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

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.

Original Research n Gastrointestinal

Jiyoung Hwang, MD Young Kon Kim, MD, PhD Woo Kyoung Jeong, MD, PhD Dongil Choi, MD, PhD Hyunchul Rhim, MD, PhD Won Jae Lee, MD, PhD

Purpose:

To compare the diagnostic performance of magnetic resonance (MR) imaging features, including those on diffusionweighted (DW) and T2-weighted images, in differentiating between hypovascular hepatocellular carcinoma (HCC) and dysplastic nodules seen as hypointense nodules at hepatobiliary phase gadoxetic acid–enhanced MR imaging.

Materials and Methods:

The institutional review board approved this retrospective study and waived the need to obtain informed patient consent. There were 53 patients (39 men and 14 women; age range, 32–75 years) with histologically proven hypovascular HCCs (n = 25) and/or dysplastic nodules (n = 31) who underwent gadoxetic acid–enhanced MR imaging at 3.0T between March 2011 and January 2014. Images of 25 HCCs and 31 dysplastic nodules were analyzed for nodule size; signal intensity on T1- and T2-weighted, portal venous phase, and DW (b value = 800 sec/mm2) images; and intralesional fat. Correlations between the hyperintensity grade of lesions and the liver-to-lesion signal intensity ratio at T2-weighted and DW imaging were determined by means of analysis with generalized estimating equations.

Results:

Hyperintensity at T2-weighted and DW imaging and hypointensity in the portal venous phase were significant features for differentiating hypovascular HCCs from dysplastic nodules (P , .05). The sensitivity of DW imaging tended to be higher than that of T2-weighted imaging (72.0% [18 of 25] vs 40.0% [10 of 25]; P = .008 for grade 2 and 3 hyperintensity). Use of the parameter of hyperintensity similar to or slightly lower than the signal intensity of the spleen on DW images (b value = 800 sec/mm2) yielded a specificity of 100% (31 of 31) for the diagnosis of hypovascular HCC by differentiating it from a dysplastic nodule.

Conclusion:

Hyperintensity at DW imaging could be a useful MR imaging feature for differentiating hypovascular HCCs from dysplastic nodules seen as hypointense nodules at gadoxetic acid–enhanced MR imaging.

1

From the Department of Radiology, Soonchunhyang University Seoul Hospital, Seoul, Republic of Korea (J.H.); and Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Irwon-dong, Gangnam-gu, Seoul 135-710, Republic of Korea (Y.K.K., W.K.J., D.C., H.R., W.J.L.). Received June 10, 2014; revision requested July 20; revision received October 8; accepted November 25; final version accepted December 16. Address correspondence to Y.K.K. (e-mail: [email protected]).

Imaging

Nonhypervascular Hypointense Nodules at Gadoxetic Acid– enhanced MR Imaging in Chronic Liver Disease: Diffusion-weighted Imaging for Characterization1

RSNA, 2015

q

Online supplemental material is available for this article.

RSNA, 2015

q

Radiology: Volume 000: Number 0—   2015 n radiology.rsna.org

1

GASTROINTESTINAL IMAGING: DW Gadoxetic Acid–enhanced MR Imaging of Hypointense Nodules in Liver Disease

T

he recently introduced hepatocyte-specific contrast agent gadoxetic acid (Primovist; Bayer Healthcare, Berlin, Germany) has opened new horizons for liver magnetic resonance (MR) imaging, with promising results for the detection and characterization of hepatocellular carcinoma (HCC). This dual-acting imaging agent provides hemodynamic information during early dynamic phases and enhances HCC detection in the hepatobiliary phase (1–3). Results of extensive studies have shown that most HCCs are observed as hypointense areas relative to the liver

Advances in Knowledge nn Hyperintensity at T2-weighted and diffusion-weighted (DW) imaging (P , .0001) and hypointensity in the portal venous phase of gadoxetic acid– enhanced MR imaging (P = .037) were significant MR imaging features for differentiating hypovascular hepatocellular carcinomas (HCCs) from dysplastic nodules. nn We found superior sensitivity of DW imaging compared with T2-weighted imaging in the diagnosis of hypovascular HCCs when we applied the parameter of hyperintensity similar to or slightly lower than the signal intensity of the spleen (72.0% [18 of 25] vs 40.0% [10 of 25]; P = .008) as the criterion for HCC. nn Applying the parameter of hyperintensity similar to or slightly lower than the signal intensity of the spleen on DW images (b value = 800 sec/mm2) achieved 100% specificity (31 of 31) for the diagnosis of hypovascular HCC by differentiating it from dysplastic nodules seen as hypointense nodules at gadoxetic acid–enhanced MR imaging. nn We found significant correlations between the hyperintensity grade of hepatic nodules and liver-tolesion signal intensity ratios at both T2-weighted imaging and DW imaging (P , .001). 2

in the hepatobiliary phase (1–5). This feature can be exploited for the diagnosis of HCC, particularly for HCCs that appear as arterial-only enhancing nodules without delayed washout on conventional dynamic images and for hypovascular early HCCs, which can be seen only in the hepatobiliary phase (3–5). However, hypointensity in the hepatobiliary phase in the setting of cirrhosis is not an exclusive finding of HCC but can also be seen in dysplastic nodules, particularly high-grade dysplastic nodules (2,6,7). Thus, differentiating between dysplastic nodules and hypovascular HCCs, including early well-differentiated HCCs, remains challenging when one is using current noninvasive HCC criteria or even gadoxetic acid (6–11). Several studies (12–17) have been performed to identify imaging features that are associated with hypovascular hypointense nodules at gadoxetic acid– enhanced MR imaging in cirrhosis that progresses to hypervascular HCC. Variable MR imaging features such as hyperintensity at diffusion-weighted (DW) imaging or T2-weighted imaging, T1 hyperintensity, large size, and intralesional fat are known to be associated with the subsequent transformation of nodules that are hypointense in the hepatobiliary phase into hypervascular HCCs (12–17). In particular, hyperintensity at T2-weighted or DW imaging in hepatocarcinogenesis is reportedly a strong indicator of HCC (8,18–20). Thus, it is reasonable to assume that these features may be useful to categorize hepatocellular nodules that fall into the gray zone of hepatocarcinogenesis or to characterize hypovascular HCC. Given that signal intensity in DW imaging is contributed to by changes in diffusivity that reflect the tumor microenvironment (eg, cellular density) as well as the T2 shine-through

