28. Gastrointestinal Imaging Papers (Liver Imaging)

Scientific Session 28—Gastrointestinal Imaging Papers (Liver Imaging)

Thursday, April 30, 2009, 1:30 pm - 3:30 pm

Abstracts 235–245

Moderators: D. Nakamoto, B. Choi

Keynote Address: Imaging of Hepatocellular Carcinoma with MR Contrast Agent—B. Choi

1:40 pm

235. Incidence of New Foci of Hepatocellular Carcinoma in Patients Who Underwent Radiofrequency Ablation of Hepatocellular Carcinoma: Role of MDCT

Bartolotta, T.*; Taibbi, A.; Putignano, L.; Galia, M.; Lo Re, G.; Malizia, G.; Midiri, M.; Lagalla, R. University Hospital, Palermo, Italy

Address correspondence to A. Taibbi (taibbiadele@hotmail.com)

Objective: The objective was to assess the incidence of new foci of hepatocellular carcinoma (HCC) by means of MDCT in patients who had undergone radiofrequency ablation (RFA) of HCC.

Materials and Methods: MDCT studies of 129 patients (92 men and 37 women; age: 48–89 years—mean: 72 years), with RFA-treated HCCs followed-up for one to 68 months (mean: 22 months), were retrospectively reviewed to detect the presence of new foci of HCC, defined as hypervascular focus in the arterial phase with wash-out in the portal-venous or equilibrium phases, arising at least at 2.1 cm from the treated nodule either in the same or in a different liver segment. All new nodules were definitively proven through biopsy and/or MR imaging and contrast-enhanced ultrasound findings.

Results: A total of 189 new HCCs (size:1–5 cm; mean: 1.7 cm) were detected in the same (n = 38; 20%) or in a different (n = 151;80%) liver segment (p < 0.001) in 70/129 (54.3%) patients, followed-up for one to 59 months (mean: 26 months); 25/70 patients had a single new HCC, whereas 45/70 patients had two (17 patients), three (10 patients), four (six patients), five (four patients), six (three patients), seven (two patients) or eight (three patients) new HCCs. The remaining 59 patients without new HCCs were followed-up for 1–68 month (mean: 15 months) (p < 0.001). Disease free interval-time was three to 54 months (mean: 14.5 months).

Conclusion: MDCT follow-up in patients with RFA-treated HCCs reveals, especially when follow-up is longer than one year, a high incidence of new HCCs mainly detected in a liver segment different from that of treated HCC.


  • * Will present paper

1:50 pm

236. Correlation Between Fatty Liver Grading and Abdominal Fat, Using Hepatic Attenuation Value and Fat Volume

Jang, S.*; Lee, C.; Kim, K.; Choi, J.; Lee, J.; Park, C. Korea University School of Medicine, Seoul, South Korea

Address correspondence to S. Jang (tiancai0505@naver.com)

Objective: It has been generally recognized that fatty liver can often be seen among the obese population. This study was conducted to evaluate the association between fatty liver grading and abdominal fat volume by using the hepatic attenuation value.

Materials and Methods: A total of 310 patients underwent fat CT scans who visited our obesity clinic over the last six months. Liver and spleen attenuation measurements were taken with the region of interest (ROI) of the liver in three areas and two areas from the spleen. The hepatic attenuation index (HAI) is measured as follows: (1) hepatic parenchymal attenuation (CTLP), (2) liver to spleen attenuation ratio (LS ratio), and (3) difference between hepatic and splenic attenuation (LSdif). Abdominal fat volume was measured using a 10 mm slice CT scan at the aortic bifurcation level and was automatically calculated with a workstation (Philips Healthcare, Bothell, WA). Abdominal fat was classified into total fat (TF), visceral fat (VF) and subcutaneous fat (SF). We used the bivariate correlation method to assess the correlation between HAI and (1) total fat (TF), (2) visceral fat (VF), and (3) subcutaneous fat (SF), respectively.

Results: There was significant negative correlation between CTLP, LS ratio and LSdif with total abdominal fat, visceral fat, and subcutaneous fat. The CTLP showed strong negative correlation with TF and VF (r = –0.412 and –0.414, respectively p = 0.000). The correlation between CTLP and SF was less significant than VF or TF (r = –0.265, p = 0.000)

Conclusion: Fat infiltration of the liver is correlated with the amount of total abdominal fat and visceral fat was found to be more strongly associated with fatty liver than subcutaneous fat.


