MRI findings in nonalcoholic steatohepatitis: correlation with histopathology and clinical staging

Available online at www.sciencedirect.com

Magnetic Resonance Imaging 27 (2009) 976 – 987

MRI findings in nonalcoholic steatohepatitis: correlation with histopathology and clinical staging Jorge Elias Jr. a,1 , Ersan Altun a , Steven Zacks b , Diane M. Armao c , John T. Woosley c , Richard C. Semelka a,⁎ a

Department of Radiology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA Department of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA c Department of Pathology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA Received 14 November 2008; revised 15 January 2009; accepted 4 February 2009 b

Abstract Purpose: To evaluate magnetic resonance imaging (MRI) findings of nonalcoholic steatohepatitis (NASH) and to determine the correlation of MRI findings with histopathology and Mayo End-Stage Liver Disease (MELD) score. Materials and Methods: Thirty patients (18 males, 12 females; mean age: 57±8.9 years; age range: 35–71 years) with histopathologically proven NASH who underwent MRI examinations between January 2001 and October 2005 were included in the study. Two radiologists retrospectively reviewed all magnetic resonance (MR) examinations in consensus to evaluate the presence and extent of predetermined findings of NASH including liver steatosis, early patchy liver enhancement indicating inflammation and liver fibrosis. The findings detected on MRI were correlated and compared to histopathological findings and MELD score by using nonparametric Spearman correlation coefficient and Kruskal–Wallis analysis of variance. Results: Liver steatosis was observed in 10 of 30 patients; early patchy liver enhancement, in 8 of 30 patients and liver fibrosis in 19 of 30 patients on MR images. Liver fibrosis was reticular in all these patients. There were statistically significant moderate correlations between MRI findings of liver steatosis and histopathologic grades of steatosis (r=0.43; Pb.05), and between MRI findings of fibrosis and histopathologic stages of fibrosis (r=0.61; Pb.001). Early patchy enhancement did not demonstrate statistically significant correlation with inflammation (P=.28). There was no statistically significant overall correlation between MRI findings of NASH and MELD score. Conclusion: MRI findings of liver steatosis and fibrosis in NASH showed moderate correlations with histopathologic grades of steatosis and stages of fibrosis, but MRI findings of NASH did not demonstrate any significant correlations with MELD score. © 2009 Elsevier Inc. All rights reserved. Keywords: NAFLD; NASH; MRI; histopathology

1. Introduction The prevalence of nonalcoholic fatty liver disease (NAFLD) in the general population is approximately 20% in the United States and presently recognized as the most common chronic liver disease in the United States and many parts of the world [1,2]. The prevalence of NAFLD varies between 57.5% and 74% in obese individuals [3]. NAFLD encompasses a spectrum of pathologic processes ⁎ Corresponding author. Tel.: +1 919 966 4400; fax: +1 919 966 9143. E-mail address: [email protected] (R.C. Semelka). 1 Jorge Elias Jr. is supported by CNPq-Brasilia/Brazil. 0730-725X/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2009.02.002

ranging from fatty liver, which is generally benign, to nonalcoholic steatohepatitis (NASH), which may progress to cirrhosis and its complications, including end-stage liver disease and hepatocellular carcinoma (HCC) [1–4]. While many patients with NAFLD have a benign course, approximately 3% of nonobese patients and 15% to 20% of obese patients have hepatocyte necrosis and inflammation with or without fibrosis, reflecting the presence of NASH [1,3,5]. Cirrhosis develops in 3–10% of the patients with NASH [6]. Histopathological evaluation is the most accurate method for the diagnosis of NASH, which is based on the assessment of macrovesicular steatosis, necroinflammatory activity and fibrosis [3,5,7].

