Small (≤1-cm) hepatocellular carcinoma: diagnostic performance and imaging features at gadoxetic acid-enhanced MR imaging.

Original Research n Gastrointestinal

Imaging

Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

Small (1-cm) Hepatocellular Carcinoma: Diagnostic Performance and Imaging Features at Gadoxetic Acid–enhanced MR Imaging1 Mi Hye Yu, MD2 Jung Hoon Kim, MD Jeong-Hee Yoon, MD Hyo-Cheol Kim, MD Jin Wook Chung, MD Joon Koo Han, MD Byung-Ihn Choi, MD

Purpose:

To assess diagnostic performance and imaging features of gadoxetic acid–enhanced magnetic resonance (MR) imaging in small (1-cm) hepatocellular carcinoma (HCC) detection in patients with chronic liver disease.

Materials and Methods:

The institutional review board approved this retrospective study and waived informed consent. Sixty patients (56 men, four women; mean age, 60.1 years) with HCC (146 lesions; 70 . 1 cm, 76  1 cm) underwent gadoxetic acid– enhanced MR imaging. HCC was confirmed at surgical resection (72 lesions; 30 . 1 cm, 42  1 cm) or by showing interval growth with typical enhancement patterns at follow-up dynamic computed tomography or MR imaging (74 lesions; 40 . 1 cm, 34  1 cm). Two radiologists assessed MR imaging features and graded likelihood of HCC with a five-point confidence scale. Jackknife alternative free-response receiver operating characteristic (JAFROC) method was used.

Results:

Mean JAFROC figure of merit for small HCC was 0.717; that for large (.1-cm) HCC was 0.973 with substantial agreement (k = 0.676). Mean sensitivity and positive predictive value (PPV) were 46.0% (70 of 152) and 48.3% (70 of 145) for small HCC versus 95.0% (133 of 140) and 78.2% (133 of 170) for large HCC, respectively. Eleven of 76 small HCCs (14%) were not seen on MR images, even after careful investigation. MR imaging features of small HCC included arterial enhancement (79%, 60 of 76), hypointensity on hepatobiliary phase (HBP) images (68%, 52 of 76), washout on 3-minute delayed phase images (50%, 38 of 76), hyperintensity on T2-weighted images (43%, 33 of 76), hypointensity on T1-weighted images (32%, 24 of 76), and restriction on diffusion-weighted images (28%, 20 of 72). Arterial enhancement and washout on 3-minute delayed phase images or hypointensity on HBP images occurred in 66% of small HCCs (50 of 76).

Conclusion:

Diagnostic performance of gadoxetic acid–enhanced MR imaging for small HCC detection is still low, with mean sensitivity of 46.0% (70 of 152) and mean PPV of 48.3% (70 of 145). By adding hypointensity on HBP images as washout, diagnostic performance for small HCC detection can be improved.

1

From the Department of Radiology, Seoul National University Hospital, 101 Daehangno, Jongno-gu, Seoul 110-744, South Korea (M.H.Y., J.H.K., J.H.Y., H.C.K., J.W.C., J.K.H., B.I.C.); and Institute of Radiation Medicine, Seoul National University College of Medicine, Seoul, South Korea (J.W.C., J.K.H., B.I.C.). Received August 23, 2013; revision requested October 17; revision received November 12; accepted December 10; final version accepted December 26. Address correspondence to J.H.K. (e-mail: jhkim2008@ gmail.com). Current address: Department of Radiology, Konkuk University Hospital, Konkuk University School of Medicine, Seoul, South Korea.

2

RSNA, 2014

q

Online supplemental material is available for this article.

RSNA, 2014

q

748

radiology.rsna.org n Radiology: Volume 271: Number 3—June 2014

GASTROINTESTINAL IMAGING: Gadoxetic Acid–enhanced MR Imaging of Small Hepatocellular Carcinoma

H

epatocellular carcinoma (HCC) is the most common primary liver cancer (1,2). Worldwide, the HCC surveillance program is well established for high-risk patients. There are many therapeutic options for HCC, including surgical resection, systemic chemotherapy, and local-regional therapies, such as radiofrequency ablation, percutaneous ethanol injection therapy, and transarterial chemoembolization (TACE). Despite the many treatment options, the highest likelihood for a cure occurs in the early stage of small HCC. There are studies indicating that the smaller the HCC, the less likely there is to be microscopic vascular invasion (3,4). In addition, the smaller the HCC, the more likely it is that local ablation will be complete (5,6). It is, therefore, important to assign an early diagnosis of HCC when the tumor is still as small as possible. However, there are also many small, benign nodules, including cirrhosis-related nodules and arterioportal shunts, which can mimic

Advances in Knowledge nn The diagnostic performance of gadoxetic acid–enhanced MR imaging for detection of small (1cm) hepatocellular carcinoma (HCC) is still low and currently has a mean sensitivity of 46.0% (70 of 152 lesions) and positive predictive value of 48.3% (70 of 145 lesions). nn The arterial enhancement and hypointensity on gadoxetic acid– enhanced MR images acquired in the hepatobiliary phase (HBP) are the most common imaging features of small (1-cm) HCC (seen in 79% of lesions [60 of 76] and 68% of lesions [52 of 76], respectively). nn When we applied the enhancement pattern by using the arterial enhancement and washout in the 3-minute delayed phase or hypointensity in the HBP for small (1-cm) HCC, 66% of small (1-cm) HCCs (50 of 76 lesions) met the diagnostic criteria for HCC.

small HCC in patients with cirrhosis. Therefore, both the early detection and the correct diagnosis of small HCC are important. The imaging criteria for the diagnosis of HCC have been established by the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases (AASLD); they detail the characteristic contrast material enhancement pattern of HCC, including hypervascularity in the arterial phase and washout in the portal or delayed phases of dynamic imaging (7,8). In current clinical practice guidelines, however, the diagnosis of HCC is achieved only for nodules larger than 1 cm in diameter that show the typical vascular enhancement pattern of HCC at dynamic imaging, owing to the low sensitivity of computed tomography (CT) and magnetic resonance (MR) imaging for small HCCs 1 cm in diameter or smaller (9–12). Furthermore, the European Association for the Study of the Liver and the AASLD do not address the diagnostic role of new contrast agents, such as hepatocytespecific agents. The hepatocyte-specific MR imaging contrast agents gadoxetic acid (Primovist; Bayer Healthcare, Berlin, Germany) and gadobenate dimeglumine (MultiHance; Bracco, Milan, Italy) have been used widely and have shown their superior effectiveness for both detection and characterization of focal liver lesions (13–17). This could be related to the characteristic dual properties of these contrast media, which allow dynamic perfusion imaging and also allow evaluation of delayed hepatocyte uptake and biliary excretion. In particular, gadoxetic acid–enhanced MR imaging usually allows detection of small liver malignancies in the hepatobiliary phase (HBP) because

Implication for Patient Care nn Washout defined as hypointensity in the 3-minute delayed phase or the HBP is the most useful criterion for detecting small (1-cm) HCCs.