  • * Will present paper

2:00 pm

237. Density Change in Colon Cancer Hepatic Metastasis Treated with Yttrium-90 on MDCT: Does it Predict PET Activity?

Tochetto, S.*; Rezai, P.; Yaghmai, V. Northwestern University, Chicago, IL

Address correspondence to V. Yaghmai (v-yaghmai@northwestern.edu)

Objective: The objective was to evaluate the correlation between PET activity and change in density of hepatic metastasis of colon cancer treated with Yttrium-90 radioembolization.

Materials and Methods: Fifty-eight colon cancer hepatic metastases in 22 patients were evaluated pre- and post-treatment with CT and PET scans. The average time interval between pre- and post-treatment scans was 44.6 ± 12.1 days (range 29 to 77 days). Pre- and post-treatment PET evaluations were compared and the results were qualitatively categorized in three groups: (A) No change; (B) Mild/moderate interval improvement; (C) Complete interval improvement. The mean density and volumetric measurements of the hepatic lesions were calculated. The percentage of change in density and volumetric measurements for the three groups was calculated comparing pre- and post-treatment measurements. p < 0.05 was considered significant.

Results: The percentage of change in density correlated with PET evaluation of tumor response to Yttrium-90 radioembolization treatment (r = 0.65). The percentage of change in density between pre- and post-treated lesions was 2.16% for Group A (SE ± 2.84%, n = 13); –9.80% for Group B (standard error ± 2.56%, n = 33) and -23.53% for Group C (SE ± 7.88, n = 12). A significant difference was observed among the groups A, B and C (ANOVA, p < 0.001). The percentage of change in volume and RECIST measurement did not correlate with PET and no significant difference was observed among the groups A, B and C (p > 0.05).

Conclusion: The change in density of colon cancer hepatic metastasis treated with Yttrium-90 correlates with PET activity and may be a better predictor of monitoring therapeutic response than volumetric and RECIST measurements.


  • * Will present paper

2:10 pm

238. Imaging Findings of Hepatic Focal Nodular Hyperplasia in Gadoxetic Acid vs. Gadobenate Dimeglumine–Enhanced MR: Preliminary Results

Taibbi, A.*; Bartolotta, T.; Putignano, L.; Galia, M.; Lo Re, G.; Toro, A.; Midiri, M.; Lagalla, R. University Hospital, Palermo, Italy

Address correspondence to A. Taibbi (taibbiadele@hotmail.com)

Objective: The objective was to evaluate imaging features of focal nodular hyperplasia (FNH) at MR performed with two different hepatocellular-specific gadolinium-based contrast agents.

Materials and Methods: Eleven patients (eight women, three men; mean age 37.8 years) with 14 confirmed FNHs (mean size 2.4 cm) underwent MR (1.5T) before (T2-weighted fast spin-echo sequences without and with fat-saturation and T1-weighted GRE in phase, out-of phase and fat-saturation sequences) and after (T1-weighted gradient echo volumetric fat-saturation sequences) intravenous bolus administration of both hepatospecific contrast agents: gadobenate dimeglumine (Gd-BOPTA): 0.1 mmol/kg and gadoxetic acid (Gd-EOB-DTPA): 0.025 mmol/kg. Arterial, portal-venous, equilibrium and hepatospecific phases—this latter at time interval of 1–2 hours (Gd-BOPTA) and 30-40 minutes (Gd-EOB-DTPA) were acquired. Contrast-enhancement patterns, central scar and, in the hepatospecific phase (HP), signal intensity (SI) of FNH compared with those of the surrounding liver parenchyma were evaluated by two blinded readers in consensus.

Results: All 14 FNHs (100%) showed strong arterial enhancement appearing hyperintense during portal-venous and equilibrium phases after Gd-BOPTA and Gd-EOB-DTPA injection without any statistically significant difference (p > 0.05). In the HP all lesions retained both contrast-media but five out of 14 FNHs revealed inhomogeneous aspect with hypointense central oval area different from central scar. This latter was evident in 10/14 (71.4%) patients studied with either Gd-BOPTA or Gd-EOB-DTPA. In the HP, SI mean values were not statistically significant different (Gd-BOPTA: 1,301.9 and Gd-EOB-DTPA: 1,415.4) (p > 0.05).

