Tatsuo Inoue 1 Tomoko Hyodo. Yasuharu Imai 5 Atsushi Higaki. Kennichi Miyoshi 8 Masahiko Koda. Hironori Ochi 11 Masashi Hirooka

Transcription

1 J Gastroenterol (2016) 51: DOI /s ORIGINAL ARTICLE LIVER, PANCREAS, AND BILIARY TRACT Kupffer phase image of Sonazoid-enhanced US is useful in predicting a hypervascularization of non-hypervascular hypointense hepatic lesions detected on Gd-EOB-DTPA-enhanced MRI: a multicenter retrospective study Tatsuo Inoue 1 Tomoko Hyodo 2 Keiko Korenaga 3,4 Takamichi Murakami 2 Yasuharu Imai 5 Atsushi Higaki 3 Takeshi Suda 6 Toru Takano 7 Kennichi Miyoshi 8 Masahiko Koda 8 Hironori Tanaka 9,10 Hiroko Iijima 9 Hironori Ochi 11 Masashi Hirooka 11 Kazushi Numata 12 Masatoshi Kudo 1 Received: 26 August 2014 / Accepted: 6 June 2015 / Published online: 15 September 2015 Springer Japan 2015 Abstract Background It remains unknown whether Kupffer-phase images in Sonazoid-enhanced ultrasonography (US) can be used to predict hypervascularization of borderline lesions. Therefore, we aimed to clarify whether Kupffer-phase images in Sonazoid-enhanced ultrasonography can predict subsequent hypervascularization in hypovascular borderline lesions detected on hepatobiliary-phase gadoliniumethoxybenzyl-diethylenetriamine pentaacetic acid (Gd- EOB-DTPA)-enhanced magnetic resonance imaging. Methods From January 2008 to March 2012, 616 lowintensity hypovascular nodules were detected in hepatobiliary-phase images of Gd-EOB-DTPA-enhanced MRI at nine institutions. Among these, 167 nodules, which were confirmed as hypovascular by Gd-EOB-DTPA-enhanced MRI and Sonazoid-enhanced US, were evaluated in this study. Potential hypervascularization factors were selected based on their clinical significance and the results of previous reports. The Kaplan Meier model and log-rank test were used for univariate analysis and the Cox regression model was used for multivariate analysis. Results The cumulative incidence of hypervascularization of borderline lesions was 18, 37, and 43 % at 1, 2, and 3 years, respectively. Univariate analyses showed that tumor size (p = ) and hypoperfusion on Kupfferphase images in Sonazoid-enhanced US (p = 0.004) were associated with hypervascularization of the tumor. Multivariate analysis showed that tumor size [HR: 1.086, 95 % & Tatsuo Inoue tatsuoinouejp@yahoo.co.jp Masatoshi Kudo m-kudo@med.kindai.ac.jp Department of Gastroenterology and Hepatology, Kinki University School of Medicine, 377-2, Ohno-Higashi, Osaka-Sayama , Osaka, Japan Department of Diagnostic Radiology, Kinki University School of Medicine, 377-2, Ohno-Higashi, Osaka-Sayama , Osaka, Japan Department of Hepatology and Pancreatology, Kawasaki Medical School, Kurashiki, Japan Department of Gastroenterology and Hepatology, Kohnodai Hospital, National Center for Global Health and Medicine, Chiba, Japan Department of Gastroenterology, Ikeda Municipal Hospital, Osaka, Japan Division of Gastroenterology and Hepatology, Graduate School of Medical and Dental Sciences, Niigata University, Niigata, Japan Division of Radiation Oncology, Graduate School of Medical and Dental Sciences, Niigata University, Niigata, Japan Division of Medicine and Clinical Science, Department of Multidisciplinary Internal Medicine, Tottori University, Tottori, Japan Ultrasound Imaging Center, Hyogo College of Medicine, Nishinomiya, Japan Division of Hepatobiliary and Pancreatic Disease, Department of Internal Medicine, Hyogo College of Medicine, Nishinomiya, Japan Department of Gastroenterology and Metabology, Ehime University Graduate School of Medicine, Matsuyama, Japan Gastroenterological Center, Yokohama City University Medical Center, Yokohama, Japan