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

Imaging studies such as ultrasound and computerized tomography (CT) are useful in demonstrating and quantifying steatosis; however, distinction between simple fatty liver and NASH cannot be made by these imaging modalities [3,6,8–12]. The evaluation of hepatic steatosis is often currently done by magnetic resonance imaging (MRI), as fat evaluation is more specific than with either ultrasound or CT, and the most commonly used MRI technique to quantify and to confirm the presence of fat in tissues is chemical-shift imaging [2,6,12–17]. Additionally, MRI has become an increasingly valuable imaging modality in the setting of diffuse liver diseases including NASH [2,3,6,12–21]. In patients with chronic hepatitis, specific early and late enhancement patterns of the liver revealed on gadolinium-enhanced dynamic magnetic resonance (MR) images demonstrate a high degree of correlation with histopathologic findings [19–21]. The presence of early patchy enhancement is indicative of hepatocellular necrosis and inflammatory activity, whereas signs of late linear or amorphous enhancement denote the presence of fibrosis [19–21]. To our knowledge, the full range of MRI findings of NASH including fatty changes, inflammation and fibrosis has not been reported in the literature. The purpose of this study is to evaluate MRI findings of NASH and to determine the correlation of MRI findings with histopathology and the Mayo End-Stage Liver Disease (MELD) score, which is used to assess the clinical severity the disease. 2. Material and methods 2.1. Patient selection Institutional review board approval with waiver of informed consent was obtained for this Health Insurance Portability and Accountability Act compliant retrospective study. Transplant surgery, hepatology, pathology and radiology department databases were searched and crossreferenced to identify all patients with histopathologically proven NASH who underwent MRI examination between the dates of January 2001 and January 2005. The final study population included 18 males (mean age of 57.4±7.5 years; range 45–70 years) and 12 females (mean age of 56.6±11.1 years; range 35–71 years). In all patients, the diagnosis of NASH was made on the basis of the presence of (a) histopathologic evidence of NASH, (b) no current excessive alcohol use as defined by average, daily consumption of less than 20 g, (c) no past history of excessive alcohol consumption and (d) negative serologic testing for viral etiologies of hepatitis [5]. Patients with the established diagnosis of cirrhosis secondary to NASH were included only if they had met the first criterion previously during the progress of the disease. None of the patients had clinical, imaging and histopathological findings of iron deposition in the liver.

977

2.2. MR imaging technique All MR examinations were performed on 1.5-T MR systems (Vision, Sonata or Avanto, Siemens Medical Solutions, Malvern, PA, USA) using a phased-array torso coil. Parameters of precontrast and postcontrast sequences are displayed in Table 1. Postcontrast imaging was performed with power-injected (Medrad, Pittsburgh, PA, USA) bolus of 0.1 mmol/kg gadolinium chelate (Omniscan, Oakville, Ontario, Canada) at 2 ml/s followed by a bolus 20 ml of saline flush in all patients. During the study period, our clinical practice has been to acquire spoiled gradient echo (SGE) sequences on the hepatic arterial dominant phase on our 1.5-T systems, which were relatively old, because of perceived higher image quality and intrinsic soft tissue contrast resolution compared to three-dimensional, gradient-echo (3D-GE) sequences. Additionally, fat suppression was not used during SGE acquisitions on the hepatic arterial dominant phase in order to increase the number of slices acquired. 2.3. MR image interpretation All MR examinations were retrospectively reviewed in consensus by two experienced abdominal radiologists (15 and 20 years of experience). The reviewers were aware that the patients had NASH but they were blinded to the clinical information and histopathologic data. When a patient had more than one MR examination, the examination which had been closest to the date of histopathological analysis was evaluated. The two readers evaluated the MR examinations for (a) steatosis, (b) patchy enhancement pattern of liver parenchyma, (c) fibrosis and (d) other hepatic and extrahepatic morphologic features. Liver steatosis was evaluated subjectively as present or absent by observing signal loss on out-of-phase images in comparison to in-phase images. Additionally, liver steatosis was evaluated objectively by obtaining the fat fraction as described by Fishbein et al [22]. A pair of in- and out-ofphase images at the correspondent level of the main portal vein was utilized to perform the signal intensity (SI) measurements, which were obtained with a standard technique using region of interest (ROI) measurements in four representative segments of the liver including right lobe anterior and posterior and left lobe medial and lateral segments. ROIs, which were placed in areas devoid of blood vessels, motion artifacts or partial volume effects, were measured to be of the maximal size in each segment and they were not less than 1 cm2. The areas of ROIs were constant for each segment between the sequences intra-individually. Mean SI levels for each ROI were recorded for each liver segment evaluated and then an average SI of all segments was calculated for each image. The fat hepatic fraction was calculated from the average SI data for each image using the formula: fat fraction=(SIin-phase−SIout-of-phase)/2SIin-phase. The presence of patchy enhancement of the liver parenchyma on early post-gadolinium imaging which fades