Conclusion: Imaging findings of FNHs evaluated with the two different hepatospecific MR contrast-agents are highly similar but Gd-EOB-DTPA allows for reduction in total time for the MR examination.


  • * Will present paper

2:20 pm

239. Focal Liver Lesions in Patients with Cavernous Transformation of the Portal Vein: MR Imaging Features

Galluzzo, A.1*; Brancatelli, G.1; Plessier, A.2; Midiri, M.1; Condat, B.2; Lagalla, R.1; Vilgrain, V.3 1. University of Palermo, Palermo, Italy; 2. Service d'hépatologie, Hopital Beaujon, AP-HP, Clichy, France; 3. Service de Radiologie, Hopital Beaujon, AP-HP, University of Paris, Clichy, France

Address correspondence to A. Galluzzo (annagalluzzo@gmail.com)

Objective: The objective was to evaluate retrospectively MR imaging findings of focal liver lesions in patients with cavernous transformation of the portal vein.

Materials and Methods: Clinical, pathological and MR imaging findings were retrospectively reviewed in 58 patients with cavernous transformation of the portal vein. Two radiologists evaluated the presence of focal liver lesions. In patients with hepatic lesions, size, number and MR features (signal and enhancement of each lesion) were assessed. Diagnosis of the liver lesions was based on typical imaging or liver biopsy.

Results: Eighteen patients showed focal liver lesions and 12 had focal nodular hyperplasia (FNH) like lesions (20.6% of all patients). In ten patients was found a solitary lesion: five FNH-like lesions and five hemangiomas. Eight patients had multiple lesions: five with FNH-like lesions, one with hemangiomas and two with the association respectively of FNH-like lesions and liver adenoma and FNH-like lesions and hemangioma. The diagnosis of FNH-like lesion and liver adenoma were confirmed at pathology in three patients. The mean size of FNH-like lesions was 12.9 mm (range: 5–42 mm). The mean delay time between diagnosis of the portal vein thrombosis and the detection of FNH-like lesions was 4.3 years (range: 0–13 years).

Conclusion: Most lesions observed in patients with cavernous transformation of the portal vein are FNH-like lesions. These lesions probably arise as a consequence of portal inflow reduction.


  • * Will present paper

2:30 pm

240. Comparison of Multislice CT and Gadolinium-Enhanced MRI for Detection of Small (<3 cm) Hepatic Lesions in Patients With Chronic Liver Disease

Wable, S.; Kaplan, M.; Wang, P.*; Naqvi, S. Albert Einstein Medical Center, Philadelphia, PA

Address correspondence to M. Kaplan (kaplanmc@einstein.edu)

Objective: The objective was to compare multislice CT and MRI for the detection of small hepatic lesions in patients with chronic liver disease.

Materials and Methods: Approximately 2,484 triple phase CT and MRI examinations were retrospectively reviewed from 2005. This yielded a study population of 30 patients with cirrhosis or chronic hepatitis who had lesions less than 3 cm and both MRI and multislice CT within a period of eight weeks. Eleven patients had CT with a 64-slice CT scanner and 19 with a 16-slice CT scanner. All but two patients had both MRI and CT within six weeks. All but six patients had no follow up or follow up for less than six months. The other patients had either follow up greater than six months to two years, had histological diagnosis or were treated with radiofrequency ablation or chemoembolization for presumed hepatocellular carcinoma based on their characteristic imaging findings.

Results: A total of 32 lesions were seen on CT and 43 on MRI. Of the 11 additional lesions on MRI, two were additional foci of hepatocellular carcinoma (HCC), three were stable for two years, one was stable for eight months, three were subcentimeter lesions and thought to be foci of arterioportal shunting and two were characteristic of hemangiomas. Of the 32 lesions seen on both MRI and CT, three lesions were better characterized on MRI, one was characteristic for a hemangioma, one represented a portosystemic venous shunt, and one was thought to be a regenerative nodule stable for two years. The two lesions better characterized on CT demonstrated interval growth on three month followup. Of the six patients who had no follow up or follow up less than six months, two had imaging characteristics for benign lesions and one was atypical on both CT and MR and has remained stable for two months. In two, both CT and MRI showed lesions suspicious for HCC. In the sixth patient, both CT and MR detected the radiofrequency ablated nonviable lesions.