2 J Gastroenterol (2016) 51: confidence interval = , p = 0.004] and hypo perfusion on Kupffer-phase images [HR: 3.684, 95 % confidence interval = , p = ] were significantly different. Conclusions Kupffer-phase images in Sonazoid-enhanced US and tumor diameter can predict hypervascularization of hypointense borderline lesions detected on hepatobiliary-phase Gd-EOB-DTPA-enhanced MRI. Keywords Kupffer phase Sonazoid-enhanced ultrasonography Gadolinium ethoxybenzyldiethylenetriamine-pentaacetic acid-enhanced MRI Hepatocellular carcinoma Hypervascular transformation Abbreviations HCC Hepatocellular carcinoma US Ultrasonography CT Dynamic computed tomography MRI Magnetic resonance imaging DNs Dysplastic nodules Gd-EOB- Gadolinium ethoxybenzyl DTPA diethylenetriamine pentaacetic acid T2WI T2-weigthed images CTHA CT during hepatic arteriography HRs Hazard ratios CIs Confidence intervals CTAP CT during arterial portography Introduction Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide and is a major cause of death in patients with cirrhosis. Recent advances in imaging modalities and periodic follow-up of chronic liver disease, particularly cirrhosis, with ultrasonography (US), dynamic computed tomography (CT), magnetic resonance imaging (MRI), or measurement of tumor markers, have aided in the detection of small, early stage HCC including early HCC, low-grade dysplastic nodules (DNs), and high-grade DNs. The majority of these lesions are hypovascular because these nodules lacking hypervascularity in the arterial phase on imaging modalities and are called borderline lesions [1, 2]. A hepatocyte-specific contrast agent, gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) (Primovist ), has unique pharmacodynamics: it is taken up by hepatocytes and subsequently excreted into the bile ducts [3]. Several previous studies showed a significant correlation between the expression level of organic anion transporting polypeptide 1B3 and enhancement ratio in HCCs in hepatobiliary-phase images of Gd-EOB-DTPA-enhanced MR [4 7]. Kogita et al. reported that reduction in Gd-EOB-DTPA uptake might be an early event of hepatocarcinogenesis that occurs before portal blood flow reduction [8]. In fact, the detection of premalignant/borderline lesions, which are difficult to detect on the basis of intratumoral hemodynamic changes, has improved dramatically with the use of Gd-EOB-DTPA-enhanced MRI [9 11]. It is important to identify the borderline lesions that are at risk of hypervascularization. Previous studies have revealed that tumor size, tumor growth rate, presence of fat, and hyperintensity on T1- and T2-weighted images indicate a high probability of hypervascularization of borderline lesions [12 20]. Sonazoid is a second-generation sonographic contrast agent that consists of perfluorobutane gas microbubbles with phospholipid monolayer shells. In addition to a realtime fine vascular image in the vascular phase [21 24], Sonazoid is phagocytosed by Kupffer cells in the liver after administration, which enables persistent and stable enhancement image, called Kupffer-phase imaging. Kupffer-phase imaging is typically performed 10 min after Sonazoid injection, at which time the normal hepatic parenchyma is enhanced, and malignant lesions that contain few or no Kupffer cells are clearly delineated as contrast defects. A few studies have suggested a relationship between enhancement patterns of the Kupffer phase in Sonazoid-enhanced US and histological grading of HCC [25, 26]. Ohama et al. reported a correlation between the hepatobiliary-phase image of Gd-EOB-DTPA-enhanced MR and the Kupffer-phase image in Sonazoid-enhanced US [27]. They concluded that the uptake of Sonazoid begins to decrease later than that of Gd-EOB-DTPA in stepwise hepatocarcinogenesis of borderline lesions. Gd-EOB- DTPA-enhanced MRI can be used to detect many borderline lesions; however, repeated MRI examinations are not suitable because of their cost and severe side effects, such as anaphylactic shock and nephrogenic systemic fibrosis although the onset of severe side effects rarely occurs if subjects are carefully selected [28]. In clinical practice, it is important to evaluate the risk of hypervascularization of borderline lesions by using cost-effective and safe methods. Therefore, we hypothesized that the Kupffer-phase image in Sonazoid-enhanced US could be evaluated to predict hypervascularization of borderline lesions. Thus far, no study has reported the usefulness of Kupffer-phase images in Sonazoid-enhanced US for prediction of hypervascularization of borderline lesions. Therefore, the aim of our study was to evaluate whether the Kupffer-phase image in Sonazoid-enhanced US can be used to predict subsequent hypervascularization in non-hyper borderline lesions detected in the hepatobiliary-phase image of Gd-EOB- DTPA-enhanced MR.