Implication for Patient Care nn DW imaging is beneficial in discriminating between hypovascular HCCs and dysplastic nodules that are both seen as hypointense nodules at hepatobiliary phase gadoxetic acid–enhanced MR imaging.

Hwang et al

effect (21), hyperintensity at DW imaging may be more sensitive than other features for the early detection of malignant changes. In this context, we performed this study to compare the diagnostic performance of MR imaging features, including DW imaging and T2-weighted imaging features, in differentiating between hypovascular HCCs and dysplastic nodules observed as hypointense nodules at hepatobiliary phase gadoxetic acid–enhanced MR imaging.

Materials and Methods Study Population Our study had institutional review board approval, and the need to obtain informed consent was waived. We retrospectively searched our institution’s pathology database between March 2011 and January 2014 to find patients with HCC or cirrhosis-associated benign hepatocellular nodules such as high-grade dysplastic nodules and low-grade dysplastic nodules that had been proven by surgical resection or biopsy (Fig 1). This search identified 230 patients who had undergone surgery or percutaneous liver biopsy because they were suspected of having HCC on the basis of MR imaging findings. The exclusion

Published online before print 10.1148/radiol.15141350 Content codes: Radiology 2015; 000:1–10 Abbreviations: ADC = apparent diffusion coefficient CI = confidence interval DW = diffusion weighted HCC = hepatocellular carcinoma Author contributions: Guarantors of integrity of entire study, J.H., Y.K.K.; study concepts/study design or data acquisition or data analysis/ interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, J.H., Y.K.K.; clinical studies, all authors; experimental studies, J.H.; statistical analysis, J.H., Y.K.K.; and manuscript editing, J.H., Y.K.K., D.C., W.J.L. Conflicts of interest are listed at the end of this article.

radiology.rsna.org n Radiology: Volume 000: Number 0—   2015


criteria for the patient group were as follows: (a) patients with HCCs or dysplastic nodules that showed arterial hypervascularity or no hypointensity at hepatobiliary phase MR imaging (n = 157); (b) patients with unavailable liver MR imaging studies, including gadoxetic acid–enhanced imaging and DW imaging studies (n = 14); and (c) patients who had undergone treatment for HCC prior to the MR imaging examination (n = 6). Hypovascular HCC was defined as when lesion showed no arterially enhancing foci on both MR vascular phase images (including subtraction imaging for T1 hyperintense nodules) and multiphasic computed tomographic (CT) images (acquired with a 64–detector row CT scanner). A total of 53 patients (39 men and 14 women; age range, 32–75 years; mean age, 57 years) who had 25 hypovascular HCCs and/or 31 dysplastic nodules were included. The reference standard for liver lesions was based on the histopathologic examination of surgical specimens for 22 HCCs, 10 low-grade dysplastic nodules, and 15 high-grade dysplastic nodules and the examination of specimens from ultrasonographically guided percutaneous biopsy for three HCCs and six dysplastic nodules (one low-grade dysplastic nodule and five high-grade dysplastic nodules). The average time between MR examination and surgery was 13 days (range, 6–30 days). The surgical procedures included segmentectomy (n = 31 patients), bisegmentectomy (n = 4), lobectomy (n = 8), and liver transplantation (n = 1). The causes of liver disease included cirrhosis or chronic hepatitis associated with hepatitis B virus (n = 49), cirrhosis or chronic hepatitis associated with hepatitis C virus (n = 2), and cirrhosis associated with both hepatitis B and C viruses (n = 2). The study group and lesion characteristics are shown in Table 1. Among the patients, 51 were classified as having Child-Pugh class A disease, and two were classified as having Child-Pugh class B disease. The 25 hypovascular HCCs consisted of 15 well-differentiated (grade I) and 10 moderately differentiated HCCs (grade II) at histologic examination. The

Hwang et al

Figure 1

Figure 1: Flowchart of the study population.

Table 1 Patient and Tumor Characteristics Characteristic No. of patients with HCC/HCC and DN/DN Mean age (y)* No. of men/no. of women† No. of patients with tumors caused by HBV/HCV/HBV and HCV No. of patients with Child-Pugh class A/B/C disease Lesion size of 25 hypovascular HCCs (cm)* Lesion size of 31 DNs (cm)* No. of lesions with Edmondson grade I/II/III No. of nodules per patient: 1/2 Diagnosis at surgery or biopsy

Datum 24/1/28 57 (32–75) 39 (55) [32–75]/14 (58) [43–68] 49/2/2 51/2/0 1.6 (0.7–2.7) 1.5 (0.6–2.5) 15/10/0 24 (HCC), 26 (10 LGDN and 16 HGDN)/1 (HCC/HGDN), 2 (HGDN/LGDN and HGDN/HGDN) 22 HCCs, 10 LGDNs, 15 HGDNs/3 HCCs, 1 LGDN, 5 HGDNs

Note.—DN = dysplastic nodule, HBV = hepatitis B virus, HCV = hepatitis C virus, HGDN = high-grade DN, LGDN = low-grade DN. * Data in parentheses are the range. †

Data in parentheses are mean ages (in years), with ranges in square brackets.

diameter of the HCCs ranged from 0.7 to 2.7 cm (mean, 1.6 cm 6 0.5 [standard deviation]), and the diameter of the dysplastic nodules ranged from 0.6 to 2.5 cm (mean, 1.5 cm 6 0.5). Of the 24 patients who had only HCCs, each patient had only a solitary HCC. One patient had one HCC and one dysplastic nodule. Of the 28 patients with only dysplastic nodules, two patients had two dysplastic nodules each, and 26 patients had one dysplastic nodule each.