Conclusion: MRI appears to have only a slight advantage over multislice CT in lesion detection and characterization. Although more lesions were seen on MRI, only two of them represented additional foci of HCC. In patients unable to tolerate long imaging times and MRI, triple phase CT examination would be an adequate study. Based on the above data multislice CT will have a similar detection rate for small hepatic lesions as MRI. More pseudolesions are seen in MRI and familiarity with these varied patterns of enhancement is essential for interpretation.


  • * Will present paper

2:40 pm

241. Differentiating a Benign From Malignant Liver Lesion in the Pretransplant Cirrhotic Liver on Triphasic MDCT: When to Worry, When to Relax?

Liu, Y.*; Shin, L.; Kamaya, A.; Jeffrey, R. Stanford University, Stanford, CA

Address correspondence to Y. Liu (yliu@stanford.edu)

Objective: Hypervascular lesions are frequently seen in end-stage cirrhotic livers. Differentiating hepatocellular carcinoma (HCC) from benign entities such as regenerative nodules or arterial-portal shunts is often difficult yet of utmost importance. Candidacy for transplantation according to the Milan criteria and priority according to the United Network for Organ Sharing (UNOS) are determined based solely on hypervascularity for diagnosis of HCC without biopsy confirmation. This study aims to identify the sensitivity and positive predictive value of arterial phase imaging in detecting HCC and the added value of portal venous and delayed phase imaging. Other characteristics, including morphology and size, were also considered.

Materials and Methods: We reviewed all patients who underwent liver transplantation at our institution from 2003 to 2007 and identified 48 arterially enhancing lesions in 25 patients within six months of transplantation. Previously chemoembolized and lesions <1 cm were excluded. Enhancement profiles and other characteristics were recorded for each hypervascular lesion. All explanted livers were serially sectioned to 10 mm slices and examined for focal lesions.

Results: The mean lesion size was 1.8 cm. Based on pathologic correlation of the explanted liver, 18 active (i.e., not chemoembolized) HCCs were identified. 14 of the 18 lesions were hypervascular on arterial phase, resulting in a 77.8% sensitivity and 29.2% positive predictive value. Three of the four lesions that were not hypervascular were well differentiated HCCs. The presence of delayed wash-out provided a much higher sensitivity (93%) and specificity (97%) compared to hypervascularity alone. Two additional lesions showed partial wash-out on the delayed phase which proved to be a mixed HCC/cholangiocarcinoma and a focally-hemorrhagic HCC. There is also significant correlation between lesion size and HCC, as well as between lesion morphology and HCC.

Conclusion: Although UNOS uses hypervascularity as criteria for diagnosis of HCC without biopsy confirmation, the presence of wash-out on delayed images has a higher sensitivity and specificity in diagnosis of HCC and may be a better indicator of malignancy than arterial hypervascularity alone which only provided 78% sensitivity. Partial wash-out may be signs of mixed malignancies or HCC complicated by hemorrhage.


  • * Will present paper

2:50 pm

242. Hepatic Lesions Imitating Neoplasms on Advanced Imaging (CT/MR/PET) – Experience in 74 Patients with Percutaneous Image-Guided Biopsy

Ma, X.*; Sahani, D.; Gervais, D.; Hahn, P.; Mueller, P. Massachusetts General Hospital, Boston, MA

Address correspondence to X. Ma (xma@partners.org)

Objective: The objective was to investigate hepatic lesions that mimic neoplasms on imaging from a cohort of 869 patients who underwent image-guided percutaneous biopsies.