3 146 J Gastroenterol (2016) 51: Methods Patients The selection of the study population is presented in a flow chart (Fig. 1). In the present study, Gd-EOB-DTPA MR was performed for routine examination or evaluation of a nodule detected by B-mode US. Sonazoid-enhanced US was performed to reevaluate tumor vascularity in the vascular phase and to evaluate uptake of Sonazoid in the Kupffer-phase detected as low intense in the hepatobiliaryphase image of Gd-EOB-DTPA-enhanced MR. Between February 2008 and March 2011, 616 hypovascular nodules with low intensity in hepatobiliary-phase images of Gd- EOB-DTPA-enhanced MR were recruited from nine institutions that participated in the present study. Of these, we excluded patients with Child-Pugh class C because of insufficient enhancement on MR imaging [29]; in addition, 214 patients who did not undergo Sonazoid-US were excluded. Finally, 112 patients with 167 hepatic nodules that were diagnosed as hypovascular hepatic nodules on Gd-EOB-DTPA-enhanced MR and Sonazoid-enhanced US were evaluated in the present study. The baseline characteristics of the patients are shown in Table 1. Our retrospective study design was approved by the institutional review board of Kinki University Faculty of Medicine. The requirement to obtain informed consent was waived. Image analysis All images were interpreted independently by an experienced board of certified radiologists, sonologists, and gastroenterologists at each institution who were aware that the patients were at risk for HCC but had no other clinical information. Discrepancies between the readers were resolved by discussion to reach consensus. MR imaging All the institutions that participated in the present study were equipped with high-field-strength (at least 1.5 T) MRI units. The pulse sequence parameters were set according to the local experience of the readers. The MR machines were 3.0 T (Achieva, Philips Medical Systems, Best, Netherlands; and MAGNETOM Skyra, Siemens, Erlangen, Germany) or 1.5 T systems (Signa Excite HDxt, GE Healthcare, Milwaukee, Wisconsin; Gyroscan Intera Nova and Achieva, Philips Medical Systems; Excelart Vantage Powered by Atlas, Toshiba Medical Systems, Tochigi, Japan; and MAGNETOM Symphony and Avanto, Siemens, Erlangen, Germany). Unenhanced, arterial, portal venous, late phase images were obtained according to the local standard of care. Hepatobiliary phase images were obtained more than 20 min after injection of Gd-EOB- DTPA at each institution. Ultrasonography Different US systems were used at the different institutions (Table 2). First, a grayscale US scan was obtained for each lesion. Each focal liver lesion was measured. After baseline US scan evaluation, the sonologist initiated the contrastspecific mode. Before intravenous injection of the microbubble contrast agent Sonazoid (Daiichi Sankyo, Tokyo, Japan), the persistence of the image display on the US machine was set to zero, the signal gain was registered below the noise threshold, and one focus was positioned below the level of the lesion. The contrast-enhanced examination consisted of the early vascular phase, late vascular phase, and Kupffer phase (at least 10 min after the injection of the contrast agent). The real-time images were stored on a hard disk so that the images could be recalled as necessary. We identified vascular patterns in the early vascular phase as isovascular or hypovascular and classified tumor perfusion patterns on Kupffer-phase images as hyper-perfusion, iso-perfusion, or hypo-perfusion. Evaluation of tumor vascularity Fig. 1 Flow chart depicting selection of the study population Tumor vascularity was evaluated by Sonazoid-enhanced US and Gd-EOB-DTPA-enhanced MRI at each institution. A non-hypervascular nodule was defined as a nodule that showed non-hypervascularity relative to the surrounding liver parenchyma during the arterial phase of dynamic imaging for both the modalities. Arterial enhancement was assessed by visual inspection. The exclusion criteria for hypointense lesions observed in hepatobiliary-phase images of Gd-EOB-DTPA-enhanced MR were as follows: (a) hypervascularity on initial dynamic MRI and/or early vascular phase of Sonazoid-enhanced US (i.e., exclusion of

