,d T1WI and 3 min late,e T1WI and HBP,f T2WI and DWI,g T2WI and AP,h T2WI and PVP,i T2WI and 3 min late,j T2WI and HBP,k DWI and AP,l DWI and PVP,m DWI and 3 min late,n DWI and HBP,o AP and PVP,p AP and 3 min late,q AP and HBP,r PVP and 3 min,s PVP and HBP,t 3 min late and HBP.u
reported (21). The presence of a fibrous capsule or internal fibrous septum causing a mosaic appearance is a characteristic finding of HCC (22,23). Thus, we postulate that fibrosis could be responsible for internal reticulation, which was partially supported by three surgically proven cases. Abundant fibrous stroma in mass-forming cholangiocarcinoma is reportedly manifest as a wide area of central hyperenhancement on delayed phase gadoxetic acid-enhanced MRI (13). Considering the peculiar characteristic of gadoxetic acid including having relatively weak, short activity as an ECS agent and early, strong activity as a hepatocyte agent (24), the use of the 3-min late phase is prudent to depict fibrotic septum, as thin fibrosis could be strongly enhanced at this stage by an ECS agent characteristic of gadoxetic acid because it needs no more delayed time as in the broad fibrotic area of cholangiocarcinoma. Meanwhile, background hypointense tumor tissues are also accentuated by homogeneously enhanced adjacent liver that already accumulates gadoxetic acid into hepatocytes. Thus, the reticular appearance in gadoxetic acid-enhanced MRI could be a reliable feature for diagnosing infiltrative HCC and 3-min late phase image is mandatory for characterizing infiltrating HCC. In our study, the mean value for lesion conspicuity was highest with DWI, followed by T2WI and HBP. This was supported by quantitative results showing better tumor-to-liver contrast ratio with DWI and T2WI or HBP than with other sequences. This result agrees well with a recent report (14), in which T2WI and DWI exhibited greater tumor-liver contrast than an early conventional gadolinium-enhanced image. The greatest merit of gadoxetic acid is the clear depiction of HCC as hypointense on HBP (15,16). However, since most tumors present with portal vein thrombosis, peritumoral liver parenchyma could show decreased enhancement on HBP, causing decreased tumor-liver contrast (Fig. 1g). In particular, in two cases depicted as amorphous infiltration among thrombosed portal veins (Fig. 2g), it was difficult to differentiate tumor infiltration and liver parenchyma with decreased hepatic function on HBP. Nevertheless, in DWI all but one tumor were scored as 4 or 3. High tumor conspicuity on DWI could be explained by the recent advances in image quality due to improved MRI gradient performance, multichannel surface
receiver coils, parallel imaging techniques, improved fat suppression schemes, and application of a dualsource parallel radiofrequency transmission system (18). Based on DWI, HBP, and T2WI, infiltrative HCCs can be classified into two morphologic types. It is well known that infiltrative HCC can fail to have a discrete mass on imaging because this tumor blends into the background of the cirrhotic liver (11). However, 16 tumors (88.9%) displayed a discrete permeative mass with an irregular border, which is higher than the rate of 57.3% in a previous report (21). Meanwhile, as for two tumors seen as amorphous infiltration among thrombosed portal veins, portal vein thrombosis was their primary imaging feature rather than parenchymal mass. Given that portal vein thrombosis from infiltrative HCC might not be as highly enhanced as typical hypervascular HCC (Figs. 1, 2), caution needs to be taken in diagnosing this pattern of infiltrative HCC because even with DWI one of two tumors was assigned as poor conspicuity (Grade 2). In consideration of growing concern for the association between gadolinium-based contrast agents and nephrogenic systemic fibrosis (25), our results offer meaningful data because infiltrative HCC might not be readily detectable in other imaging modalities, while DWI, T1WI, and T2WI provide high tumor conspicuity. Furthermore, the sensitivity of DWI for detecting small nodular HCC is reportedly comparable to gadoxetic acid-enhanced MRI (16). In that sense, our observations could indicate that unenhanced MRI including DWI deserves to be introduced into HCC surveillance program. Further studies are needed. A major limitation of this study was that most tumors were not surgically confirmed, so the correlation between imaging and histological composition was not made. In particular, although this study focused on tumor conspicuity based on clear tumor margin at different MR sequences, a reference standard was not enough to determine exact tumor boundary. In addition, biopsied specimen could not represent whole tumor histology. Second, it was performed on a 3.0T system. Thus, our results might not necessarily be transferable to a 1.5T system. Third, due to the retrospective study design, we did not assess image acquired at variable delay timing after contrast administration. Fourth, most tumors were
Infiltrative HCC on 3T MRI