MR Imaging MR images were acquired by using a 3.0-T MR imaging system (Intera


Achieva 3.0 T; Philips Healthcare, Best, the Netherlands) equipped with a dualsource parallel radiofrequency transmission system and a quadrature body coil. Baseline MR imaging included a T1-weighted turbo field-echo opposed (first-echo echo time) and in-phase (second-echo echo time) sequence, a breath-hold multishot T2-weighted sequence (echo train length, 14), and a respiratory-triggered multishot T2weighted turbo spin-echo sequence (echo train length, 82) with a field of view of 32–38 cm. DW imaging was performed by using respiratory-triggered single-shot echo-planar imaging 3


Hwang et al

Table 2 Parameters of MR Imaging Sequences

Sequence T1-weighted imaging 3D dual GRE BH MS T2-weighted imaging RT MS T2-weighted imaging T1-weighted imaging 3D GRE DW imaging

Repetition Time (msec)/Echo Time (msec)

Flip Angle (degrees)

Section Thickness (mm)

Matrix Size

3.5/1.15–2.3

10

6

256 3 194

1623/70

90

5

324 3 235

1156/160

90

5

376 3 270

3.1/1.5

10

2

256 3 256

1600/70

90

5

112 3112

Measured/Reconstructed Voxel Size (mm)

Bandwidth (Hz/pixel)

1.86 3 1.99 3 6.0/1.09 3 1918.6/0.226 1.10 3 6.0 1.46 3 1.47 3 5.0/0.63 3 255.3/1.702 0.62 3 0.5 0.99 3 1.38 3 5.0/0.73 3 388.9/1.117 0.73 3 0.5 1.49 3 1.49 3 4.0/0.96 3 723.4/0.601 0.96 3 2.0 3.5 3 3.5 3 5.0/1.4 3 79.5/5.467 1.4 3 5.0

Fat Suppression Opposed/in phase SPAIR SPIR SPAIR SPIR

Acquisition Time (sec)

No. of Signals Acquired

14

1

55

1

120

2

16.6 126

1 2

Note.—BH = breath hold, GRE = gradient echo, MS = multishot, SPAIR = spectral attenuated inversion recovery technique, SPIR = spectral presaturation with inversion recovery, RT = respiratory triggered, 3D = three-dimensional.

with b values of 0, 100, and 800 sec/ mm2. The apparent diffusion coefficient (ADC) was calculated by using a monoexponential function with b values of 100 and 800 sec/mm2 to minimize perfusion effects. For gadoxetic acid–enhanced imaging, unenhanced, arterial phase (20–35 seconds), portal phase (60 seconds), late phase (3 minutes), and 20-minute hepatobiliary phase images were obtained by using a T1-weighted threedimensional turbo-field-echo sequence (T1 high-resolution isotropic volume examination, THRIVE; Philips Healthcare). The time for arterial phase imaging was determined by using the MR fluoroscopic bolus detection technique. The contrast agent (0.25 mol/L) was administered intravenously by using a power injector at a rate of 1 mL/sec for a dose of 0.025 mmol per kilogram of body weight, followed by a 20-mL saline flush. The detailed parameters of the MR imaging sequences used are shown in Table 2.

Image Analysis All MR imaging features were evaluated in an anonymized and randomized manner by two gastrointestinal radiologists (J.H. and Y.K.K., with 5 and 13 years of experience in liver MR imaging, respectively) who were blinded to the final pathologic 4

diagnosis. All images were evaluated by using a picture archiving and communication system (Pathspeed; GE Medical Systems Integrated Imaging Solutions, Mount Prospect, Ill). The readers were free to alter the window level and the window width at their discretion. First, the two observers independently graded the signal intensity of all 25 HCCs and 31 dysplastic nodules relative to that of the surrounding liver parenchyma on the T2-weighted images and the DW images obtained at a b value of 800 sec/ mm2 with the following initial grading system: a grade of 0 indicated that the lesion was not observed or was isointense; a grade of 1, that the lesion had minimal perceptible hyperintensity; a grade of 2, that the lesion was hyperintense but had signal intensity slightly lower than that of the spleen; and a grade of 3, that the lesion was hyperintense with a signal intensity similar to that of the spleen. No patient had previously undergone splenectomy. Hypointense lesions or nondetectable lesions were given a rating of 0. Second, to investigate image features that could reliably differentiate between dysplastic nodules and HCCs that both showed no arterial hyperenhancement, the readers evaluated in consensus the following features: (a) lesion size (≤1.0 cm, 1.1–2.0 cm, or > 2.1 cm), (b) the

presence of low signal intensity or hyperintensity at T1-weighted imaging, (c) the presence of hyperintensity at T2-weighted imaging, (d) the presence of hyperintensity at DW imaging, (e) the presence of hypointensity in the portal venous phase, and (f) the presence of intralesional fat. The presence of a fat component in the nodule was determined on the basis of signal intensity decrease at opposed-phase compared with in-phase T1-weighted imaging. All fat-containing lesions were proved at histologic examination. The portal venous phase images were included to assess washout of the lesions. The observers were asked to record the presence and the location of the lesions, which was assessed by using the Couinaud classification system. To avoid discrepancies between the findings of the recorded lesions and the findings of the standard of reference, each observer recorded the image number for each lesion. After individual review, joint evaluation was performed until a consensus was reached on the results. Quantitative measurements were performed by an abdominal radiologist (W.K.J., with 10 years of experience in liver MR imaging) who did not participate in the qualitative image analysis to determine the reliability of qualitative assessment by measuring the signal