4 J Gastroenterol (2016) 51: Table 1 Baseline characteristics of the patients and imaging findings of the nodules Parameters Hypervascularization at follow-up examinations p value* Yes (n = 43) No (n = 124) Tumor diameter 12 ± ± Coexistence of hypervascular HCC (yes/no) 25/18 51/ Liver disease (chronic hepatitis/liver cirrhosis/unknown) 8/28/7 44/ Etiology (HCV/HBV/other/unknown) 19/12/6/6 69/29/16/ Child Pugh score (A/B/C/unknown) 34/3/0/6 97/14/0/ Previous history of HCC treatment (yes/no) 25/18 51/ Unenhanced T1-weighted images on MRI (low intensity/iso-hyper intensity) b 29/14 84/ Fat-containing lesions on in- and opposed-phase images on MRI (yes/no) b 10/33 26/ Unenhanced T2-weighted images on MRI (hyper intensity/iso-low intensity) b 9/34 16/ Kupffer-phase image of Sonazoid-enhanced US (hypo-perfusion/iso-hyper perfusion) b 15/28 17/ Arterial-phase image of dynamic study a (hypovascular/isovascular) b 23/20 53/ Portal-phase image of dynamic study a (hypovascular/isovascular) b 15/28 52/ Serum a-fetoprotein level (AFP) (\12 ng/ml/[12 ng/ml/na) 17/26/0 65/57/ Serum des-c-carboxy prothrombin (DCP) (\21mAU/ml/ 16/24/3 60/57/ [21 mau/ml/na) Observation period Data are mean ± SD When CTHA and CTAP were performed, we assessed the respective imaging findings HCC hepatocellular carcinoma, HCV hepatitis C virus, HBV hepatitis B virus, NA not assessed * p \ 0.05 indicates statistical significance a Dynamic study refers to any of the available modalities (Gd-EOB-DTPA enhanced MRI, dynamic CT, contrast-enhanced US, CTHA, and CTAP) b Qualitative assessment Table 2 US equipment and contrast-enhanced modes Equipment, manufacturer Scanning mode No. of lesions scanned Amount of Sonazoid a Scan time of Kupffer phase (min) Mechanical index (MI) High MI burst b GE logiq 7 Coded harmonic angio ml/body c GE logiq E9 Phase inversion amplitude ml/kg 0.26 modulation Toshiba Aplio XG Toshiba Aplio XV HITACHI Aloka Prosound alpha- 10 Pulse subtraction ml/kg , 1.52 Pure harmonic detection mode a The amount of Sonazoid was set at each institution 0.5 ml/body ml/kg ml/body b To eliminate background B-mode findings, the high MI contrast mode was used when the tumor showed hyper echogenicity on B-mode US c A total of 0.2 ml Sonazoid was injected classical HCC and other hypervascular tumors); (b) delayed enhancement on initial dynamic MRI and/or late vascular phase of Sonazoid-enhanced US (i.e., exclusion of slow-filling hemangiomas); (c) strong high intensity on T2- weigthed images (T2WI) (i.e., excluding cysts); and (d) lesion size of less than 3 mm (because the slice thickness for the hepatobiliary phase of Gd-EOB-DTPA-enhanced MR was 3 mm).

5 148 J Gastroenterol (2016) 51: Table 3 Stepwise multiple Cox regression analysis Results of univariate analysis Parameters p value* Age (\70/C70 years) 0.76 Sex (male/female) 0.29 Tumor size (continuous value) Coexistence of hypervascular HCC (yes/no) 0.17 Liver disease (chronic hepatitis/liver cirrhosis/unknown) 0.87 Etiology of liver disease (HCV/HBV/other/unknown) 0.89 Child-Pugh score (A/B/C/unknown) 0.13 Unenhanced T1-weighted images on MRI (low intensity/iso-hyper intensity) 0.57 Previous history of HCC treatment (yes/no) 0.17 Fat-containing lesions on in- and opposed-phase images on MRI (yes/no) 0.93 Unenhanced T2-weighted images on MRI (hyper intensity/iso-low intensity) 0.43 Kupffer-phase image of Sonazoid-enhanced US (hypo-perfusion/iso-hyper perfusion) Arterial-phase image in dynamic study (hypovascular/isovascular) 0.91 Portal-phase image in dynamic study (hypovascular/isovascular) 0.07 Serum a-fetoprotein level (AFP) (\12 ng/ml/c20 ng/ml/na) 0.22 Serum des-c-carboxy prothrombin (DCP) (\ 21 mau/ml/c 21 mau/ml/na) 0.14 Results of multivariate analysis Parameters Ex (B) 95 % CI p value* Tumor diameter (mm) Kupffer phase (hypo-perfusion) Coexistence of hypervascular HCC History of local therapy for HCC Image of fat-suppressed MR T2-weighted images Fat-containing lesions on in- and opposed-phase images * p \ 0.05 indicates statistical significance Definition of hypervascular transformation During the follow-up period, when a CT hepatic arteriography (CTHA) image, Sonazoid-enhanced US, dynamic CT, and MR in the early phase indicated a region of hyperattenuation relative to the area surrounding the nodule, it was described as hypervascularization. Hypervascularization was confirmed by more than one modality, and the earliest date of the imaging examinations was used as the reference point. Statistical analysis