Materials and Methods: From a cohort of 869 patients having had image-guided percutaneous focal liver lesion biopsy with both core biopsy and fine needle aspiration (FNA), (724 without liver cirrhosis and 145 with cirrhosis), 74 lesions were considered as pseudolesions based on the histopathology results, lesion stability/resolution on imaging follow-up more than six months or surgery diagnosis. Preprocedure image studies done within three months were reviewed for the reasons of biopsy. The prevalence of pseudolesions in cirrhotic and noncirrhotic liver and incidence of different lesion sizes (≤3cm vs. >3cm) was assessed for statistical significance using Chi-square test. Data was also reviewed to study lesion characteristics on imaging (CT/MRI/PET).

Results: Seventy-four (8.5%) of 968 lesions biopsied were considered pseudolesions. Lesion biopsy was undertaken due to suspicious imaging appearance/enhancing features on MRI (n = 37), CT (n = 32), or increased FDG uptake on PET (n = 5). Histopathology revealed 23 (31%) inflammatory changes, 21 cirrhotic nodules, 20 focal steatosis, five regenerative changes and five perfusion abnormalities. On six months follow up imaging, 58 lesions were stable, whereas 15 disappeared and one patient underwent resection. 53/74 (71.6%) were in the 724 noncirrhotics, whereas 21/74 (28.4%) were found in the 145 cirrhotics (p = 0.0048). Fifty-four (73%) pseudolesions measured <3cm and 20 (27%) in lesion size >3cm. (p < 0.001) Small size (<3cm), poor conspicuity on delayed phase images, lack of T2 signal abnormalities, or mild FDG-uptake were features discovered in these lesions.

Conclusion: Pseudolesions are commonly encountered on liver imaging studies, more frequently in cirrhotics and they can present suspicious features. However, small lesion size and lack of visualization on delayed phase or T2-weighted images should raise the possibility of a benign diagnosis. A second imaging test or follow up can be a reasonable alternate as the majority disappears or remain stable.


  • * Will present paper

3:00 pm

243. Acoustic Radiation Force Impulse Technology and Fibroscan in the Evaluation of Liver Diseases

Clevert, D.1*; Zachoval, R.2; Reiser, M.1 1. Department of Clinical Radiology, University of Munich-Grosshadern Campus, Munich, Germany; 2. Internal Medicine II., Grosshadern Campus, Munich, Germany

Address correspondence to D. Clevert (Dirk.Clevert@med.uni-muenchen.de)

Objective: Fibroscan is an accepted method in the detection of fibrotic and cirrhotic changes of the liver. The aim of our study was to compare this method with the new acoustic radiation force impulse (ARFI)-technology in patients with liver diseases.

Materials and Methods: Twenty-five healthy patients (group 1) age 25 to 37 years were examined with the fibroscan and with ARFI-technology. In addition, 28 patients age 48–77 years with histological proven liver cirrhosis by hepatitis C were examined (group 2). The examination time and the stiffness of the liver were measured. In addition, the healthy probands were examined through subcostal and intercostal windows with the ARFI technology.

Results: Fibroscan and ARFI showed a good correlation in the 25 healthy probands as well as in the 28 patients with histologically proven liver cirrhosis. The velocity of ARFI was between 0.7–4.89 m/s and the stiffness of Fibroscan was between 3.7–75 kPa. There were no significant differences in the velocity between the subcostal and intercostal measurement using the ARFI technology. The mean examination time for Fibroscan/ARFI was 5.2/1.7 minutes in group one and 6.8/3.1 minutes in group two. Due to ARFI one unknown hepatocellular carcinoma was found.

Conclusion: The measured data correlated well between the ARFI technology and Fibroscan in both groups. The examination time with ARFI can be reduced up to 67% in comparison to Fibroscan. The subcostal ARFI examination has advantages for patients with considerable ascites. Due to the principle of ARFI, additional information is available with impact in the treatment of the patients.


  • * Will present paper

3:10 pm

244. Characterization of Hypoechoic Focal Hepatic Lesions in Patients with Fatty Liver: Diagnostic Performance and Confidence of Contrast-Enhanced Ultrasound

Bartolotta, T.*; Taibbi, A.; Galia, M.; La Grutta, L.; Malizia, G.; Midiri, M.; Lagalla, R. University Hospital, Palermo, Italy

Address correspondence to A. Taibbi (taibbiadele@hotmail.com)

Objective: The objective was to assess the diagnostic performance of contrast-enhanced ultrasound (CEUS) in the characterization of hypoechoic focal hepatic lesions (HFHLs) in patients with fatty liver (FL) in comparison with baseline ultrasound (US).