large, so whether these results might be exportable to small infiltrative HCC cannot be determined. Fifth, our study was also potentially limited by consensus review because we did not assess interobserver variability. Finally, this study included only infiltrative HCCs without control patients. Further investigation with hypovascular other hepatic tumors is needed. In conclusion, infiltrative HCCs showed two morphologic types: discrete permeative mass and indistinct infiltrations among thrombosed portal vein branches. DWI provided the highest tumor conspicuity compared to unenhanced T1WI and T2WI, and gadoxetic acid-enhanced dynamic and HBP as most tumors with portal vein thrombosis were clearly depicted as hyperintense on high-b value DWI. The gadoxetic acid-enhanced 3-min late image was useful in characterizing infiltrative HCC, as it clearly depicted internal reticulation in all tumors. POTENTIAL CONFLICT OF INTEREST We certify that all authors have had no relevant financial interests or personal affiliations in connection with the content of this article. REFERENCES 1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108. 2. El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999;340:745–750. 3. Baer HU, Gertsch P, Matthews JB, et al. Resectability of large focal liver lesions. Br J Surg 1989;76:1042–1044. 4. Okuda K, Noguchi T, Kubo Y, et al. A clinical and pathological study of diffuse type hepatocellular carcinoma. Liver 1981;1:280– 289. 5. Trevisani F, Caraceni P, Bernardi M, et al. Gross pathologic types of hepatocellular carcinoma in Italian patients. Relationship with demographic, environmental, and clinical factors. Cancer 1993; 72:1557–1563. 6. Kanematsu M, Semelka RC, Leonardou P, et al. Hepatocellular carcinoma of diffuse type: MR imaging findings and clinical manifestations. J Magn Reson Imaging 2003;18:189–195. 7. Bruix J, Sherman A, American Association for the Study of Liver Disease. Management of hepatocellular carcinoma: an update. Hepatology 2011;53:1020–1021. 8. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. J Hepatol 2001;35:421–430. 9. Myung SJ, Yoo JH, Kim KM, et al. Diffuse infiltrative hepatocellular carcinomas in a hepatitis B-endemic area: diagnostic and therapeutic impediments. Hepatogastroenterology 2006;53:266– 270.
1245 10. Benvegnu L, Noventa F, Bernardinello E, et al. Evidence for an association between the aetiology of cirrhosis and pattern of hepatocellular carcinoma development. Gut 2001;48:110–115. 11. Demirjian A, Peng P, Geschwind JF, et al. Infiltrating hepatocellular carcinoma: seeing the tree through the forest. J Gastrointest Surg 2011;15:2089–2097. 12. Kim YK, Han YM, Kim CS. Comparison of diffusion hepatocellular carcinoma and intrahepatic cholangiocarcinoma using sequentially acquired gadolinium-enhanced and Resovist-enhanced MRI. Eur J Radiol 2009;70:94–100. 13. Chong YS, Kim YK, Lee MW, et al. Differentiating mass-forming intrahepatic cholangiocarcinoma from atypical hepatocellular carcinoma using gadoxetic acid-enhanced MRI. Clin Radiol 2012; 67:766–773. 14. Rosenkrantz AB, Lee L, Matza BW, Kim S. Infiltrative hepatocellular carcinoma: comparison of MRI sequences for lesion conspicuity. Clin Radiol 2012;67:e105–111. 15. Haradome H, Grazioli L, Tinti R, et al. Additional value of gadoxetic acid-DTPA-enhanced hepatobiliary phase MR imaging in the diagnosis of early-stage hepatocellular carcinoma: comparison with dynamic triple-phase multidetector CT imaging. J Magn Reson Imaging 2011;34:69–78. 16. Park MJ, Kim YK, Lee WJ, et al. Small hepatocellular carcinomas: improved sensitivity by combining gadoxetic acid-enhanced and diffusion-weighted MR imaging patterns. Radiology 2012;264: 761–770. 17. Parikh T, Drew SJ, Lee VS, et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008; 246:812–822. 18. Taouli B, Koh DM. Diffusion-weighted MR imaging of the liver. Radiology 2010;254:47–66. 19. Piana G, Trinquart L, Meskine N, Barrau V, Beers BV, Vilgrain V. New MR imaging criteria with a diffusion-weighted sequence for the diagnosis of hepatocellular carcinoma in chronic liver diseases. J Hepatol 2011;55:126–132. 20. Holzapfel K, Eiber MJ, Fingerie AA, et al. Detection, classification, and characterization of focal liver lesions: value of diffusionweighted MR imaging, gadoxetic acid-enhanced MR imaging and the combination of both methods. Abdom Imaging 2012;37:74– 82. 21. Kneuertz PJ, Demirjian A, Firoozmand A, et al. Diffuse infiltrative hepatocellular carcinoma: assessment of presentation, treatment, and outcomes. Ann Surg Oncol 2012;19:2897–2907. 22. Stevens WR, Gulino SP, Batts KP, Stephens DH, Johnson CD. Mosaic pattern of hepatocellular carcinoma: histologic basis for a characteristic CT appearance. J Comput Assist Tomogr 1996;20: 337–342. 23. Ishiazaki M, Ashida K, Higashi T, et al. The formation of capsule and septum in human hepatocellular carcinoma. Virchows Arch 2001;438:574–580. 24. Tamada T, Ito K, Sone T, et al. Dynamic contrast-enhanced magnetic resonance imaging of abdominal solid organ and major vessel: comparison of enhancement effect between Gd-EOB-DTPA and Gd-DTPA. J Magn Reson Imaging. 2009;29:636–640. 25. Sadowski EA, Bennett LK, Chan MR, et al. Nephrogenic systemic fibrosis: risk factors and incidence estimation. Radiology 2007; 243:148–157.