All analyses were conducted at the nodule level. R software (Version ; R Foundation for Statistical Computing, Vienna, Austria) was used for statistical analysis. To evaluate the independent prognostic significance of baseline covariates for subsequent hypervascularization, a multivariate Cox proportional hazard model was used. Because 26 patients had multiple nodules detected at two or more follow-up examinations, we used the coxph function from the survival package in the R software, with the cluster option. This method accounted for correlation induced by having multiple nodules per patient and used robust variance estimates [30]. Before model selection, bivariate analysis was performed by using Spearman rank correlations to test for collinearity among independent variables. As a result, Spearman correlation coefficients for variables were generally below 0.5, which suggests that multicollinearity was not a concern. Hazard ratios (HRs) and 95 % confidence intervals (CIs) were calculated. Continuous variables were presented as the median and range. Continuous variables such as tumor size and observation period were compared with Mann Whitney U test, and other categorical variables were compared by Fisher s exact test or the Chi-squared test. Based on the analysis of lesions, the cumulative risk of a non-hypervascular tumor transforming into classical HCC was calculated according to the Kaplan Meier method. We calculated the relative risk using Cox proportional hazard regression analysis. Actuarial analysis of the cumulative incidence of vascularization was performed with the Kaplan Meier method, and the differences were tested by

6 J Gastroenterol (2016) 51: Fig. 2 Cumulative rates for hypervascularization of hypointense lesions. The overall cumulative incidence of hypervascular transformation was 18 % at 12 months, 37 % at 24 months, and 43 % at 18 months. Forty-three (25.7 %) out of the 167 cases showed hypervascular transformation in the arterial phase of dynamic imaging during the follow-up period the log-rank test. A p value \0.05 was considered to denote a statistically significant difference. Results Nodule characteristics Table 1 shows the characteristics of the nodules that were vascularized and of those that were not. Of the 167 nodules, 43 (25.7 %) showed hypervascular transformation in the arterial phase of dynamic imaging during the follow-up period. Fisher s exact test showed that at the start of follow-up, nodules with and without vascularization showed significant differences with respect to the average tumor diameter (p = 0.04) and Kupffer-phase images in Sonazoid-enhanced US (p = 0.005). Cumulative incidence of hypervascular transformation The overall cumulative incidence of hypervascular transformation was 18 % at 12 months, 37 % at 24 months, and 43 % at 18 months (Fig. 2). Univariate analysis using the log-rank test revealed that hypoperfusion on Kupffer-phase images of contrast-enhanced US using Sonazoid and tumor diameter were correlated with hypervascular transformation (Table 3). Next, the two significant factors identified by univariate analysis and another four variables, including coexistence of hypervascular HCC, history of local therapy for HCC, fatsuppressed T2-weighted images, and fat-containing lesions on in- and opposed-phase images, were further analyzed by multivariate analysis using the Cox regression model because these 4 variables have been described as important predictors [14]. Consequently, tumor diameter (HR = 1.086; p = 0.004, 95 % CI: 1.027, 1.148) and hypo-perfusion on Kupffer-phase images in Sonazoid-enhanced US (HR = 3.684; p = , 95 % CI: 1.798, 7.546) were identified as independent factors for hypervascular transformation (Table 3). Subsequently, we compared the incidence of hypervascular transformation of tumors with these risk factors based on the Kaplan Meier curve. The incidence of hypervascularization was significantly higher in the groups with these prognostic factors (Figs. 3, 4). The optimum cut-off point for tumor size was estimated to be 8.6 mm by receiver operating characteristic (ROC) curves (figure not shown). Discussion To the best of our knowledge, this study is the first to show the usefulness of Kupffer-phase images in Sonazoid-enhanced US for predicting hypervascularization of non-hypervascular borderline lesions detected in