Radiology: Volume 271: Number 3—June 2014 n radiology.rsna.org

Yu et al

of the higher contrast between a focal liver lesion and the background liver parenchyma (18,19). In clinical practice, however, the detection and characterization of small hepatic focal lesions are challenging because they frequently show atypical enhancement patterns at dynamic imaging. In fact, it is well documented that the earliest detectable HCC nodules rarely show typical radiologic changes because the reduction in the portal supply to the nodule may have occurred, while arterial hypervascularization has not fully developed and a decrease in the portal supply may not be detectable at imaging (20). Until now, most previous studies regarding gadoxetic acid–enhanced MR imaging in cirrhotic patients have not focused on small hepatic nodules 1 cm in diameter or smaller. Therefore, for a more accurate noninvasive diagnosis of small (1-cm) HCC, we first need to know the imaging features of small HCC tumors on gadoxetic acid– enhanced MR images, as well as the diagnostic performance for detection of small HCC at this stage. Therefore, the purpose of this study was to

Published online before print 10.1148/radiol.14131996 Content codes: Radiology 2014; 271:748–760 Abbreviations: AASLD = American Association for the Study of Liver Diseases DW = diffusion weighted HBP = hepatobiliary phase HCC = hepatocellular carcinoma JAFROC = jackknife alternative free-response receiver operating characteristic PPV = positive predictive value TACE = transarterial chemoembolization Author contributions: Guarantors of integrity of entire study, M.H.Y., J.H.K., J.H.Y.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, M.H.Y., J.H.K., H.C.K., J.K.H.; clinical studies, M.H.Y., J.H.K., J.H.Y., H.C.K., J.W.C., J.K.H.; experimental studies, M.H.Y.; statistical analysis, M.H.Y.; and manuscript editing, M.H.Y., J.H.K., H.C.K., J.W.C., B.I.C. Conflicts of interest are listed at the end of this article.

749


Yu et al

Figure 1

Figure 1: Flowchart of the study population.

assess the diagnostic performance and imaging features of gadoxetic acid–enhanced MR imaging for the detection of small HCC in patients with chronic liver disease. We also investigated the role of HBP images in the detection of small HCC.

Materials and Methods Patient Selection and Standard of Reference This retrospective study was approved by the institutional review board at our hospital. The required informed consent was waived because of the retrospective study design. To identify small HCC, we used two standards of reference for small HCC. One standard of reference was based on the pathologic examination of surgical specimens (group 1), and the other was C-arm CT-based diagnostic imaging criteria, established through careful image review during a follow-up period as long as 2 years (group 2). 750

We searched the pathology and medical records at our institution from March 2009 to October 2012. We selected patients with pathologically proven small HCC. During this period, 498 consecutive patients underwent hepatic resection at our hospital. Among them, 52 patients had small HCC. Of these 52 patients, we excluded 25 for the following criteria: (a) gadoxetic acid–enhanced MR examination was not performed within 20 days before surgery (n = 7); (b) there were multiple HCCs—that is, more than 10 lesions in one patient and more than five in one segment (n = 7); (c) previous treatment with radiofrequency ablation or TACE was performed for small HCC (n = 9); and (d) MR images were obtained with poor image quality, owing to the respiration artifact (n = 2). In total, 27 patients with pathologically proven small HCC (n = 42) were included in group 1 (Fig 1). The mean time interval between MR examination and surgery 6 standard deviation was 7.93 days 6 5.90 (range, 1–19 days).

The surgeries included total hepatectomy for liver transplantation (n = 20), hemihepatectomy (n = 5), and tumorectomy (n = 2). We also established the standard of reference for small HCC by using C-arm CT. Since C-arm CT provides high spatial and contrast resolution, we selected C-arm CT-based diagnostic criteria for small HCC instead of dynamic CT or MR imaging–based diagnostic criteria. C-arm CT allows a section thickness and in-plane resolution of less than 1 mm because of the high frame rates, and the intraarterial bolus injection of contrast medium for C-arm CT provides better contrast resolution than CT or MR imaging with intravenous injection of contrast medium (21,22). Furthermore, radiologists confidentially determined the same location of the two nodules on the prior and follow-up C-arm CT images by referring to the hepatic artery branching pattern as a landmark because of its high spatial resolution. During the study period detailed previously, 3166 patients suspected of



Yu et al

Figure 2

Figure 2: Images in a 62-year-old man with small hypervascular HCC. (a) At axial C-arm CT, a small subcentimeter-sized hyperenhancing nodule (arrow) is seen in the segment 7 dome. (b) Three months later at follow-up axial C-arm CT, there has been interval growth in the nodule (arrow), which is now more than 1 cm. On dynamic CT images obtained 3 months after the initial C-arm CT examination (in a), this nodule demonstrates arterial enhancement and portal venous washout (see Fig E1 [online]). Therefore, the previous subcentimetersized, hyperenhancing nodule at initial C-arm CT (in a) was diagnosed as a small HCC (group 2). (c–e) On axial MR images obtained at the time of initial C-arm CT (a), this small HCC (arrow) appears as a discrete nodule with arterial enhancement (c), with washout on the 3-minute delayed phase image (d) and hypointensity seen on the HBP image (e). Both of the reviewers considered this lesion to be an HCC. A large hepatic cyst (arrowhead) is partially imaged in the dome of the liver, near the small HCC (arrow).

having HCC underwent TACE at our institution. Among them, 2348 patients underwent C-arm CT immediately before TACE. Of these 2348 patients, 518 underwent C-arm CT twice or more. Of these 518 patients, 239 underwent gadoxetic acid–enhanced MR examination within 20 days before TACE. Through the careful imaging review by two attending abdominal radiologists (M.H.Y. and J.K.H., with 8 and 25 years of clinical experience in abdominal imaging, respectively), we searched subcentimeter-sized, hyperenhancing nodules seen at prior C-arm CT, which were observed to have grown at follow-up C-arm CT (mean time interval, 227.0

days 6 157.2; range, 61–646 days). If a nodule then satisfied all of the following three conditions, we concluded that this subcentimeter-sized, hyperenhancing nodule seen at prior C-arm CT was a small HCC: (a) The nodule showed interval growth of more than 1 cm at follow-up C-arm CT; (b) at the same time as the follow-up C-arm CT examination was conducted, the grown nodule showed the typical enhancement pattern of HCC—that is, arterial enhancement and portal washout on dynamic CT or MR images; (c) the nodule showed dense, compact iodized oil uptake (Lipiodol; Andre Guerbet, AulnaySous Bois, France) at CT performed


after TACE (23). By using this standard of reference, we found 34 small HCCs in 33 patients (group 2) (Figs 1, 2). The mean time interval between C-arm CT and MR examinations was 13.58 days 6 5.97 (range, 0–20 days). In total, 60 patients (mean age, 60.1 years 6 8.7; age range, 40–78 years) with 76 small HCCs were included in this study by using these two kinds of reference standards. There were 56 men (mean age, 60.0 years 6 8.7; age range, 40–78 years) and four women (mean age, 61.0 years 6 10.5; age range, 46–70 years). The large (.1-cm) HCCs were diagnosed according to the pathology findings or the noninvasive diagnostic 751


criteria of the AASLD guidelines (5). In the 60 included patients, there were 70 large HCCs in 48 patients. In group 1, 36 large HCCs were found at pathologic examination. However, we excluded six lesions because they had been treated previously with TACE. They showed more than 90% tumor necrosis, with residual sparse tumor cells seen at pathologic examination. There was no nodule in group 1 that met the AASLD guidelines but was not confirmed to be HCC at pathologic examination. Therefore, 30 large HCCs were included in group 1. In group 2, there were 40 large HCCs that showed a typical enhancement pattern (according to AASLD guidelines) on MR images and showed compact iodized oil uptake after TACE. In summary, 146 HCCs (76 HCCs  1 cm, 70 HCCs . 1 cm) were identified in the 60 study patients. In group 1, there were 42 small HCCs and 30 large HCCs; and in group 2, there were 34 small HCCs and 40 large HCCs. The mean number of lesions per patient was 2.44 6 1.1 (range, 1–6), and the distributions were as follows: There were seven patients with one lesion, 33 with two lesions, 13 with three lesions, three with four lesions, two with five lesions, and two with six lesions. We also searched our pathology and medical databases during the study period and selected patients without HCC as a control group. During this period, 209 consecutive patients underwent total hepatectomy for liver transplantation at our hospital for various reasons, including HCC, hepatic failure, and fulminant hepatitis. Among these patients, 161 had pathologically proven HCC, and the other 48 had no identifiable HCC at pathologic examination. We excluded 18 of 48 patients for the following reasons: (a) gadoxetic acid– enhanced MR examination was not performed (n = 4) or was performed at an outside hospital (n = 2) and (b) MR images were obtained with poor image quality, owing to the large amount of ascites (n = 9) or respiration artifacts (n = 3). Finally, 30 patients (mean age, 54.8 years 6 6.8; age range, 37–64 years) were included as a control group. There were 26 men (mean age, 54.8 years 6 752

7.2; age range, 37–64 years) and four women (mean age, 55.3 years 6 3.8; age range, 52–59 years).