intensity of the liver and all lesions by using operator-defined regions of interest (ROIs) on T2-weighted images and the DW images obtained at a b value of 800 sec/mm2. For measurements within the lesion, the round ROI was positioned manually so as to avoid necrotic foci as much as possible (mean ROI, 1.34 cm2; range, 0.78–1.74 cm2). For signal intensities of the liver, round ROIs were drawn in the same location on each image devoid of a large intrahepatic vessel. The shapes and sizes of the ROIs were identical for all images as much as possible and were drawn larger than 1.0 cm2 (mean ROI, 1.52 cm2; range, 1.40–1.65 cm2). The liverto-lesion signal intensity ratio (SIR) was calculated by using the following equation: SIR = (SIlesion − SIliver)/SIparaspi, where SIlesion is the signal intensity of nal the lesion, SIliver is the signal intensity of the liver, and SIparaspinal is the signal intensity of the paraspinal muscle. We also measured the ADC for each nodule on the ADC map using the b values of 100 and 800 sec/mm2.

Statistical Analysis For data involving lesions from the same patients, methods considering correlation among lesions within the same patient were used. Analyses with generalized estimating equations (GEEs) were performed to determine the univariate statistical significances of each categoric variable in differentiating between 25 hypovascular HCCs and 31 dysplastic nodules. Multivariate analysis with GEEs could not be performed because the sample sizes for the categories of lesion size, T1weighted imaging, and DW imaging were too small. Instead, we compared the sensitivity and specificity of T2weighted imaging with those of DW imaging for differentiating hypovascular HCCs from dysplastic nodules. The sensitivities of T2-weighted imaging and DW imaging for the diagnosis of hypovascular HCC by differentiating it from a dysplastic nodule were compared by using the exact McNemar test because each patient had only one HCC. The specificities of T2-weighted imaging and DW imaging

were estimated with consideration of the patient as a cluster and were compared by using GEE analysis. We determined the correlation between the qualitative signal intensity grades and the liver-to-lesion signal intensity ratios at T2-weighted and DW imaging by using GEE. The k or weighted k statistic for two observers, with consideration of the patient as a cluster, was calculated to assess interobserver agreement regarding the degree of hyperintensity of lesions at T2-weighted and DW imaging and the signal intensity at T1-weighted imaging and in the portal venous phase. The 95% confidence intervals (CIs) of the weighted k statistics were estimated from 1000 bootstrapped samples by means of the cluster bootstrap method (22). GEEs were used to determine the significance of differences in ADC between HCCs and dysplastic nodules. Statistical analysis was performed by using SAS, version 9.3 (SAS Institute, Cary, NC). The null hypothesis of no difference was rejected for P values of less than .05.

Results Univariate and Multivariate Analysis Univariate analysis revealed that hyperintensity at T2-weighted and DW imaging (P , .0001) and hypointensity in the portal venous phase (P = .037) were significant MR imaging features among six categories to differentiate between hypovascular HCCs and dysplastic nodules (Figs 2, 3; Table 3). At T2-weighted imaging, 19 (76.0%) of 25 hypovascular HCCs and four (12.9%) of 31 dysplastic nodules showed hyperintensity; the remaining lesions were isointense. Nine, four, and six HCCs were assigned grade 1, 2, and 3 hyperintensity at T2-weighted imaging, respectively, whereas four dysplastic nodules were assigned grade 1 hyperintensity. At DW imaging, 23 (92.0%) hypovascular HCCs showed hyperintensity, with five designated as showing grade 1, six as showing grade 2 (Fig 4), and 12 as showing grade 3 (Fig 5) hyperintensity. The remaining two well-differentiated HCCs with


Hwang et al

diffuse fatty infiltration (90% intralesional fat at histologic examination) were iso- or hypointense at DW imaging (Fig E1 [online]). Five (16.1%) of the 31 dysplastic nodules showed hyperintensity at DW imaging and were designated as grade 1, and the remaining 26 (83.9%) were isointense. Thus, two high-grade dysplastic nodules and two low-grade dysplastic nodules showed hyperintensity at both T2-weighted and DW imaging. An additional high-grade dysplastic nodule appeared hyperintense at DW imaging but not at T2weighted imaging. In the portal venous phase, 13 (52.0%) of the HCCs and seven (22.6%) of the dysplastic nodules were hypointense, while the remaining lesions were isointense. There were no significant differences between hypovascular HCCs and dysplastic nodules with respect to lesion size (P = .690), signal intensity at T1weighted imaging (P = .064), and presence of intralesional fat (P = .480). When hyperintensity (grades 1, 2, and 3) at T2-weighted imaging or at DW imaging (b value = 800 sec/mm2) was used as the criterion for hypovascular HCC, the sensitivity of DW imaging (92.0% [23 of 25]; exact 95% CI: 74.0%, 99.0%) was higher than that of T2-weighted imaging (76.0% [19/25]; exact 95% CI: 54.9%, 90.6%); however, this difference was not statistically significant (P = .125). We found no significant difference in specificity between DW imaging (83.9% [26 of 31]; 95% CI: 70.5%, 97.2%) and T2weighted imaging (87.1% [27 of 31]; 95% CI: 75.0%, 99.2%) (P = .311). In addition, for multivariate analysis considering only hyperintensity at T2weighted and DW imaging, after adjusting for the effect of T2-weighted imaging (P = .890), this parameter at DW imaging was still statistically significant (P = .003). When applying hyperintensity grades of 2 or 3 on DW images (b value = 800 sec/mm2) or T2-weighted images as the HCC criteria, we could achieve 100% specificity (31 of 31) for both kinds of image and corresponding sensitivities of 72.0% (18 of 25; exact 95% CI: 50.6%, 87.9%) and 40.0% (10 of 25; exact 95% CI: 21.1%, 61.3%; P = 5