TR, Repetition time; TE, echo time; FOV, field of view; RARE, rapid acquisition with relaxation enhancement; STIR, Short tau inversion recovery. a Rectangular FOV. b Postcontrast SGE sequence was acquired at 18 seconds on the hepatic arterial dominant phase, and three dimensional gradient echo (3D-GE) sequences were acquired at 60 and 120 seconds on portal venous and interstitial phases.

19 19 Breath-hold Breath-hold 1 1 128×256 144×320 350 (75%) 360 (80%) 1.6 0.7 8 3.5 80 10 Not used Used Transverse Transverse

140/4.4 4.3/1.7

490 350

1 118×256 350 (72%) 1.6 8 300 150 3830/64 Used Transverse

180 ∞/90

651

8

1.6

400 (81%)

192×256

1

Breathing32 independent Breathing32 independent Breath-hold 40 1 400 (81%-100%) 192×256 1.6 8 180

Transverse/ Not used coronal Transverse Used

∞/90

651

19 Breath-hold 1 128×256 350 (75%) 1.6 8 80 140/2.2

490

Breath-hold 1 128×256 350 (75%) a 1.6 8 490 80 140/4.4

Transverse/ Not used coronal Transverse Not used

Precontrast T1-weighted in-phase SGE T1-weighted out-of-phase SGE T2-weightedhalfFourier RARE T2-weighted HalfFourier RARE STIR Postcontrast b T1-weighted SGE T1-weighted 3D-GE

Matrix size No. of signals Respiratory averaged control FatTR/TE (msec) Flip Angle (°) Bandwidth/ Section Thickness Interslice gap FOV (mm) suppression pixel (Hz) (mm) (mm) Plane Sequences

Table 1 Parameters of precontrast and postcontrast sequences

19

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987 Acquisition time (s)

978

or tends to fade to isointensity with the remaining liver parenchyma on delayed post-gadolinium imaging was evaluated and defined as the indicator of necroinflammatory activity [19,21]. The imaging pattern of liver fibrosis was characterized as absent, reticular, confluent or both reticular and confluent. Reticular fibrosis was defined as fine lines that are hypointense on short TE out-of-phase imaging and/or show prominent enhancement on delayed post-gadolinium MR images. A four-point scoring system was used to evaluate the extent of reticular fibrosis (0: none, 1: mild, 2: moderate, 3: severe). We defined mild reticular fibrosis as a fine network of linear fibrous tissue with a thickness less than 2 mm without obvious liver surface nodularity, moderate reticular fibrosis as linear fibrotic strands measuring between 2 and 5 mm with liver surface nodularity caused by intervening bands of fibrosis and severe reticular fibrosis as thick fibrotic strands measuring greater than 5 mm [20]. Confluent fibrosis was defined as a region of amorphous fibrosis that is isolated or associated with reticular fibrosis, greater than 2 cm in diameter and shows the same pre- and post-contrast MR characteristics as reticular fibrosis [20]. Other morphologic liver findings that were assessed included liver volume, modified caudate–right lobe (C-RL) ratio, surface irregularity and the presence of hypovascular and hypervascular nodules [22–26]. Liver volume was determined by manually tracing the liver boundary on consecutive sections of a data acquisition encompassing the entire liver. The modified C-RL ratio was obtained as described by Awaya et al. [23]. Surface nodularity was assessed by observing irregularity along the liver surface. Liver nodules were determined by using previously reported criteria [27]. Additional criteria embraced the evaluation of findings of portal hypertension including varices, ascites, splenomegaly (defined as greater than 13 cm in maximum dimension); portal vein thrombosis; and periportal and portocaval lymphadenopathy (short axis greater than 1 cm). 2.4. Correlation with histopathological analyses and MELD score evaluation Histopathological analyses included liver explant evaluation in 10 patients and needle liver biopsy evaluations in the remaining 20 patients. All histopathologic evaluations were performed according to the Brunt et al. [7] grading and staging scheme which includes the grading of macrovesicular steatosis and necroinflammatory activity and staging of fibrosis (Table 2). While comparing MRI findings and histopathological findings, the discrepancies resulting from the comparison of microscopic findings to macroscopic findings were minimized by simplifying classifications for grade of steatosis and stage of fibrosis. However, because of many criteria affecting the assessment of necroinflammatory