MR Examination MR imaging was performed with either a 1.5-T (Signa HDx [n = 32] or Signa Excite [n = 4]; GE Medical Systems, Milwaukee, Wis) or a 3-T (Verio [n = 13] or Trio [n = 9], Siemens Medical Solutions, Erlangen, Germany; or Signa Excite [n = 2], GE Medical Systems) superconducting system by using either an eight-channel (Signa HDx and Signa Excite) or a 32-channel (Verio and Trio) phased-array coil. Our routine liver MR imaging protocol consisted of a breath-hold fat-saturated T2-weighted fast spinecho or turbo spin-echo sequence, a breath-hold T1-weighted dualecho (in-phase and opposed-phase) sequence, dynamic three-dimensional fat-saturated T1-weighted sequences, and free-breathing diffusion-weighted (DW) imaging by using a single-shot echo-planar imaging sequence. Before and after administration of contrast media, dynamic three-dimensional fatsaturated T1-weighted sequences with a spatial resolution of 1.2–1.7 mm and a section thickness of 3–6 mm were performed. The contrast agent was injected intravenously as a bolus of 10 mL and at a rate of 1 or 1.5 mL/ sec by using a power injector (Spectris Solaris EP; Medrad, Warrendale, Pa), followed by a 25-mL saline flush. After contrast material administration, imaging delay times were determined by using real-time MR imaging fluoroscopic monitoring. Images in the arterial phase were acquired 7 seconds after the contrast medium had arrived at the thoracic aorta, and images in the portal venous phase, delayed phase, and HBP were subsequently acquired 60 seconds, 3 minutes, and 20 minutes, respectively, after the beginning of contrast medium injection. DW imaging with simultaneous respiratory triggering was performed before 20 minutes in the delayed phase. For each patient, the repetition time was matched to the length of the respiratory cycle; every patient had b

Yu et al

values of 0 sec/mm2 and 500 sec/mm2. Among the 60 patients, DW images were available for 56 patients with 138 HCCs (.1 cm, n = 66; 1 cm, n = 72). All sequences were performed in the axial plane. The detailed imaging parameters of the MR equipment used are summarized in Table 1.

TACE Technique and C-Arm CT Acquisition TACE was performed in an interventional procedure room equipped with a commercially available digital subtraction angiography unit (AXIOM Artis dTA/VB30; Siemens). Two clinically experienced interventional radiologists (H.C.K., with 5 years of clinical practice; and J.W.C., with 18 years of clinical practice) performed all of the angiographic examinations. Before chemotherapeutic agent infusion, a single series of threedimensional, rotational, C-arm angiographic images of the common or proper hepatic artery was obtained for 8 seconds during a single breath hold and with 211° of circular trajectory. The contrast medium (Pamiray 300; Dongkook Pharmaceutical, Seoul, South Korea) was injected by using a power injector at a flow rate of 2–4 mL/sec for 12 seconds. The images were then obtained 4 seconds after injection. The parameters of C-arm CT were as follows: 0.5° increment, 512 3 512 matrix in projections, 211° total angle and approximately 26° rotation per second, a system dose of approximately 0.36 mGy per frame, and a total of 419 projections acquired. The acquired images were transferred to a commercially available, dedicated workstation (Leonardo with Dyna CT; Siemens Healthcare), where images were reconstructed within 1 minute with a reconstructed section thickness of 0.4 mm. If there was a variation in the hepatic arterial anatomy, such as the left hepatic artery arising from the left gastric artery and the right hepatic artery coming from the superior mesenteric artery, three-dimensional rotational C-arm angiographic images of the left hepatic artery and the right hepatic artery were obtained separately. Therefore, the entire liver was


128 3 128 144 3 144 136 3 136 128 3 128 192 3 192 400 3 400 380 3 380 380 3 380 380 3 380 350 3 350 7 7 7 7 8 6000/60.2 90 4850/68 90 5000/52 180 5000/52 180 4500/53 90 320 3 256 320 3 224 384 3 250 384 3 250 320 3 256 360 3 360 380 3 380 380 3 309 380 3 309 350 3 350 6 4.8 3.5 3 5.2 12 12 11 11 15 4.6/2.2 5/2.2 3.4/1.2 3.6/1.3 4.2/1.9 256 3 256 320 3 224 320 3 256 320 3 224 512 3 384


Yu et al

fully covered in the imaging range in all patients (24,25).

Signa Excite, GE Healthcare.

Verio, Siemens Healthcare

Trio, Siemens Healthcare.

Signa Excite, GE Healthcare.

†

‡

§

||

* Signa HDx, GE Healthcare.

Note.—TR/TE = repetition time (msec)/echo time (msec).

360 3 360 380 3 380 380 3 380 380 3 380 350 3 350 7 5 7 7 8 2000/84 90 7500/878 90 3000/91 140 3000/91 140 12587.1/100.7 90

MR Unit No. TR/TE

1 (1.5 T)* 2 (1.5 T)† 3 (3 T)‡ 4 (3 T)§ 5 (3 T)||

TR/TE

Section Flip Angle Thickness (degrees) (mm) Field of View Matrix Section Flip Angle Thickness (degrees) (mm) Field of View Matrix

Fat-saturated T2-weighted Sequence

MR Imaging Parameters

Table 1

Dynamic Three-dimensional Fat-saturated T1-weighted Sequence

TR/TE

DW Imaging

Section Flip Angle Thickness (degrees) (mm) Field of View Matrix


Image Analysis Two clinically experienced abdominal radiologists (J.H.K. and J.H.Y., with 16 and 7 years of clinical experience in abdominal imaging, respectively) independently analyzed the gadoxetic acid– enhanced liver MR images to assess diagnostic performance. They understood that the patients might have HCC smaller than 1 cm in diameter and that there was a control group, although they were blinded to the number, size, and location of the lesions. Each observer was asked to record the size and segmental location of the hepatic lesions they identified. Therefore, each reviewer (a) selected the image sequence that most clearly depicted the lesion and (b) measured the size of the lesion in a maximum diameter on the image displayed on the picture archiving and communication system in a one-by-one setting, with one image displayed on one monitor. When a patient had multiple lesions, the reviewers added the image number and special comments regarding each lesion to the review sheet to avoid confusion in the data analysis. On MR images, each reviewer assessed the signal intensities on T1- and T2-weighted images, the arterial enhancement and signal intensities on the 3-minute delayed phase images and HBP images, and the restriction seen on DW images. They then graded the possibility of HCC by using a fivepoint confidence scale for each lesion as follows: 1, definitely benign lesion; 2, probably benign lesion; 3, indeterminate lesion; 4, probably HCC; and 5, definitely HCC. The diagnostic criteria for HCC were based on the MR imaging features—that is, if a nodule showed arterial enhancement, washout on 3-minute delayed images, or hypointensity on HBP images, we rated it with a score of 5. If a nodule showed no definite arterial enhancement but showed hypointensity on the 3-minute delayed images or HBP images and an additional imaging feature among the hyperintensity on T2-weighted images or hyperintensity on DW images, it was rated with a 753