Hwang et al

Figure 2

Figure 2: Axial MR images show well-differentiated HCC in 66-year-old man. (a) On a 20-minute hepatobiliary phase image obtained after administration of gadoxetic acid, a liver mass (arrow) is seen as hypointense only in the hepatobiliary phase, without definitive arterial hypervascularity. On (b) a T2-weighted image and (c) a single-shot echo-planar DW image (b value = 800 sec/mm2 ) the mass (arrow) is seen as subtly hyperintense and was rated as grade 1 during imaging interpretation. For an additional image in this patient, please see Figure E2 (online).

Figure 3

Figure 3: Axial MR images show high-grade dysplastic nodule in 59-year-old man. (a) On a 20-minute hepatobiliary phase image obtained after administration of gadoxetic acid, a liver mass (arrow) is seen as hypointense only in the hepatobiliary phase. On (b) a T2-weighted image and (c) a single-shot echo-planar DW image (b value = 800 sec/mm2 ), the mass is not identified. For an additional image in this patient, please see Figure E3 (online).

.008) for the diagnosis of hypovascular HCC by differentiating it from a dysplastic nodule.

Correlation between Signal Intensity Grades of All Hepatic Lesions and the Liver-to-Lesion Signal Intensity Ratio at T2-weighted and DW Imaging Table 4 shows the signal intensity grade of 25 HCCs and 31 dysplastic nodules relative to that of the surrounding liver parenchyma on T2-weighted and DW images (b value = 800 sec/mm2). We found a significant correlation between qualitative signal intensity grading and 6

the liver-to-lesion signal intensity ratio at T2-weighted imaging (mean, 0.29 6 0.54) and DW imaging (mean, 0.55 6 0.80) (P , .001). There was no significant difference in mean ADC between hypovascular HCCs (1.06 6 0.13) and dysplastic nodules (1.09 6 0.13) (P = .256) (Figure E1 [online]).

Interobserver Agreement The weighted k values for the two observers were 0.863 (95% bootstrap CI: 0.860, 0.865) for T2-weighted imaging and 0.914 (95% bootstrap CI: 0.912, 0.917) for DW imaging, thus

indicating good or excellent interobserver agreement with regard to characterizing the hyperintensity of HCCs and dysplastic nodules at T2-weighted and DW imaging. At DW imaging, there were six cases of discordance between observers in assigning grade 2 or 3 (n = 3) and grade 1 or 2 (n = 3) and no discordance in assigning grade 0. At T2-weighted imaging, there were seven cases of discordance between observers, including one case of discordance in assigning grade 0 to dysplastic nodules and the remaining six in assigning grades 1–3. Regarding



Table 3 Results of Univariate Analysis for MR Imaging Findings between Dysplastic Nodules and Hypovascular HCC Parameter and Feature Lesion size (cm) 1.0 1.1–2.0 2.1 T1-weighted imaging Isointense Hypointense Hyperintense T2-weighted imaging Isointense Hyperintense† DW imaging Isointense Hyperintense† Portal venous phase imaging Isointense Hypointense Presence of fat

Dysplastic Nodules (n = 31)*

Hypovascular HCCs (n = 25)

P Value

5 (16.1) [1/4] 24 (77.4) [8/16] 2 (6.5) [0/2]

5 (20.0) 17 (68.0) 3 (12.0)

.690

26 (83.9) [10/16] 2 (6.4) [1/1] 3 (9.7) [0/3]

16 (64.0) 8 (32.0) 1 (4.0)

.064

27 (87.1) [9/18] 4/0/0 (12.9) [2/2]

6 (24.0) 9/4/6 (76.0)

,.0001

26 (83.9) [9/17] 5/0/0 (16.1) [2/3]

2 (8.0) 5/6/12 (92.0)

,.0001

24 (77.4) [8/16] 7 (22.6) [3/4] 6 (19.4) [3/3]

12 (48.0) 13 (52.0) 3 (12.0)

.037 .480

Note.—Data in parentheses are percentages. * Data in brackets are numbers of low-grade/high-grade dysplastic nodules. †

Data in brackets for hyperintensity at T2-weighted imaging and DW imaging are numbers of grade 1/grade 2/grade 3 lesions.

Figure 4

Figure 4: Axial MR images show moderately differentiated HCC in 52-year-old man. (a) On the 20-minute hepatobiliary phase image obtained after administration of gadoxetic acid, a liver mass (arrow) is seen as hypointense, without definitive arterial hypervascularity. (b) On a single-shot echo-planar DW image (b value = 800 sec/mm2 ), the mass (arrow) is seen as subtly hyperintense. The tumor was rated as grade 2 on b during image interpretation. For additional images in this patient, please see Figure E4 (online).

the evaluation of signal intensity at T1weighted imaging and in the portal venous phase, the k values for the two observers were 0.871 (95% bootstrap

CI: 0.867, 0.875) and 0.840 (95% CI: 0.835, 0.844), indicating excellent interobserver agreement. There were three and four cases of discordance


Hwang et al

between the observers at T1-weighted imaging and in the portal venous phase, respectively. There was no discordance between observers for the presence of a fat component (k = 1.0) (Tables E1 and E2 [online]).