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987 Table 2 Histopathological evaluation: grading and staging of NASH Grading of NASH for steatosis and steatohepatitis 1. Macrovesicular steatosis Grade 0 None Grade 1 (mild) Up to 33% Grade 2 (moderate) 33–66% Grade 3 (severe) N66% 2. Necroinflammatory activity Grade 1 (mild) Steatosis up to 66%, occasional ballooned hepatocyte (mainly zone 3), scattered intra-acinar neutrophils (PMN)±lymphocytes, no or mild portal inflammation Grade 2 (moderate) Steatosis of any degree, obvious zone III ballooning degeneration, intra-acinar PMNs, zone III perisinusoidal fibrosis may be present, mild to moderate, portal and intra-acinar inflammation Grade 3 (severe) Panacinar steatosis, widespread ballooning, intra-acinar inflammation, PMNs associated with ballooned hepatocytes, mild to moderate portal inflammation Staging of NASH for fibrosis Stage 1 Zone III perisinusoidal/pericellular fibrosis; focally or extensively present Stage 2 Zone III perisinusoidal/pericellular fibrosis with focal or extensive periportal fibrosis Stage 3 Zone III perisinusoidal/pericellular fibrosis and portal fibrosis with focal or extensive bridging fibrosis Stage 4 Cirrhosis PMN, polymorphonuclear leukocytes.

activity at the microscopic level which does not have direct macroscopic correlates, a simplified classification could not be established for this criterion. Patients were classified in one of three groups according to grade of steatosis: Group 1 included the patients with Grade 0 steatosis, Group 2 included the patients with Grade 1 steatosis and Group 3 included the patients with Grade 2 and 3 steatosis. Patients were classified in one of three groups according to stage of fibrosis: Group 1 included the patients with no fibrosis and stage 1 fibrosis, Group 2 included the patients with Stage 2 and 3 fibrosis, and Group 3 included the patients with Stage 4 fibrosis. The MELD score was used to measure the clinical severity of NASH. The MELD score is calculated on the basis of the total serum bilirubin level, the creatinine level and the international normalized ratio for prothrombin time [22]. Patients were classified into one of three groups according to the MELD score: group 1 had MELD scores lower than 10, group 2 had scores of 10–19 and group 3 had scores higher than 19 [28]. The mean time interval between the MR examinations and the histopathologic evaluations was 71.7±50.8 days, ranging from 2 to 180 days. The mean interval between the MR examinations and the determination of MELD scores was 7.2±16.1 days, ranging from 0 to 60 days.

979

2.5. Statistical analyses In the analysis of patient characteristics, data are expressed as mean±S.D. The prevalence of MR imaging findings was estimated as a percentage of the patients displaying each abnormality. Correlations between MR findings, histopathologic grades and stages of findings and MELD scores were analyzed using the Spearman coefficient and Fisher's Exact Test. The Kruskal-Wallis analysis of variance was used to evaluate the differences between MR findings when compared across groups of histopathologic findings and MELD score groups. All P values were derived from two-tailed tests, and a level of less than .05 was accepted as statistically significant.