score of 4. If a nodule was only seen as hypointense on the HBP images or had only arterial enhancement with a triangular or irregular shape, it was given a rating of 1 or 2, according to subjective judgment. To evaluate the MR imaging features of small and large HCCs, two additional experienced attending radiologists (M.H.Y. and J.K.H., with 8 and 25 years of clinical experience in abdominal imaging, respectively), who did not participate in the diagnostic performance study, reviewed the MR images in consensus. In this review, the information regarding the location and size of each HCC was given to the reviewers. They assessed the signal intensities seen on T1- and T2-weighted images, arterial enhancement, signal intensities on the 3-minute delayed phase images and HBP images, and restriction seen on DW images. We also analyzed the false-positive lesions on MR images. A false-positive lesion was defined as a lesion detected as HCC on MR images but not confirmed to be HCC at pathologic examination or at C-arm CT-based follow-up imaging. All image reviews were performed by using picture archiving and communication system software (Maroview 5.4; Infinitt, Seoul, South Korea) on a computer workstation (XW6200; HewlettPackard, Palo Alto, Calif) with monitors that had a spatial resolution of 1600 3 1200 (Totoku, Tokyo, Japan).

Statistical Analysis To analyze the diagnostic performance of MR imaging for the detection of HCC, a jackknife alternative freeresponse receiver-operating-characteristic (JAFROC) analysis was performed by using JAFROC software (JAFROC, version 4.1; http://www. devchakraborty.com) (26,27). The mean diagnostic accuracy was calculated according to the mean figure of merit, which was defined as the probability that on normal images, a lesion is rated higher than the highest rated nonlesion seen on control MR images of the liver (28). The sensitivity and positive predictive value (PPV) for detecting HCC on a per-lesion basis and 754

Yu et al

Table 2 Diagnostic Performance of Gadoxetic Acid–enhanced MR Imaging for Detection of HCC: Per-Lesion Analysis Parameter

Small (1-cm) HCC (n = 76) Large (.1-cm) HCC (n = 70)

Figure of merit Reviewer 1 0.700 (0.634,0.767) Reviewer 2 0.733 (0.658,0.808) Mean 0.717 Sensitivity (%) Reviewer 1 42 (31.6, 53.3) [32/76] Reviewer 2 50 (39.0, 61.0) [38/76] Mean 46 (38.3, 54.0) [70/152] PPV (%) Reviewer 1 55 (42.5, 67.3) [32/58] Reviewer 2 44 (33.7, 54.2) [38/87] Mean 48 (40.3, 56.4) [70/145]

0.977 (0.951, 1.002) 0.968 (0.940, 0.997) 0.973

Overall (n = 146) 0.834 (0.789, 0.877) 0.847 (0.799, 0.893) 0.881

96 (88.1, 98.5) [67/70] 94 (86.2, 97.8) [66/70] 95 (90.0, 97.6) [133/140]

67.8 (59.9, 74.9) [99/146] 71.2 (63.4, 78.0) [104/146] 69.5 (64.0, 74.5) [203/292]

86 (76.5, 91.4) [67/78] 72 (61.8, 79.9) [66/92] 78 (71.5, 83.8) [133/170]

72.8 (64.8, 79.6) [99/136] 58.1 (50.8, 65.1) [104/179] 64.4 (59.0, 69.5) [203/315]

Note.—Numbers in parentheses are 95% confidence intervals, and numbers in brackets are raw data.

also on a per-patient basis were then calculated. In a per-patient analysis, the reviewer’s assessment was assumed as positive when at least one lesion was detected in the liver. The weighted k value was used to evaluate the interobserver agreement of the confidence scale regarding the possibility of HCC detected on MR images. The scale for the k coefficients for interobserver agreement was as follows: less than 0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; and 0.81–1.00, almost perfect. To evaluate statistical differences in the MR imaging features of small and large HCCs, multivariable logistic regression analysis was used. The independent t test was used for comparing the size of small HCCs in groups 1 and 2. The sensitivities and PPVs of the two groups were compared by using the Fisher exact test. P , .05 was considered to indicate a significant difference. Statistical analysis was performed by using SPSS version 19.0 software (SPSS, Chicago, Ill) and Medcalc version 10.4.0.0 software for Windows (MedCalc Software, Mariakerte, Belgium).

Results The mean size of all of the HCCs was 1.63 cm 6 2.0 (range, 0.2–12 cm). The

mean size of small HCCs was 0.71 cm 6 0.21 (range, 0.2–1.0 cm) and that of large HCCs was 2.38 cm 6 2.50 (range, 1.1–12 cm). The size of the small HCCs in group 1 (0.77 cm 6 0.21) was larger than that of the small HCCs in group 2 (0.63 cm 6 0.19) (P , .05). The JAFROC figure of merit, sensitivity, and PPV for detecting HCC on a per-lesion basis are summarized in Table 2. The figure of merit of small HCC was 0.700 and 0.733 for reviewer 1 and reviewer 2, respectively, while the figure of merit for large HCC was 0.977 and 0.968 for reviewer 1 and reviewer 2, respectively. The mean sensitivity and PPV for the detection of HCC were 46.0% (70 of 152) and 48.3% (70 of 145) in small HCC and 95.0% (133 of 140) and 78.2% (133 of 170) in large HCC, respectively. According to the patient group, the mean sensitivity and PPV for detection of small HCC were as follows: 51% (43 of 84) in group 1 versus 40% (27 of 68) in group 2 (P . .05) and 58% (43 of 74) in group 1 versus 37% (27 of 73) in group 2 (P = .017), respectively. In a per-patient analysis, the sensitivity and PPV for small HCC were 45% (27 of 60) and 77% (27 of 35) for reviewer 1 and 52% (31 of 60) and 64% (31 of 48) for reviewer 2, respectively (Table 3). The interobserver agreement regarding the possibility of HCC was substantial (k = 0.676).



Table 3 Diagnostic Performance of Gadoxetic Acid–enhanced MR Imaging for Detection of HCC: Per-Patient Analysis Parameter Sensitivity (%) Reviewer 1 Reviewer 2 PPV (%) Reviewer 1 Reviewer 2

Small (1-cm) HCC (n = 60)

Large (.1-cm) HCC (n = 48)

Overall (n = 60)

45 (33.1, 57.5) [27/60] 52 (39.3, 63.8) [31/60]

98 (89.1, 99.6) [47/48] 98 (89.1, 99.6) [47/48]

87 (75.8, 93.1) [52/60] 87 (75.8, 93.1) [52/60]

77 (61.0, 87.9) [27/35] 65 (50.4, 76.6) [31/48]

92 (81.5, 96.9) [47/51] 92 (81.5, 96.9) [47/51]

93 (83.0, 97.2) [52/56] 87 (75.8, 93.1) [52/60]

Note.—Numbers in parentheses are 95% confidence intervals, and numbers in brackets are raw data.