Discussion Our univariate analysis revealed that hyperintensity at T2-weighted and DW imaging and hypointensity in the portal venous phase were significant MR imaging features for differentiating between hypovascular HCCs and dysplastic nodules (P , .05). Two prior studies (13,14) showed that hyperintensity at DW imaging or T2-weighted imaging is an independent and strongly associated factor for subsequent arterial hypervascularization of hypointense nodules in the hepatobiliary phase in cirrhosis. However, no study, to our knowledge, has correlated imaging features of such indeterminate nodules with pathologic findings. In consideration of the multistep nature of the hepatocarcinogenetic pathway, our result is reasonable. It is well known that hyperintensity at T2-weighted imaging in hepatocarcinogenesis is strongly indicative of HCC (18–20), and the current practice at many centers involves consideration of T2 hyperintensity as a sign of HCC (23). In our study, 19 hypovascular nodules (76.0%) showed hyperintensity at T2-weighted imaging. Prior studies (24,25) showed that 42.1%–53% of HCCs appeared iso- or hypointense at T2-weighted imaging. Different selection criteria, as well as recent advances in MR image quality, and differences in the etiology and severity of cirrhosis, which may influence the signal intensity of background liver parenchyma, may be responsible for different results among studies. We found superior sensitivity for DW imaging than for T2-weighted imaging in the diagnosis of hypovascular HCCs when hyperintensity similar to or slightly lower than the signal intensity of the spleen (72.0% [18 of 25] vs 40.0% [10 of 25]; P = .008) was used as the HCC criterion. Interestingly, all 19 HCCs with hyperintensity at 7


Hwang et al

Figure 5

Figure 5: Axial MR images show well-differentiated HCC in 54-year-old man. (a) On the 20-minute hepatobiliary phase image obtained after administration of gadoxetic acid, a liver mass (arrow) is seen as hypointense only in the hepatobiliary phase, without arterial hypervascularity. (b) On a T2-weighted image, the mass is not seen. (c) On single-shot echo-planar DW image (b value = 800 sec/mm2 ), the mass (arrow) is clearly seen as hyperintense and was rated as grade 3 during imaging interpretation. Clear hyperintensity of this mass at DW imaging helps to make the diagnosis of HCC. For an additional image in this patient, please see Figure E5 (online). For images of a well-differentiated HCC in a 70-year-old woman, please see Figure E6 (online).

Table 4 Results of Hyperintensity Grading for 25 HCCs and 31 Dysplastic Nodules at T2-weighted Imaging and DW Imaging Grade at DW Imaging Grade at (b Value = 800 sec/mm2) T2-weighted Imaging 0 1 2 3 0 1 2 3 Total

2/26 0/0 0/0 0/0 2/26

3/1 2/4 0/0 0/0 5/5

Total

0/0 1/0 6/27 5/0 2/0 9/4 1/0 3/0 4/0 0/0 6/0 6/0 6/0 12/0 25/31

Note.—Data are numbers of HCCs/number of dysplastic nodules. A grade of 0 = not observed or isointense or hypointense, a grade of 1 = minimal perceptible hyperintensity, a grade of 2 = hyperintense but slightly lower than signal intensity of spleen, and a grade of 3 = hyperintense and similar in signal intensity to spleen.

T2-weighted imaging also showed hyperintensity at DW imaging. DW imaging could target cellular density and architectural changes through differences in diffusivity as well as vascular changes. In addition, changes in tissue T2 can influence appearance at DW imaging independent of tissue diffusivity, which is called the T2 shine-through or T2 blackout effect (21,26). Thus, the mixed contribution of such variable factors to lesion signal intensity 8

at DW imaging may be responsible for the higher sensitivity of DW imaging to early hepatocarcinogenesis than T2-weighted imaging. We surmise that the most probable reason for our result may be increased cellular density during the histologic transition from dysplastic nodule to HCC (27). However, although we found a tendency toward a higher mean ADC in dysplastic nodules than in HCCs, no significant difference was found. This is consistent with previous reports (28,29) of the superiority of visual assessment at DW imaging to ADCs in the characterization of focal liver lesions. Therefore, it can be said that T2 and ADC weightings combine synergistically to provide a useful parameter for characterizing HCC in our MR imaging setting. Hyperintensity at T2-weighted or DW imaging is a well-known HCC feature (2,8,18–20). However, most cases in previous reports were advanced hypervascular HCCs. Our study concerned nonhypervascular HCCs, which are early (before developing arterial hypervascularization), and dysplastic nodules seen as hypointense nodules in the hepatobiliary phase, even though some HCCs were advanced HCCs, being hypointense at pre–T1-weighted imaging (n = 8) and grade II at histologic examination (n = 10). Thus, the clinical

impact of our observation differs from that of observations in prior studies because it provides a possible guideline for nonhypervascular nodules with hypointensity in the hepatobiliary phase. We found no significant difference in specificity between DW imaging (83.9% [26 of 31]) and T2-weighted imaging (87.1% [27 of 31]) (P = .317). Consistent with findings in previous reports (2,8,30), in our study, two high-grade dysplastic nodules and two low-grade dysplastic nodules showed hyperintensity at both T2-weighted and DW imaging. Thus, our study leaves uncertainty regarding the differentiation of low-grade dysplastic nodules and high-grade dysplastic nodules at MR imaging. An additional high-grade dysplastic nodule appeared hyperintense at DW imaging but not at T2-weighted imaging. Given that dysplastic nodules can be hyperintense at T2-weighted imaging because of varying degrees of fibrosis or infarction (31,32) and signal intensity at DW imaging is closely related to T2 shine-through, this result is not surprising. In addition, the nodule-to-liver contrast can be affected by findings in the surrounding liver parenchyma, such as the degree of fibrosis, iron or fat deposition, and inflammation. Interestingly, all five dysplastic nodules that showed hyperintensity at