3. Results 3.1. MRI findings Hepatic steatosis was observed in 10 (33.3%) of 30 patients on the basis of the subjective evaluation of the decrease in signal intensity on out-of-phase MR images. The mean value of the hepatic fat fraction was 9.2%±10%, ranging from 0.32% to 40.8%. Patchy liver enhancement (Fig. 1) on early post-gadolinium images was observed in 8 (26.7%) of 30 patients. Of the 30 patients, 11 (36.7%) had no imaging findings of fibrosis. Reticular fibrosis was observed in 19 patients (63.3%) with a mean grade of 2.05±0.77 while confluent fibrosis was not observed in any patient. Mild fibrosis (Fig. 2) was diagnosed in 5 (26.3%) of 19 patients, moderate fibrosis (Fig. 3) in 8 (42.1%) of 19 patients and severe fibrosis (Fig. 4) in 6 (31.6%) of 19 patients. Calculated liver volumes ranged between 943.9 and 2501 ml, with a mean volume of 1510±422 ml. The modified C-RL ratio ranged between 0.32 and 0.96, with a mean of 0.59±0.19. Liver surface irregularity was observed in 16 (53.3%) of 30 patients. Hypervascular nodules were observed in 2 (6.7%) of 30 patients — one diagnosed as a hepatocellular carcinoma and the other as a high grade dysplastic nodule. Findings of portal hypertension included varices in 19 (63.3%) of 30 patients, ascites in 18 (60%) of 30 and splenomegaly in 24 (80%) of 30. Portal vein thrombosis was observed in 2 (6.7%) of 30 patients — partial thrombosis in one and total in the other. Lymphadenopathy was observed in 1 (3.3%) of 30 patients. 3.2. Histopathologic analyses Of 30 patients, 5 (16.7%) had no steatosis, 13 (43.3%) had mild steatosis, 9 (30%) had moderate steatosis and 3 (10%) had severe steatosis. Considering the grade of steatosis groups classification, 5 of 30 patients were classified as Group 1, 13 of 30 patients as Group 2 and

980

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

Fig. 1. Transverse T1-weighted (A) in-phase SGE, (B) out-of-phase SGE, postgadolinium (C) hepatic arterial dominant phase SGE and (D) interstitial phase fatsuppressed SGE images in a 38-year-old female patient with NASH who had Grade 2 steatosis, Grade 2–3 necroinflammatory activity and Stage 2–3 fibrosis histopathologically. Diffuse fatty infiltration of the liver which shows signal drop on (B) out-of-phase SGE image compared to (A) in-phase SGE image is detected. The liver demonstrates diffuse patchy enhancement on (C) hepatic arterial dominant phase SGE, which fades to isointensity with the remaining liver parenchyma on (D) interstitial phase fat-suppressed SGE images. Note that no fibrosis is seen on MR images and the liver is larger than normal. The time interval between histopathologic evaluation and MR examination was 2 days.

12 of 30 patients as Group 3. Of 30 patients, 12 (40%) had no, 9 (30%) had mild, 6 (20%) had moderate and 3 (10%) had severe necroinflammatory activity. Of 30 patients, 1 (3.3%) had no fibrosis, 3 (10%) had Stage 1 fibrosis, 3 (10%) had Stage 2 fibrosis, 6 (20%) had Stage 3 fibrosis and 17 (56.7%) had Stage 4 fibrosis. Considering the stage of fibrosis groups classification; 4 of 30 patients were classified as Group 1, 9 of 30 as Group 2 and 17 of 30 as Group 3. 3.3. MELD score evaluation The distribution of patients according to MELD score showed that 11 (36.7%) of 30 patients were classified as Group 1 with a mean score of 5.27±2.72; 13 (43.3%) of 30 patients were classified as Group 2 with a mean score of 13.69±2.49 and 6 (20%) of 30 patients were classified as Group 3 with a mean score of 22.67±4.13.