hepatobiliaryphase images of Gd-EOB-DTPA-enhanced MR. Previously, Kupffer cells were thought to be absent in overt HCC tissues, but recently investigators have shown that Kupffer cells exist in early stage HCCs [31]. Furthermore, the histologic grade of HCC has been shown to correlate with the number of Kupffer cells present. Kupffer cells, the resident liver macrophages, constitute 31 % of the sinusoidal cells [32]. They are more numerous (43 %) in the periportal zone of the lobule. In addition to being more numerous, periportal Kupffer cells are larger, have more lysosomes, and take up more particles than do middle- or central-zone Kupffer cells [33]. Sugihara et al. [34] reported that cancerous tissue of well-differentiated HCCs possesses blood spaces that are similar to the normal sinusoids. Therefore, blood spaces are expected to possess morphologic and functional characteristics similar to those of normal sinusoids but that as tumors grew in size and came to have a lower histologic grade, the blood spaces increased in apparent capillarization and became morphologically different from normal sinusoids. When neovascularization suggested by unpaired artery [35] occurred, normal sinusoids were gradually destroyed and the portal supply declined. Therefore, we speculate that Kupffer cells lose their functional ability to take up microbubbles. Although all most all premalignant/borderline lesions possess portal areas and blood spaces similar to those of normal sinusoids may take up microbubbles in a manner similar to non-nodular adjacent tissue, some of the

7 150 J Gastroenterol (2016) 51: Fig. 3 Cumulative rates for hypervascularization of hypointense lesions. Lesions are stratified according to the tumor size and Kupfferphase image in Sonazoid-enhanced US. The incidence of hypervascularization was significantly higher when the tumor diameter was greater than 8.6 mm Fig. 4 Case presentation of borderline lesions (arrowheads indicate the tumor). a, b Tumors showing isovascularity on CTHA (a) and CTAP(b). Initial Gd-EOB-DTPA enhanced MRI (c f). Tumors showing: Isointensity on in-phase (c) and opposed-phase images (d). e Isointensity on T2-weighted image. f Low intensity in hepatobiliaryphase image of Gd-EOB-DTPA-enhanced. MRI Initial Sonazoidenhanced US (g i). g, k Monitor mode, h, i, l contrast mode. Tumors showing: g, h Hypovascularity in early vascular phase. i Hypoperfusion in Kupffer-phase. j Hypervascular spot in nodule in arterial-phase of Gd-EOB-DTPA-enhanced MRI at 30 months after start of followup. k, l hypervascular spot in nodule in early vascular phase in Sonazoid-enhanced US

8 J Gastroenterol (2016) 51: premalignant/borderline lesions might have lost their ability to possess microbubbles resulting in the tumorous perfusion defects observed in Kupffer phase imaging before hypervascularization detected by other imaging modalities. In the present study, 135 out of 167 nodules showed iso-perfusion on the Kupffer-phase image. Although the improved detection ability of Gd-EOB- DTPA-enhanced MRI enabled us to detect many premalignant/borderline lesions, which are difficult to detect on the basis of intratumoral hemodynamic changes, the new challengeisindecidingwhichlesionsshouldbemonitoredmore carefully.however,byusingsonazoid-enhancedus,kupffer cellfunctioncanbeevaluatedfortumorsthatcannotbeevaluated by other imaging modalities. Therefore, we evaluated the Kupffercellfunctionofthesepremalignant/borderlinelesions using Sonazoid, and we found that it can be a useful tool for predicting hypervascularization. Sonazoid-enhanced US is easytoperform,anditismorecost-effectivethanmrianddoes not involve exposure to ionizing radiation, unlike contrast-enhancedct,ctha,andctduringarterialportography(ctap). In general, when non-hypervascular hypointense lesions are detectedbygd-eob-dtpa-enhancedmri, we stronglysuggest that Sonazoid-enhanced US should be performed to evaluatekupffercellfunctionandpredicthypervascularization. Although we conducted a cooperative study, collecting several non-hypervascular tumors and investigating the natural outcome of hypointense lesions in a nationwide manner, this study did have some limitations. The principal limitation of this study was the variation in the equipment used at different institutions. Although this limitation may have been inevitable because of the multicenter