There were one large HCC and 36 small HCCs that were not identified by any reviewer on MR images. Among the 36 small HCCs, 11 were not seen on gadoxetic acid–enhanced MR images, even after careful investigation by two additional abdominal radiologists who were aware of the information regarding the location and size of the HCCs. Eight lesions were too small (mean, 4.75 mm 6 0.46) to be detected on MR images without information regarding the location and size of the lesions. Nine lesions were seen as nodules that showed only arterial enhancement (n = 6) (Fig 3) or hyperintensity on T2-weighted images without arterial enhancement (n = 3). The remaining eight lesions were thought to not have been detected because of the presence of accompanying multiple arterioportal shunts (n = 3) and because of the lesion locations (n = 5)—that is, three were in the right hepatic dome, one abutted the portal vein, and the other one was located near the right kidney. Table 4 summarizes the factor analysis of the HCCs that were missed by both reviewers. The two reviewers detected 37 and 75 false-positive lesions on MR images, respectively. For the small HCCs, reviewer 1 detected 26 of the false-positive lesions, and reviewer 2 detected 49 of these lesions. On MR images, most of them were hypervascular lesions, including small hemangiomas and nodular arterioportal shunts or dysplastic nodules (Fig E2 [online]).

The MR imaging features of the small and large HCCs are summarized in Table 5. The arterial enhancement (79% [60 of 76] of small HCCs and 96% [67 of 70] of large HCCs) and hypointensity seen on the HBP images (68% [52 of 76] of small HCCs and 94% [66 of 70] of large HCCs) were the most common MR findings in both large and small HCC. At multivariable logistic regression, hyperintensity on T2-weighted images and hyperintensity on DW images were significantly less frequent in small HCC than in large HCC (odds ratio = 0.3 and 0.4, respectively; P , .05). The typical enhancement pattern of HCC—that is, arterial enhancement and washout seen on the 3-minute delayed phase images—was seen in 43% (33 of 76) of small HCCs and 74% (52 of 70) of large HCCs. In addition, 36% (27 of 76) of small HCCs showed arterial enhancement and isointensity on 3-minute delayed phase images. Arterial enhancement and hypointensity on HBP images were seen in 62% (47 of 76) of small HCCs (Fig 4) and 90% (63 of 70) of large HCCs. The arterial enhancement and washout seen on the 3-minute delayed phase images or hypointensity on HBP images could be seen in 66% (50 of 76) of small HCCs and 91% (64 of 70) of large HCCs.

Discussion Our study results show that the diagnostic performance of gadoxetic acid– enhanced MR imaging for small HCC


Yu et al

detection was lower than that for large HCC (mean figure of merit, 0.717 vs 0.973). The mean sensitivity and PPV for small HCC detection on gadoxetic acid–enhanced MR images was 46.0% (70 of 152) and 48.3% (70 of 145), respectively. With regard to the MR imaging features, the arterial enhancement and hypointensity on the HBP images were the most common MR imaging features of small HCC. When we applied the enhancement pattern by using the arterial enhancement and washout on the 3-minute delayed phase images or hypointensity on the HBP images for small HCC, 66% of small HCCs (50 of 76) met the diagnostic criteria for HCC. According to some previous studies (10–12,29–31), the sensitivities of CT and MR imaging varied for detection of small HCC. However, MR imaging generally showed superior sensitivity to that of CT (range, 20%–46% with CT vs 38%–60% with MR imaging). Regarding gadoxetic acid–enhanced MR imaging, Akai et al (31) reported that the sensitivities of 64-detector CT and MR imaging for surgically proven small HCCs were both 62% (10 of 16) for observer 1; for observer 2, they were 56% (nine of 16) for multidetector CT versus 75% (12 of 16) for MR imaging. In another study performed by Baek et al (12), gadoxetic acid–enhanced MR imaging also showed a higher sensitivity for detection of small HCCs than did multidetector CT (64% [nine of 14] vs 14% [two of 14] for observer 1; and 43% [six of 14] vs 21% [three of 14] for observer 2, respectively). Park et al (32) determined that the combination of gadoxetic acid–enhanced MR imaging and DW imaging yielded better diagnostic performance for the detection of small HCC (area under the receiver operating curve value of 0.911 and mean sensitivity of 84.8% [140 of 165]) than did gadoxetic acid–enhanced MR imaging alone (area under the receiver operating curve value of 0.733 and mean sensitivity of 58.8% [97 of 165]). Compared with the previous study, the diagnostic performance of our study, which also involved the use of both gadoxetic acid–enhanced 755


Figure 3

Figure 3: Images in a 54-year-old man with HCC who underwent liver transplantation. He had two HCC nodules in liver segment 8. The sizes of two nodules were 2.4 cm and 0.8 cm (group 1). (a–c) Axial MR images show a well-defined mass (arrowhead) in the segment 8 dome of the liver with obvious arterial enhancement (a), washout on the 3-minute delayed image (b), and hypointensity on the HBP image (c). Another small arterial enhancing nodule (arrow in a) is identified near the large mass in segment 8 of the liver. However, the smaller nodule is not clearly identified on the 3-minute delayed phase image (b) or the HBP image (c). The two reviewers interpreted the larger mass as being an HCC; however, both reviewers considered the other, smaller lesion to be an arterioportal shunt. (d) On a photograph of the explanted liver specimen, a smaller HCC (arrow) is seen as a yellowish solid mass in segment 8. (The larger HCC is not shown.)

MR imaging and DW imaging, seems to be markedly lower. This discrepancy could be explained by the difference in the study populations—that is, they included a small number of patients with explanted livers, and they excluded patients who had undergone previous treatment for HCC. As Park et al had already commented on their study limitation, they included a small number 756

of patients (12 of 130) who underwent transplantation in their study, and this may have resulted in overestimation of the diagnostic performance of MR imaging by decreasing the number of false-negative lesions. In addition, the patients who underwent treatment previously, including TACE and radiofrequency ablation for HCC, may have affected the diagnostic performance of

Yu et al

MR imaging, because not only treated HCCs, with iodized oil uptake or parenchymal defect, but also parenchymal changes, including atrophy and ischemic changes, could make it difficult to detect newly appearing HCC. Furthermore, in general, operable patients have better liver function than those who are awaiting transplantation or repeat TACE. Impaired hepatic function decreases the uptake of contrast media, and it may also affect the detection and characterization of hepatic lesions. Regarding the results of the subgroup comparison, the mean sensitivity in group 1 was higher than that in group 2, but there was no significant difference (51% [43 of 84] in group 1 vs 40% [27 of 68] in group 2, P . .05). The mean PPV in group 1 was also higher than that in group 2 (58% [43 of 74] in group 1 vs 37% [27 of 73] in group 2, P , .05). This result may be related to the size of small HCCs, as the small HCCs in group 1 (0.77 cm 6 0.21) were larger than those in group 2 (0.63 cm 6 0.19) (P , .05). Furthermore, as we mentioned earlier, multiple previous TACE treatments may have also caused the lower diagnostic performance seen in group 2, as opposed to group 1. With regard to the MR imaging features of small HCC, 79% (60 of 76) of the small HCCs showed arterial enhancement in our study. Nevertheless, the typical enhancement pattern of HCC—that is, arterial enhancement and washout on the 3-minute delayed phase images—was seen in 43% of small HCCs (33 of 76), while arterial enhancement and isointensity on the 3-minute delayed phase images was seen in 36% of small HCCs (27 of 76). Although arterial enhancement and washout on the 3-minute delayed phase images are the most common and most important imaging characteristics of HCC, previous studies have shown that some small HCCs are seen only during the arterial phase and without early washout (9,33,34), as washout is less pronounced in smaller HCC than in larger HCC. van den Bos et al (35) also maintained that smaller HCC often showed more intense arterial enhancement and less pronounced