3.4. Correlation of MRI and histopathologic findings There was a statistically significant correlation between the subjective MR evaluation of steatosis and the histopathologic grades of steatosis (r=0.49; 95% confidence interval: 0.15–0.73; Pb.01), as it was also observed with the hepatic fat fraction (r=0.43; 95% confidence interval: 0.07–0.69; Pb.05). There were statistically significant differences between subjective steatosis, hepatic fat fraction, mean liver volumes and mean modified C-RL ratio evaluations on MRI across the groups of histopathologic grade of steatosis (Pb.01, .01, .05 and .05; respectively) (Table 3). No significant differences were observed between patchy liver enhancement, fibrosis and liver surface irregularity frequency evaluations on MRI across the groups of histopathologic grade of steatosis (P=.79, .08 and .19; respectively) (Table 3). There was no correlation between the presence or absence of patchy liver enhancement pattern and the grade

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

981

Fig. 2. Transverse T1-weighted (A) in-phase spoiled gradient echo (SGE), (B) out-of-phase SGE, postgadolinium (C) hepatic arterial dominant phase SGE and (D) interstitial phase fat-suppressed SGE images in a 52-year-old female patient with NASH who had Grade 2 steatosis, Grade 1 necroinflammatory activity and Stage 1 fibrosis histopathologically. Diffuse fatty infiltration of the liver which shows signal drop on (B) out-of-phase SGE image compared to (A) in-phase SGE image is detected. Hepatic segments 4 and 2–3 demonstrates patchy enhancement (arrows) on (C) hepatic arterial dominant phase SGE, which fades to isointensity with the remaining liver parenchyma on (D) interstitial phase fat-suppressed SGE images. Note that mild reticular fibrosis is seen on (D) interstitial phase fat-suppressed SGE image and the liver is larger than normal. The time interval between histopathologic evaluation and MR examination was 2 days.

of necroinflammatory activity as evaluated by histopathology (P=.28). There was a statistically significant correlation between MRI findings of liver fibrosis and the histopathologic stages of fibrosis (r=0.61; 95% confidence interval: 0.31– 0.80; Pb.001), which was also observed as a statistically significant difference between MRI findings of fibrosis across the groups of histopathologic stage of fibrosis (Pb.05) (Table 4). There were also statistically significant differences between liver contour irregularity frequencies, mean liver volumes and mean modified C-RL ratios evaluated on MRI across the groups of histopathologic stage of fibrosis (Pb.01, .05 and .01; respectively)

(Table 4). There was no significant difference between subjective steatosis, hepatic fat fraction and patchy liver enhancement pattern evaluations on MRI across the groups of histopathologic stage of fibrosis (P=.43, .05 and .35; respectively) (Table 4). 3.5. Correlation of histopathologic findings with MELD score There was a statistically significant inverse correlation between the histopathologic grades of steatosis and the mean MELD scores (r=−0.41; 95% confidence interval: −0.67 to −0.05; Pb.05). There was no significant difference between

982

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

Fig. 3. Transverse T1-weighted (A) in-phase spoiled gradient echo (SGE), (B) out-of-phase SGE, postgadolinium (C) hepatic arterial dominant phase SGE and (D) interstitial phase fat-suppressed SGE images in a 58-year-old female patient with NASH who had Grade 1 steatosis, Grade 1 necroinflammatory activity and Stage 4 fibrosis histopathologically. There is no fatty infiltration appreciated as signal drop on (B) out-of-phase SGE image compared to (A) inphase SGE image. There is also no patchy enhancement on (C) hepatic arterial dominant phase SGE image. Note that moderate reticular fibrosis together with liver contour irregularities is seen on (D) interstitial phase fat-suppressed SGE image. The time interval between histopathologic evaluation and MR examination was 3 months.

the mean MELD scores across the groups of histopathologic grade of steatosis (P=.06) (Table 3). There was a statistically significant correlation between the histopathologic stages of fibrosis and the mean MELD scores (r=0.54; 95% confidence interval: 0.21–0.76; Pb.01). There was a statistically significant difference between the mean MELD scores across the groups of histopathologic stage of fibrosis (Pb.05) (Table 4).

subjective steatosis (P=.52), hepatic fat fraction evaluations (P=.06), patchy liver enhancement patterns (P=.25), fibrosis (P=.13), liver surface irregularities (P=.10), liver volumes (P=.37) and modified C-RL ratios (P=.16) on MRI across the MELD score groups (Table 5).