nature of the study, the different sensitivities of the different equipment used should be considered in future investigations. Second, this was a retrospective study, which may have introduced bias in data homogeneity. Moreover, the interval between the follow-up examinations varied for each individual and among the patients. Prospective studies with consistent follow-up intervals must be performed to overcome this limitation. In this study, most of the follow-up studies were performed at intervals of 3 and 6 months, which is consistent with the usual practice for follow-up of patients at high risk of HCC [36, 37]. Third, on imaging analysis, our study was potentially limited by consensus review. To minimize operator bias, well-trained radiologists and physicians reviewed the images. However, in the future, assessment of interobserver variability is warranted. Fourth, the ability of Sonazoid-enhanced US to detect hypervascularization in a nodule varies considerably according to depth from the surface of the liver, coarseness of liver parenchyma, and patient s obesity. It would be appropriate to use Gd-EOB-DTPA-enhanced MR only for the evaluation. Finally, no pathologic proof existed for the nodules evaluated in the present study. However, the major purpose of the present study was to evaluate whether the Kupffer-phase image in Sonazoid-enhanced US can predict hypervascular transformation in borderline lesions; we did not aim to distinguish among dysplastic nodules, early HCCs, and well-differentiated HCCs. In conclusion, this study shows that the Kupffer-phase image in Sonazoidenhanced US is useful for the prediction of hypervascularization of non-hypervascular hypointense hepatic lesions detected on Gd-EOB-DTPA-enhanced MRI. Conflict of interest of interest. References The authors declare that they have no conflict 1. Kudo M. Imaging diagnosis of hepatocellular carcinoma and premalignant/borderline lesions. Semin Liver Dis. 1999;19(3): Krinsky G. Imaging of dysplastic nodules and small hepatocellular carcinomas: experience with explanted livers. Intervirology. 2004;47(3 5): Schuhmann-Giampieri G, Schmitt-Willich H, Press WR, et al. Preclinical evaluation of Gd-EOB-DTPA as a contrast agent in MR imaging of the hepatobiliary system. Radiology. 1992;183(1): Kitao A, Zen Y, Matsui O, et al. Hepatocellular carcinoma: signal intensity at gadoxetic acid-enhanced MR imaging correlation with molecular transporters and histopathologic features. Radiology. 2010;256(3): Narita M, Hatano E, Arizono S, et al. Expression of OATP1B3 determines uptake of Gd-EOB-DTPA in hepatocellular carcinoma. J Gastroenterol. 2009;44(7): Leonhardt M, Keiser M, Oswald S, et al. Hepatic uptake of the magnetic resonance imaging contrast agent Gd-EOB-DTPA: role of human organic anion transporters. Drug Metab Dispos. 2010;38(7): Kitao A, Matsui O, Yoneda N, et al. The uptake transporter OATP8 expression decreases during multistep hepatocarcinogenesis: correlation with gadoxetic acid enhanced MR imaging. Eur Radiol. 2011;21(10): Kogita S, Imai Y, Okada M, et al. Gd-EOB-DTPA-enhanced magnetic resonance images of hepatocellular carcinoma: correlation with histological grading and portal blood flow. Eur Radiol. 2010;20(10): Inoue T, Kudo M, Komuta M, et al. Assessment of Gd-EOB- DTPA-enhanced MRI for HCC and dysplastic nodules and comparison of detection sensitivity versus MDCT. J Gastroenterol Kudo M. Early hepatocellular carcinoma: definition and diagnosis. Liver Cancer. 2013;2(2): Ichikawa T, Sano K, Morisaka H. Diagnosis of pathologically early HCC with EOB-MRI: experiences and current consensus. Liver Cancer. 2014;3(2): Kumada T, Toyoda H, Tada T, et al. Evolution of hypointense hepatocellular nodules observed only in the hepatobiliary phase of gadoxetate disodium-enhanced MRI. AJR Am J Roentgenol. 2011;197(1): Akai H, Matsuda I, Kiryu S, et al. Fate of hypointense lesions on Gd-EOB-DTPA-enhanced magnetic resonance imaging. Eur J Radiol. 2012;81(11): Motosugi U, Ichikawa T, Sano K, et al. Outcome of hypovascular hepatic nodules revealing no gadoxetic acid uptake in patients