Table 4 Factor Analysis of HCCs Missed by Both Reviewers Presumed Factors Responsible for Missing Lesions on MR Images Corresponding lesions not seen, even with lesion information Lesion too small to detect without lesion information Lesion showed only arterial enhancement Hyperintensity on T2-weighted MR images without arterial enhancement Lesion not detected because of accompanying multiple arterioportal shunts Location made the tumor difficult to detect

Small (1-cm) HCC (n = 36)

Large (.1-cm) HCC (n = 1)

11 8 6 3

0 0 0 1

3

0

5

0

Table 5 MR Imaging Features of Small (1-cm) and Large (.1-cm) HCC MR Imaging Features Hyperintensity on T2-weighted images Hypointensity on T1-weighted images Arterial enhancement Washout on 3-minute delayed phase images Hypointensity on HBP images Hyperintensity on DW images*

Small HCC (n = 76)

Large HCC (n = 70)

Odds Ratio

95% Confidence Interval

P Value†

33 (43)

58 (83)

0.3

0.1, 0.8

.018

24 (32)

37 (53)

1.4

0.6, 3.3

.504

60 (79) 38 (50)

67 (96) 55 (79)

0.3 0.7

0.1, 1.8 0.3, 1.7

.208 .389

52 (68) 20 (28)

66 (94) 47 (71)

0.4 0.4

0.1, 1.7 0.1, 0.9

.224 .024

Note.—Numbers in parentheses are percentages. * Diffusion-weighted images were available for 138 lesions (56 patients)—that is, 72 small and 66 large HCCs. †

P value was obtained from multivariable logistic regression analysis of MR imaging features between small and large HCC.

delayed washout compared with larger HCC. Moreover, the hepatocyte uptake of gadoxetic acid in gadoxetic acid–enhanced MR imaging starts with the first pass and is certainly perceivable within the first 90 seconds (36). Therefore, the portal venous phase finding on gadoxetic acid–enhanced MR images is different from that of extracellular gadolinium-based contrast agents. In our study, arterial enhancement and hypointensity on HBP images was seen in 62% of small HCCs (47 of 76). Eventually, when we applied the HCC criteria with arterial enhancement and washout on the 3-minute delayed phase images or hypointensity on HBP images, even for small nodules 1 cm in diameter or smaller, 66% of small HCCs (50 of 76) met these enhancement patterns in

the present study. Currently, gadoxetic acid–enhanced MR imaging is a clinically attractive imaging examination of the liver because of its combined properties of extracellular-space contrast agent imaging during the early vascular interstitial phases and hepatocyteselective imaging during the delayed phase. These properties make gadoxetic acid–enhanced MR imaging desirable for both early detection and accurate characterization of HCC. As seen in previous studies (37,38), there were many benign, small, enhancing nodules in patients with cirrhosis. Jeong et al (37) also reported that only 13% of arterial enhancing nodules (nine of 68 nodules , 2 cm) were HCC in patients with cirrhosis. In this situation, HBP imaging is useful


Yu et al

for differentiating benign hypervascular nodules from HCC because HCC usually appears as areas of hypointensity on HBP images. Sun et al (39) reported that hypointensity on HBP images was the most useful for differentiating HCC nodules (2 cm) from arterial enhancing pseudolesions. Even though we applied hypointensity on the HBP images in the detailed characterization of small arterial enhancing nodules (1 cm), in our study, the mean PPV for the detection of small HCC was 48.3% (70 of 145), and 36 of 76 small HCCs were not verified by any observer on MR images. There were also many false-positive lesions (26 for observer 1 and 49 for observer 2). Although gadoxetic acid–enhanced MR imaging with HBP imaging improved both the detection and the accurate characterization of HCC, it still seems to be difficult to detect small HCC in patients with cirrhosis. Kim et al (19) reported that 90.7% (98 of 108) and 73.1% (79 of 108) of hypervascular small HCCs showed hyperintensity on T2-weighted images and hyperintensity on DW images, respectively. They therefore concluded that hyperintensity on both T2-weighted images and DW images is helpful in the diagnosis of hypervascular small HCC. In our study, hyperintensity on T2-weighted images and hyperintensity on DW images were seen in only 43% (33 of 76) and 28% (20 of 72) of small HCCs, respectively. Although investigators in the previous study reviewed the MR imaging features of only “hypervascular” small HCCs, there are still some differences in the imaging features of small HCC seen on T2-weighted images and on DW images. During the retrospective review of our study, however, as many small HCCs were inconspicuous on both T2-weighted images and DW images because of their small size, they were therefore interpreted as being isointense on T2-weighted images and on DW images. In the multivariable logistic regression analysis, hyperintensity on T2-weighted images and hyperintensity on DW images were also seen significantly less frequently in small HCC than in large HCC (odds ratio = 0.3 and 0.4, respectively; P , .05). On the basis 757


Yu et al

Figure 4

Figure 4: Images in a 59-year-old man with HCC. (a) Axial C-arm CT image demonstrates a subcentimeter-sized hyperenhancing nodule (arrow) identified in segment 3, the subcapsular portion of the liver. (b) On a follow-up axial C-arm CT image obtained 2 months later, this nodule (arrow) shows interval growth and compact iodized oil uptake at CT performed after TACE (see Fig E3 [online]). Therefore, the previous subcentimeter-sized hyperenhancing nodule seen at initial C-arm CT (a) was diagnosed as small HCC (group 2). (c–e) On axial MR images obtained at the time of initial C-arm CT, this small HCC (arrow) shows arterial enhancement (c) and isointensity on the 3-minute delayed phase image (d), but it shows hypointensity on the HBP image (e). The two reviewers interpreted the small nodule as an HCC.

of our study results, small HCC often seems to appear isointense rather than hyperintense on both T2-weighted images and DW images. Our study had several limitations. First, as it was designed retrospectively, a selection bias was unavoidable. Furthermore, gadoxetic acid– enhanced MR examinations were performed by using various kinds of MR units, because of the retrospective study design. Despite this heterogeneity of our MR units, however, we used the acceptable and adequate MR image quality to evaluate HCC. Second, DW imaging was not performed in all of the study patients, and we evaluated DW images by using a single pair of b values, 0 sec/mm2 and 500 sec/mm2. 758

These images can be confounding for identification of the true restricted diffusion, as they have a great deal of perfusion effect. Third, we used two kinds of reference standards for small HCC—that is, pathology findings and C-arm CT-based imaging criteria—in the two patient groups. Even though pathology findings constitute the most reliable reference standard, it is difficult to detect small HCC in the specimen, especially if the study is designed retrospectively. Even though, at our hospital, all resected livers were sectioned at no more than 5-mm to 10mm intervals in the sagittal plane, this limitation was also unavoidable. Until now, C-arm CT-based imaging criteria of HCC have been less reliable than

pathology findings. In addition, hypovascular HCC is not demonstrated at C-arm CT. In clinical practice, however, it is difficult to confirm small HCC histopathologically. Moreover, most small HCCs are diagnosed by means of follow-up imaging in actual clinical practice. With regard to this aspect, despite the heterogeneity of our study population, our population is closer to the actual clinical situation. Fourth, in our study, it was difficult to determine specificity, as it is difficult to select control nodules in patients with liver cirrhosis. There are so many cirrhosis-related focal lesions—that is, dysplastic nodules, regenerative nodules, and benign nodules. In this special situation, it is difficult to define