Materials and Methods: A total of 105 consecutive patients with 105 HFHLs (52 malignant and 53 benign; mean size: 2.8 cm) in FL underwent CEUS after sulphur hexafluoride (Bracco, Milan, Italy) administration. Two blinded readers independently reviewed baseline US and CEUS scans classifying each lesion as malignant or benign on a five-point scale of confidence by means of definite diagnostic criteria. Moreover they recorded if further imaging was needed for lesion characterization. Sensitivity, specificity, areas under the receiver operating characteristic (ROC) curve (Az) and interobserver agreement were calculated.

Results: Diagnostic confidence improved after reviewing CEUS scans for both readers (Az = 0.706 and 0.999, 0.665 and 0.990 at baseline US and CEUS, respectively: p < 0.0001). Inter-reader agreement increased (weighted k = 0.748 at baseline US vs. 0.882 at CEUS). For both readers, after CEUS correctly characterized lesions increased (from 27/105 [27.5%] to 94/105 [89.5%] and from 19/105 [18.1%] to 93/105 [88.6%], respectively: p < 0.0001) and the need for further imaging decreased (from 93/105 [88.6%] to 26/105 [24.8%] and from 96/105 [91.4%] to 40/105 [38.1%], respectively, (p < 0.0001).

Conclusion: CEUS improves the diagnostic performance of radiologists in the characterization of HFHLs arising in FL reducing the need for further radiological work-up.


  • * Will present paper

3:20 pm

245. Role of Contrast-Enhanced Ultrasound in the Characterization of Incidentally Discovered Focal Liver Lesions

Bartolotta, T.*; Taibbi, A.; Galia, M.; Lo Re, G.; La Grutta, L.; Midiri, M.; Lagalla, R. University Hospital, Palermo, Italy

Address correspondence to A. Taibbi (taibbiadele@hotmail.com)

Objective: The objective was to assess the role of contrast-enhanced ultrasound (CEUS) to characterize incidentally discovered focal hepatic lesions (IDFHLs).

Materials and Methods: A total of 155 patients (112 women, 43 men; age range: 18-77 years, mean: 48.2 years) with 155 IDFHLs (size range: 0.5-13.1 cm; mean: 3.1 cm) at baseline ultrasound (US) underwent CEUS after sulphur hexafluoride (Bracco, Milan, Italy) administration. Standard of reference was histology (n = 3: two focal nodular hyperplasia [FNH], one hepatocellular adenomas), core-biopsy (n = 21: two hepatocellular adenomas, 16 FNHs, one intrahepatic extramedullary hematopoiesis nodule, one solitary necrotic nodule, one inflammatory pseudotumor) and typical CT and/or MRI findings (n = 131: 86 hemangiomas, 28 FNHs, one echinococcus cyst, 16 hyper/hyposteatosis areas). Two blinded readers independently reviewed baseline US and CEUS scans and classified each lesion as malignant or benign on a five-point scale of confidence and recorded if further imaging work-up was needed. Sensitivity, specificity, areas under the receiver operating characteristic (ROC) curve (Az) and interobserver agreement were calculated.

Results: Diagnostic confidence improved after reviewing CEUS scans for both readers (Az = 0.606 and 0.899, 0.565 and 0.890 at baseline US and CEUS, respectively: p < 0.0001). Inter-reader agreement increased (weighted k = 0.648 at baseline US vs. 0.782 at CEUS). For both readers, after CEUS correctly characterized lesions increased (from 44/155 [28.4%] to 138/155 [89%] and from 32/155 [20.6%] to 135/155 [87.1%], respectively, (p < 0.0001) and the need for further imaging decreased (from 132/155 [85.2%] to 48/155 [30.1%] and from 140/155 [90.3%] to 52/155 [33.5%], respectively, (p < 0.0001).

Conclusion: CEUS improves the diagnostic performance of radiologists in the characterization of IDFHLs and reduces the need for further imaging work-up.


  • * Will present paper

| Table of Contents

Navigate This Article

  1. Top

This Article

  1. AJR May 2009 vol. 192 no. 5 Supplement A70-A73

    Navigate This Article

    1. Top