3.6. Correlation of MR findings with MELD score

NASH was originally described to be more prevalent in females [3,5], although most recent series have reported a higher prevalence in males [3,5], which was observed in our study with a male-to-female ratio of 1.5:1. NASH has been reported as most common in middle-age individuals [3,5],

There was no significant correlation between MRI findings of fibrosis and the MELD score groups (P=.13). There was no significant difference between findings of

4. Discussion

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

983

Fig. 4. Transverse T1-weighted (A) in-phase spoiled gradient echo (SGE), (B) out-of-phase SGE, postgadolinium (C) hepatic arterial dominant phase SGE and (D) interstitial phase fat-suppressed SGE images in a 62-year-old female patient with NASH who had Grade 2 steatosis, Grade 1 necroinflammatory activity and Stage 4 fibrosis histopathologically. Focal patchy fatty infiltration of the liver which shows signal drop on (B) out-of-phase SGE image compared to (A) in-phase SGE image is detected. The liver demonstrates diffuse patchy enhancement on (C) hepatic arterial dominant phase SGE, which tends to fade to isointensity with the remaining liver parenchyma on (D) interstitial phase fat-suppressed SGE images. Note that severe reticular fibrosis together with liver contour irregularities is seen on (D) interstitial phase fat-suppressed SGE image. The time interval between histopathologic evaluation and MR examination was 3 months.

which we also observed in our study with the mean patient age being 57±8.9 years. It has generally been accepted that liver biopsy is the only reliable method for the differentiation of NASH from hepatic

steatosis and the determination of extent of steatosis, necroinflamatory activity and fibrosis [3,7]. The drawbacks of using liver biopsy in the diagnosis of NASH and assessment of its severity include its sampling error and

Table 3 Correlation between the groups of histopathologic grades of steatosis and MRI evaluation for steatosis, liver enhancement, fibrosis, liver surface irregularities, liver volume and modified C-RL ratio, and correlation between the groups of histopathologic degrees of steatosis and MELD score Grade of steatosis–histopathology

MRI evaluation

a

Number of patients.

Subjective steatosis evaluation Fat hepatic fraction (mean %±S.D.) Patchy liver enhancement Fibrosis (mean±S.D.) Liver surface irregularity Liver volume (ml) (mean±S.D.) Modified C-RL ratio (mean±S.D.) MELD score (mean±S.D.)

Group 1 (n=5) a

Group 2 (n=13)

Group 3 (n=12)

P

1/5 (20%) 5.26%±2.59% 1/5 (20%) 1.00±0.70 3/5 (60%) 1408±554.7 0.63±0.15 14.80±4.08

1/13 (7.6%) 3.81%±3.05% 3/13 (23%) 1.85±1.21 9/13 (69.2%) 1350±418.9 0.70±0.18 14.69±8.15

8/12 (66.7%) 17.08%±12.28% 4/12 (33.3%) 0.83±1.11 4/12 (33.3%) 1713±297.8 0.48±0.15 8.91±5.55

b.01 b.01 .79 .08 .19 b.05 b.05 .06

984

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

Table 4 Correlation between the groups of histopathologic stage of fibrosis and MRI evaluation for steatosis, liver enhancement, fibrosis, liver surface irregularity, liver volume, and modified C-RL ratio; and correlation between the groups of histopathologic stage of fibrosis and MELD score Fibrosis stage–histopathology

MRI evaluation

a

Subjective steatosis evaluation Fat hepatic fraction (mean %±S.D.) Patchy liver enhancement Fibrosis (mean±S.D.) Liver surface irregularity Liver volume (ml) (mean±S.D.) Modified C-RL ratio (mean±S.D.) MELD score (mean±S.D.)