the true-negative lesions. Therefore, in many previous studies regarding the diagnostic performance of the JAFROC method for HCC detection, they focused on the sensitivity and PPV rather than the specificity (21,33,40). In addition, because our study patients had multiple HCC lesions, our study results may have a potential bias because of the clustering effect. Therefore, we analyzed the diagnostic performance of MR imaging on the basis of both a per-lesion analysis and a perpatient analysis. Fifth, gadoxetic acid– enhanced MR imaging is theoretically weak in allowing delineation of hypervascular HCC during the early vascular phase, as well as enhancement of the hepatic blood vessels, owing to the lower molar concentration of gadolinium. However, Feuerlein et al (41) reported that portal vein–to-liver contrast during gadoxetic acid–enhanced MR imaging cannot be improved within a dose spectrum of 0.025–0.06 mmol per kilogram of body weight. In addition, according to previous reports (16,42), gadoxetic acid–enhanced MR imaging shows high diagnostic performance for the detection of HCC, even when it is compared with that obtained by using an extracellular contrast agent. Therefore, we believe that the problem does not greatly affect the diagnostic performance of gadoxetic acid–enhanced MR imaging for the detection of HCC. Last, the assessment of MR imaging features of small HCCs can be made subjectively to some degree because of small lesion size. To overcome this drawback, two attending reviewers assessed in consensus all of the HCC nodules. Despite these many limitations, we believe our study is valuable because it focused primarily on the diagnostic performance and imaging features of gadoxetic acid–enhanced MR imaging for small HCC. In conclusion, the diagnostic performance of gadoxetic acid–enhanced MR imaging for detection of small HCC is still low, with a mean sensitivity of 46.0% (70 of 152) and a mean PPV of 48.3% (70 of 145). The arterial enhancement and hypointensity on HBP images are the most common

MR imaging features of small HCC. If we apply the washout as hypointensity on the 3-minute delayed phase images or HBP images, the diagnostic performance of gadoxetic acid–enhanced MR imaging for detection of small HCC can be improved. Acknowledgments: We thank Bonnie Hami, South Euclid, Ohio, and Jeehyun Kim, Berkeley, Calif, for editorial assistance in the preparation of this manuscript. Disclosures of Conflicts of Interest: M.H.Y. No relevant conflicts of interest to disclose. J.H.K. No relevant conflicts of interest to disclose. J.H.Y. No relevant conflicts of interest to disclose. H.C.K. No relevant conflicts of interest to disclose. J.W.C. No relevant conflicts of interest to disclose. J.K.H. No relevant conflicts of interest to disclose. B.I.C. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: author received a grant from Samsung Electronics. Other relationships: none to disclose.

References 1. Parkin DM, Bray F, Ferlay J, Pisani P. Estimating the world cancer burden: Globocan 2000. Int J Cancer 2001;94(2):153–156. 2. Bosch FX, Ribes J, Díaz M, Cléries R. Primary liver cancer: worldwide incidence and trends. Gastroenterology 2004;127(5 Suppl 1):S5–S16. 3. Tomizawa M, Kondo F, Kondo Y. Growth patterns and interstitial invasion of small hepatocellular carcinoma. Pathol Int 1995;45(5):352–358. 4. Kojiro M. Focus on dysplastic nodules and early hepatocellular carcinoma: an Eastern point of view. Liver Transpl 2004; 10(2 Suppl 1):S3–S8. 5. Nakashima Y, Nakashima O, Tanaka M, Okuda K, Nakashima M, Kojiro M. Portal vein invasion and intrahepatic micrometastasis in small hepatocellular carcinoma by gross type. Hepatol Res 2003;26(2):142–147. 6. Sala M, Llovet JM, Vilana R, et al. Initial response to percutaneous ablation predicts survival in patients with hepatocellular carcinoma. Hepatology 2004;40(6):1352–1360. 7. Bruix J, Sherman M; American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma: an update. Hepatology 2011;53(3):1020–1022. 8. European Association for the Study of the Liver; European Organisation for Research and Treatment of Cancer. EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol


Yu et al

2012;56(4):908–943. [Published correction appears in J Hepatol 2012;56(6):1430.] 9. Forner A, Vilana R, Ayuso C, et al. Diagnosis of hepatic nodules 20 mm or smaller in cirrhosis: prospective validation of the noninvasive diagnostic criteria for hepatocellular carcinoma. Hepatology 2008;47(1): 97–104. 10. Hirakawa M, Yoshimitsu K, Irie H, et al. Performance of radiological methods in diagnosing hepatocellular carcinoma preoperatively in a recipient of living related liver transplantation: comparison with step section histopathology. Jpn J Radiol 2011;29(2): 129–137. 11. Park MS, Kim S, Patel J, et al. Hepatocellular carcinoma: detection with diffusionweighted versus contrast-enhanced magnetic resonance imaging in pretransplant patients. Hepatology 2012;56(1):140–148. 12. Baek CK, Choi JY, Kim KA, et al. Hepatocellular carcinoma in patients with chronic liver disease: a comparison of gadoxetic acid–enhanced MRI and multiphasic MDCT. Clin Radiol 2012;67(2):148–156. 13. Ringe KI, Husarik DB, Sirlin CB, Merkle EM. Gadoxetate disodium-enhanced MRI of the liver. Part 1. Protocol optimization and lesion appearance in the noncirrhotic liver. AJR Am J Roentgenol 2010;195(1):13–28. 14. Cruite I, Schroeder M, Merkle EM, Sirlin CB. Gadoxetate disodium-enhanced MRI of the liver. Part 2. Protocol optimization and lesion appearance in the cirrhotic liver. AJR Am J Roentgenol 2010;195(1):29–41. 15. Zech CJ, Reiser MF, Herrmann KA. Im aging of hepatocellular carcinoma by computed tomography and magnetic resonance imaging: state of the art. Dig Dis 2009;27(2):114–124. 16. Park G, Kim YK, Kim CS, Yu HC, Hwang SB. Diagnostic efficacy of gadoxetic acid–enhanced MRI in the detection of hepatocellular carcinomas: comparison with gadopentetate dimeglumine. Br J Radiol 2010;83(996): 1010–1016. 17. Lee JM, Zech CJ, Bolondi L, et al. Consensus report of the 4th international forum for gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid magnetic resonance imaging. Korean J Radiol 2011;12(4):403–415. 18. Ichikawa T, Saito K, Yoshioka N, et al. Detection and characterization of focal liver lesions: a Japanese phase III, multicenter comparison between gadoxetic acid disodium-enhanced magnetic resonance imaging and contrastenhanced computed tomography predominantly in patients with hepatocellular carci-

759


noma and chronic liver disease. Invest Radiol 2010;45(3):133–141. 19. Kim JE, Kim SH, Lee SJ, Rhim H. Hypervascular hepatocellular carcinoma 1 cm or smaller in patients with chronic liver disease: characterization with gadoxetic acid–enhanced MRI that includes diffusion-weighted imaging. AJR Am J Roentgenol 2011;196(6): W758–W765. 20. Kudo M. Multistep human hepatocarcino genesis: correlation of imaging with pathology. J Gastroenterol 2009;44(Suppl 19):112–118. 21. Iwazawa J, Ohue S, Hashimoto N, Abe H, Hamuro M, Mitani T. Detection of hepatocellular carcinoma: comparison of angiographic C-arm CT and MDCT. AJR Am J Roentgenol 2010;195(4):882–887. 22. Meyer BC, Witschel M, Frericks BB, et al. The value of combined soft-tissue and vessel visualisation before transarterial chemoembolisation of the liver using C-arm computed tomography. Eur Radiol 2009;19(9): 2302–2309. 23. Lencioni R, Pinto F, Armillotta N, et al. Intrahepatic metastatic nodules of hepatocellular carcinoma detected at lipiodol CT: imaging–pathologic correlation. Abdom Imaging 1997;22(3):253–258. 24. Kim HC, Chung JW, Lee IJ, et al. Intercostal artery supplying hepatocellular carcinoma: demonstration of a tumor feeder by C-arm CT and multidetector row CT. Cardiovasc Intervent Radiol 2011;34(1):87–91. 25. Yu MH, Kim JH, Yoon JH, et al. Role of C-arm CT for transcatheter arterial chemoembolization of hepatocellular carcinoma: diagnostic performance and predictive value for therapeutic response compared with gadoxetic acid–enhanced MRI. AJR Am J Roentgenol 2013;201(3):675–683. 26. Chakraborty DP. Analysis of location spe cific observer performance data: validated extensions of the jackknife free-response (JAFROC) method. Acad Radiol 2006;13(10): 1187–1193.