T2-weighted or DW imaging were designated as having grade 1 hyperintensity. Thus, identifying grade 2 and 3 hyperintensity at DW imaging yielded a specificity of 100% (31 of 31) for the diagnosis of hypovascular HCC by differentiating it from a dysplastic nodule. This visual assessment is reproducible, because good or excellent interobserver agreement regarding signal intensity grading of hepatic nodules was demonstrated, and signal intensity grades were strongly correlated with liver-tolesion signal intensity ratios. This study had several limitations. First, the study population could be biased because this was a retrospective study, excluding patients who had hypovascular nodules with hypointensity in the hepatobiliary phase with no histopathologic confirmation. Second, some lesions were diagnosed by means of percutaneous biopsy, which might cause sampling error. Third, the clinical impact of our results might be diminished in centers where patients with high-grade dysplastic nodules are also registered as treatment candidates, because the value of differentiating between patients with preoperative imaging would be diminished. Fourth, our data came from a single institution in an endemic area for hepatitis B virus infection. Thus, our results might not be applicable to other centers with different etiologies of liver cirrhosis or different MR imaging units with a variety of hardware and software (eg, field strength, choice of b value, and dual-source radiofrequency). Last, hypovascular HCCs were determined on the basis of the arterial phase of MR imaging and CT. Thus, there was a possibility of missing arterial tumor vascularity because of inadequate arterial phase timing. In conclusion, hyperintensity at DW imaging could be a useful MR imaging feature to differentiate hypovascular HCCs from dysplastic nodules seen as hypointense nodules in the hepatobiliary phase of gadoxetic acid–enhanced MR imaging. Disclosures of Conflicts of Interest: J.H. disclosed no relevant relationships. Y.K.K. disclosed no relevant relationships. W.K.J. dis-

closed no relevant relationships. D.C. disclosed no relevant relationships. H.R. disclosed no relevant relationships. W.J.L. disclosed no relevant relationships.

References 1. Ahn SS, Kim MJ, Lim JS, Hong HS, Chung YE, Choi JY. Added value of gadoxetic acidenhanced hepatobiliary phase MR imaging in the diagnosis of hepatocellular carcinoma. Radiology 2010;255(2):459–466. 2. 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. 3. 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. 4. Sun HY, Lee JM, Shin CI, et al. Gadoxetic acid-enhanced magnetic resonance imaging for differentiating small hepatocellular carcinomas (, or =2 cm in diameter) from arterial enhancing pseudolesions: special emphasis on hepatobiliary phase imaging. Invest Radiol 2010;45(2):96–103. 5. Motosugi U, Ichikawa T, Sou H, et al. Distinguishing hypervascular pseudolesions of the liver from hypervascular hepatocellular carcinomas with gadoxetic acid-enhanced MR imaging. Radiology 2010;256(1):151–158. 6. Bartolozzi C, Battaglia V, Bargellini I, et al. Contrast-enhanced magnetic resonance imaging of 102 nodules in cirrhosis: correlation with histological findings on explanted livers. Abdom Imaging 2013;38(2):290– 296. 7. Kogita S, Imai Y, Okada M, et al. Gd-EOBDTPA-enhanced magnetic resonance images of hepatocellular carcinoma: correlation with histological grading and portal blood flow. Eur Radiol 2010;20(10):2405–2413. 8. Lee MH, Kim SH, Park MJ, Park CK, Rhim H. Gadoxetic acid-enhanced hepatobiliary phase MRI and high-b-value diffusionweighted imaging to distinguish well-differentiated hepatocellular carcinomas from benign nodules in patients with chronic liver disease. AJR Am J Roentgenol 2011; 197(5):W868–W875. 9. Sersté T, Barrau V, Ozenne V, et al. Accuracy and disagreement of computed tomography and magnetic resonance imaging for the diagnosis of small hepatocellular carci-


Hwang et al

noma and dysplastic nodules: role of biopsy. Hepatology 2012;55(3):800–806. 10. Bolondi L, Gaiani S, Celli N, et al. Characterization of small nodules in cirrhosis by assessment of vascularity: the problem of hypovascular hepatocellular carcinoma. Hepatology 2005;42(1):27–34. 11. Golfieri R, Grazioli L, Orlando E, et al. Which is the best MRI marker of malignancy for atypical cirrhotic nodules: hypointensity in hepatobiliary phase alone or combined with other features? Classification after Gd-EOB-DTPA administration. J Magn Reson Imaging 2012;36(3):648–657. 12. 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):1377–1383. 13. Hyodo T, Murakami T, Imai Y, et al. Hy povascular nodules in patients with chronic liver disease: risk factors for development of hypervascular hepatocellular carcinoma. Radiology 2013;266(2):480–490. 14. Kim YK, Lee WJ, Park MJ, Kim SH, Rhim H, Choi D. Hypovascular hypointense nodules on hepatobiliary phase gadoxetic acid-enhanced MR images in patients with cirrhosis: potential of DW imaging in predicting progression to hypervascular HCC. Radiology 2012;265(1):104–114. 15. Joishi D, Ueno A, Tanimoto A, et al. Natural course of hypovascular nodules detected on gadoxetic acid-enhanced MR imaging: presence of fat is a risk factor for hypervascularization. Magn Reson Med Sci 2013;12(4):281–287. 16. 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):472–479. 17. Nakamura S, Nouso K, Kobayashi Y, et al. The diagnosis of hypovascular hepatic lesions showing hypo-intensity in the hepatobiliary phase of Gd-EOB- DTPA-enhanced MR imaging in high-risk patients for hepatocellular carcinoma. Acta Med Okayama 2013; 67(4):239–244. 18. Kelekis NL, Semelka RC, Worawattanakul S, et al. Hepatocellular carcinoma in North America: a multiinstitutional study of appearance on T1-weighted, T2-weighted, and serial gadolinium-enhanced gra-

9


Hwang et al

dient-echo images. AJR Am J Roentgenol 1998;170(4):1005–1013.

and MRI. Expert Rev Gastroenterol Hepatol 2013;7(3):269–279.

ing focal liver masses. Acad Radiol 2009; 16(10):1208–1214.