9 152 J Gastroenterol (2016) 51: with chronic liver disease. J Magn Reson Imaging. 2011;34(1): Takayama Y, Nishie A, Nakayama T, et al. Hypovascular hepatic nodule showing hypointensity in the hepatobiliary phase of gadoxetic acid-enhanced MRI in patients with chronic liver disease: prediction of malignant transformation. Eur J Radiol. 2012;81(11): Kobayashi S, Matsui O, Gabata T, et al. Relationship between signal intensity on hepatobiliary phase of gadolinium ethoxybenzyl diethylenetriaminepentaacetic acid (Gd-EOB-DTPA)-enhanced MR imaging and prognosis of borderline lesions of hepatocellular carcinoma. Eur J Radiol. 2012;81(11): Hyodo T, Murakami T, Imai Y, et al. Hypovascular nodules in patients with chronic liver disease: risk factors for development of hypervascular hepatocellular carcinoma. Radiology. 2013;266(2): Higaki A, Ito K, Tamada T, et al. High-risk nodules detected in the hepatobiliary phase of Gd-EOB-DTPA-enhanced MR imaging in cirrhosis or chronic hepatitis: incidence and predictive factors for hypervascular transformation, preliminary results. J Magn Reson Imaging. 2013;37(6): Takechi M, Tsuda T, Yoshioka S, et al. Risk of hypervascularization in small hypovascular hepatic nodules showing hypointense in the hepatobiliary phase of gadoxetic acid-enhanced MRI in patients with chronic liver disease. Jpn J Radiol. 2012;30(9): Inoue T, Hyodo T, Murakami T, et al. Hypovascular hepatic nodules showing hypointense on the hepatobiliary-phase image of Gd-EOB-DTPA-enhanced MRI to develop a hypervascular hepatocellular carcinoma: a nationwide retrospective study on their natural course and risk factors. Dig Dis. 2013;31(5 6): Moriyasu F, Itoh K. Efficacy of perflubutane microbubble-enhanced ultrasound in the characterization and detection of focal liver lesions: phase 3 multicenter clinical trial. AJR Am J Roentgenol. 2009;193(1): Murakami T, Imai Y, Okada M, et al. Ultrasonography, computed tomography and magnetic resonance imaging of hepatocellular carcinoma: toward improved treatment decisions. Oncology. 2011;81(Suppl 1): Kudo M, Hatanaka K, Kumada T, et al. Double-contrast ultrasound: a novel surveillance tool for hepatocellular carcinoma. Am J Gastroenterol. 2011;106(2): Luo W, Numata K, Morimoto M, et al. Role of Sonazoid-enhanced three-dimensional ultrasonography in the evaluation of percutaneous radiofrequency ablation of hepatocellular carcinoma. Eur J Radiol. 2010;75(1): Korenaga K, Korenaga M, Furukawa M, et al. Usefulness of Sonazoid contrast-enhanced ultrasonography for hepatocellular carcinoma: comparison with pathological diagnosis and superparamagnetic iron oxide magnetic resonance images. J Gastroenterol. 2009;44(7): Sugimoto K, Moriyasu F, Saito K, et al. Comparison of Kupfferphase Sonazoid-enhanced sonography and hepatobiliary-phase gadoxetic acid-enhanced magnetic resonance imaging of hepatocellular carcinoma and correlation with histologic grading. J Ultrasound Med. 2012;31(4): Ohama H, Imai Y, Nakashima O, et al. Images of Sonazoidenhanced ultrasonography in multistep hepatocarcinogenesis: comparison with Gd-EOB-DTPA-enhanced MRI. J Gastroenterol Grobner T, Prischl FC. Gadolinium and nephrogenic systemic fibrosis. Kidney Int. 2007;72(3): Willatt JM, Hussain HK, Adusumilli S, et al. MR imaging of hepatocellular carcinoma in the cirrhotic liver: challenges and controversies. Radiology. 2008;247(2): Lin DY. Cox regression analysis of multivariate failure time data: the marginal approach. Stat Med. 1994;13(21): Tanaka M, Nakashima O, Wada Y, et al. Pathomorphological study of Kupffer cells in hepatocellular carcinoma and hyperplastic nodular lesions in the liver. Hepatology. 1996;24(4): Bouwens L, Baekeland M, De Zanger R, et al. Quantitation, tissue distribution and proliferation kinetics of Kupffer cells in normal rat liver. Hepatology. 1986;6(4): Ruttinger D, Vollmar B, Wanner GA, et al. In vivo assessment of hepatic alterations following gadolinium chloride-induced Kupffer cell blockade. J Hepatol. 1996;25(6): Sugihara S, Kojiro M, Nakashima T. Ultrastructural study of hepatocellular carcinoma with replacing growth pattern. Acta Pathol Jpn. 1985;35(3): Kojiro M, Wanless I, Alves V, et al. Pathologic diagnosis of early hepatocellular carcinoma: a report of the international consensus group for hepatocellular neoplasia. Hepatology. 2009;49(2): Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. Hepatology. 2011;53(3): Kudo M, Izumi N, Kokudo N, et al. Management of hepatocellular carcinoma in Japan: consensus-based clinical practice guidelines proposed by the Japan Society of Hepatology (JSH) 2010 updated version. Dig Dis. 2011;29(3):