760

27. Chakraborty DP, Berbaum KS. Observer studies involving detection and localization: modeling, analysis, and validation. Med Phys 2004;31(8):2313–2330.

Yu et al

imaging findings, including signal intensity and dynamic enhancement patterns. J Magn Reson Imaging 2007;26(6):1548–1555.

28. Zheng B, Chakraborty DP, Rockette HE, Maitz GS, Gur D. A comparison of two data analyses from two observer performance studies using Jackknife ROC and JAFROC. Med Phys 2005;32(4):1031–1034.

36. Weinmann HJ, Schuhmann-Giampieri G, Schmitt-Willich H, Vogler H, Frenzel T, Gries H. A new lipophilic gadolinium chelate as a tissue-specific contrast medium for MRI. Magn Reson Med 1991;22(2):233– 237; discussion 242.

29. Ahn SS, Kim MJ, Lim JS, Hong HS, Chung YE, Choi JY. Added value of gadoxetic acid– enhanced hepatobiliary phase MR imaging in the diagnosis of hepatocellular carcinoma. Radiology 2010;255(2):459–466.

37. Jeong YY, Mitchell DG, Kamishima T. Small (,20 mm) enhancing hepatic nodules seen on arterial phase MR imaging of the cirrhotic liver: clinical implications. AJR Am J Roentgenol 2002;178(6):1327–1334.

30. Sangiovanni A, Manini MA, Iavarone M, et al. The diagnostic and economic impact of contrast imaging techniques in the diagnosis of small hepatocellular carcinoma in cirrhosis. Gut 2010;59(5):638–644.

38. Holland AE, Hecht EM, Hahn WY, et al. Importance of small (20-mm) enhancing lesions seen only during the hepatic arterial phase at MR imaging of the cirrhotic liver: evaluation and comparison with whole explanted liver. Radiology 2005;237(3):938–944.

31. Akai H, Kiryu S, Matsuda I, et al. Detection of hepatocellular carcinoma by Gd-EOBDTPA-enhanced liver MRI: comparison with triple phase 64 detector row helical CT. Eur J Radiol 2011;80(2):310–315. 32. Park MJ, Kim YK, Lee MW, et al. Small hepatocellular carcinomas: improved sensitivity by combining gadoxetic acid–enhanced and diffusion-weighted MR imaging patterns. Radiology 2012;264(3):761–770. 33. Kim YK, Kim CS, Han YM, Park G. Detection of small hepatocellular carcinoma: can gadoxetic acid–enhanced magnetic resonance imaging replace combining gadopentetate dimeglumine–enhanced and superparamagnetic iron oxide–enhanced magnetic resonance imaging? Invest Radiol 2010;45(11):740–746. 34. Park MJ, Kim YS, Lee WJ, Lim HK, Rhim H, Lee J. Outcomes of follow-up CT for small (5–10-mm) arterially enhancing nodules in the liver and risk factors for developing hepatocellular carcinoma in a surveillance population. Eur Radiol 2010;20(10):2397–2404. 35. van den Bos IC, Hussain SM, Dwarkasing RS, et al. MR imaging of hepatocellular carcinoma: relationship between lesion size and

39. Sun HY, Lee JM, Shin CI, et al. Gadoxetic acid–enhanced magnetic resonance imaging for differentiating small hepatocellular carcinomas (2 cm in diameter) from arterial enhancing pseudolesions: special emphasis on hepatobiliary phase imaging. Invest Radiol 2010;45(2):96–103. 40. Marin D, Di Martino M, Guerrisi A, et al. Hepatocellular carcinoma in patients with cirrhosis: qualitative comparison of gadobenate dimeglumine–enhanced MR imaging and multiphasic 64-section CT. Radiology 2009;251(1):85–95. 41. Feuerlein S, Boll DT, Gupta RT, Ringe KI, Marin D, Merkle EM. Gadoxetate disodium– enhanced hepatic MRI: dose-dependent contrast dynamics of hepatic parenchyma and portal vein. AJR Am J Roentgenol 2011;196(1): W18–W24. 42. Liu X, Zou L, Liu F, Zhou Y, Song B. Gadoxetic acid disodium–enhanced magnetic resonance imaging for the detection of hepatocellular carcinoma: a meta-analysis. PLoS ONE 2013;8(8):e70896.


Hepatocellular carcinoma: MR imaging.

Diagnostic performance of gadoxetic acid (Primovist)-enhanced MR imaging versus CT during hepatic arteriography and portography for small hypervascular hepatocellular carcinoma.

Morphologic and Molecular Features of Hepatocellular Adenoma with Gadoxetic Acid-enhanced MR Imaging.

Early hepatocellular carcinoma: MR imaging.

Small hepatocellular carcinoma versus small cavernous hemangioma: differentiation with MR imaging at 2.0 T.

Differentiation of lipid poor angiomyolipoma from hepatocellular carcinoma on gadoxetic acid-enhanced liver MR imaging.

MR imaging of hepatocellular carcinoma at 1.5 Tesla.

MR imaging in cirrhosis and hepatocellular carcinoma.

Diagnostic accuracy of gadoxetic acid-enhanced MR for small hypervascular hepatocellular carcinoma and the concordance rate of Liver Imaging Reporting and Data System (LI-RADS).

Imaging features of subcentimeter hypointense nodules on gadoxetic acid-enhanced hepatobiliary phase MR imaging that progress to hypervascular hepatocellular carcinoma in patients with chronic liver disease.

Experimental hepatocellular carcinoma: MR receptor imaging.

Hepatocellular carcinoma: diagnostic criteria by imaging techniques.

Hepatocellular carcinoma: Advances in diagnostic imaging.

Imaging and Timing Performance of 1cm × 1cm Position-sensitive Solid-state Photomultiplier.

CT and MR imaging diagnosis and staging of hepatocellular carcinoma: part II. Extracellular agents, hepatobiliary agents, and ancillary imaging features.

Hepatocellular carcinoma and other hepatic malignancies: MR imaging.

Hepatocellular carcinoma: evaluation with color Doppler US and MR imaging.

Small hepatocellular carcinoma treated with percutaneous ethanol injection: MR imaging findings.

MR imaging of small hepatocellular carcinoma: effect of intratumoral copper content on signal intensity.

MR imaging features of medulloblastomas.

Nonhypervascular Hypointense Nodules at Gadoxetic Acid-enhanced MR Imaging in Chronic Liver Disease: Diffusion-weighted Imaging for Characterization.

Diagnostic performance of contrast-enhanced multidetector computed tomography and gadoxetic acid disodium-enhanced magnetic resonance imaging in detecting hepatocellular carcinoma: direct comparison and a meta-analysis.

Hepatocellular carcinoma in Budd-Chiari syndrome: enhancement patterns at dynamic gadolinium-enhanced T1-weighted MR imaging.

Comparison of diffusion-weighted imaging and gadoxetic acid-enhanced MR images in the evaluation of hepatocellular carcinoma and hypovascular hepatocellular nodules.