19. Kim YK, Lee YH, Kim CS, Han YM. Added diagnostic value of T2-weighted MR imaging to gadolinium-enhanced three-dimensional dynamic MR imaging for the detection of small hepatocellular carcinomas. Eur J Radiol 2008;67(2):304–310.

24. Hussain HK, Syed I, Nghiem HV, et al. T2weighted MR imaging in the assessment of cirrhotic liver. Radiology 2004;230(3):637– 644.

29. Battal B, Kocaoglu M, Akgun V, et al. Diffusion-weighted imaging in the characterization of focal liver lesions: efficacy of visual assessment. J Comput Assist Tomogr 2011;35(3):326–331.

20. Willatt JM, Hussain HK, Adusumilli S, Marrero JA. MR Imaging of hepatocellular carcinoma in the cirrhotic liver: challenges and controversies. Radiology 2008;247(2):311– 330. 21. Taouli B. Diffusion-weighted MR imaging for liver lesion characterization: a critical look. Radiology 2012;262(2):378–380. 22. Kang C, Qaqish B, Monaco J, Sheridan SL, Cai J. Kappa statistic for clustered dichotomous responses from physicians and patients. Stat Med 2013;32(21):3700–3719. 23. Tang A, Cruite I, Sirlin CB. Toward a standardized system for hepatocellular carcinoma diagnosis using computed tomography

10

25. Hecht EM, Holland AE, Israel GM, et al. Hepatocellular carcinoma in the cirrhotic liver: gadolinium-enhanced 3D T1-weighted MR imaging as a stand-alone sequence for diagnosis. Radiology 2006;239(2):438– 447. 26. Silvera S, Oppenheim C, Touzé E, et al. Spontaneous intracerebral hematoma on diffusion-weighted images: influence of T2shine-through and T2-blackout effects. AJNR Am J Neuroradiol 2005;26(2):236–241. 27. Kojiro M, Roskams T. Early hepatocellular carcinoma and dysplastic nodules. Semin Liver Dis 2005;25(2):133–142. 28. Sandrasegaran K, Akisik FM, Lin C, Ta hir B, Rajan J, Aisen AM. The value of diffusion-weighted imaging in characteriz-

30. Xu PJ, Yan FH, Wang JH, Shan Y, Ji Y, Chen CZ. Contribution of diffusion-weighted magnetic resonance imaging in the characterization of hepatocellular carcinomas and dysplastic nodules in cirrhotic liver. J Comput Assist Tomogr 2010;34(4): 506–512. 31. Ohtomo K, Baron RL, Dodd GD 3rd, Federle MP, Ohtomo Y, Confer SR. Confluent hepatic fibrosis in advanced cirrhosis: evaluation with MR imaging. Radiology 1993;189(3):871–874. 32. Kim T, Baron RL, Nalesnik MA. Infarcted regenerative nodules in cirrhosis: CT and MR imaging findings with pathologic correlation. AJR Am J Roentgenol 2000;175(4):1121– 1125.


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.

Non-hypervascular hypointense nodules ≥1 cm on the hepatobiliary phase of gadoxetic acid-enhanced magnetic resonance imaging in cirrhotic livers.

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

Regenerating nodules of liver cirrhosis: MR imaging with pathologic correlation.

Focal manifestations of diffuse liver disease at MR imaging.

Hypointense renal cell carcinoma: MR imaging with pathologic correlation.

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

The MR imaging diagnosis of liver diseases using gadoxetic acid: emphasis on hepatobiliary phase.

Incremental value of liver MR imaging in patients with potentially curable colorectal hepatic metastasis detected at CT: a prospective comparison of diffusion-weighted imaging, gadoxetic acid-enhanced MR imaging, and a combination of both MR techniques.

Is gadoxetic acid-enhanced MRI limited in tumor characterization for patients with chronic liver disease?

The utility of gadoxetic acid-enhanced MR imaging to characterize atypical cirrhotic nodules detected on dynamic CT images.

Differentiation of hepatocellular carcinomas from hyperplastic nodules induced in rat liver with ferrite-enhanced MR imaging.

MR imaging in hydatid disease.

Evaluation of hypointense liver lesions during hepatobiliary phase MR imaging in normal and cirrhotic livers: is increasing flip angle reliable?

MR imaging of diffuse liver disease: from technique to diagnosis.

Dermatomyositis: correlative MR imaging and P-31 MR spectroscopy for quantitative characterization of inflammatory disease.

Liver Imaging Reporting and Data System with MR Imaging: Evaluation in Nodules 20 mm or Smaller Detected in Cirrhosis at Screening US.

Receptor imaging: application to MR imaging of liver cancer.

Hypointense paranasal sinus foci: differential diagnosis with MR imaging and relation to CT findings.

Chronic kidney disease: pathological and functional assessment with diffusion tensor imaging at 3T MR.

Imaging Crohn disease: MR enterography.

Hydatid disease: MR imaging study.

Half-dose gadoxetic acid-enhanced liver magnetic resonance imaging in patients at risk for nephrogenic systemic fibrosis.

Contrast-enhanced MR imaging of the liver.