Group 1 (n=4) a

Group 2 (n=9)

Group 3 (n=17)

P

2/4 (50%) 15.73%±16.78% 0/4 (0%) 0 0/4 (0%) 1973±302.4 0.38±0.09 8.25±5.12

4/9 (44.4%) 13.44%±11.45% 2/9 (22.2%) 1.00±1.32 3/9 (33.3%) 1602±542.7 0.53±0.18 8.33±6.38

4/17 (23.5%) 4.97%±4.39% 6/17 (35.2%) 1.76±0.97 13/17 (76.4%) 1343±255.1 0.69±0.15 15.53±6.41

.43 .05 .35 b.05 b.01 b.05 b.01 b.05

Number of patients.

lethal complications [12,29,30]. Because the biopsy procedure is usually not targeted and the biopsy sample represents only 1/50.000th of the liver, sampling error is a critical issue in diseases of heterogeneous distribution, such as NASH [12,29]. Liver biopsy is also an invasive procedure and carries a potential risk for lethal complications (0.009– 0.12%) [30]. MRI is a noninvasive and comprehensive test which has the potential for the diagnosis and assessment of NASH. The role of MR imaging in the evaluation of NAFLD including NASH is evolving and it has been mainly directed to steatosis quantification [2,6,13,22]. However, in our study, we analyzed the full spectrum of MRI findings in correlation with histopathological analyses. In our study, the correlation coefficients calculated for MRI findings of steatosis and histopathologic grades of steatosis (r=0.43, 0.49) is in the range of correlation coefficients (r=0.25–0.78) reported in the literature [2,13]. The correlation coefficient calculated for MRI findings of fibrosis and histopathologic stages of fibrosis (r=0.61) is also in the range of correlation coefficients (r=0.473– 0.638) reported in the literature [31,32]. There was no correlation between early patchy enhancement on MRI and histopathologic changes of acute inflammation and hepatocellular necrosis. We believe that a combination of factors including the relatively high number of patients

with cirrhosis, the absence of standardization and targeting of biopsy location and long time interval between MRI and histopathologic specimen acquisitions probably prevented us from reaching higher extent of correlations for steatosis and fibrosis and obtaining significant correlation for inflammation. Since a proportion of our patient population was drawn from the databases of Transplant Surgery and Pathology Departments, more advanced disease was present in our patients, reflecting the established presence of NASH. Prior reports have shown a significant reduction and disappearance of steatosis as the disease process progresses with increasing fibrosis in NASH [5,33]. In our study, the mean hepatic fat fraction demonstrated a decrease across the groups of histopathologic stage of fibrosis, concurring with previous descriptions of decrease in hepatic fat content as fibrosis and cirrhosis progress [5,33]. This finding did not, however, demonstrate statistical significance in our study, which we believe reflects the small size of the study population, but a trend was shown with the P value equal to .05. No prior study has attempted to correlate the MELD score with histopathologic and MR findings of NASH. In our study, the correlation between MRI findings and MELD scores was not significant, while the MELD scores showed a significant difference across the histopathologic stages of

Table 5 Correlation between MELD groups and MRI evaluation for steatosis, liver enhancement, fibrosis, liver surface irregularities, liver volume, modified CR-L ratio and steatosis MELD groups

MRI evaluation

a

Number of patients.

Subjective steatosis evaluation Fat hepatic fraction (mean %±S.D.) Patchy liver enhancement Fibrosis (mean±S.D.) Liver surface irregularity Liver volume (ml) (mean±S.D.) Modified C-RL ratio (mean±S.D.)

b10 (n=11) a

10 to 19 (n=13)

N19 (n=6)

P

5/11 (45.4%) 13.68%±12.76% 1/11 (9%) 0.81±1.25 3/11 (27.2%) 1657±475.2 0.50±0.14

3/13 (23%) 3.88%±3.72% 5/13 (38.5%) 1.38±0.96 9/13 (69.2%) 1446±387.1 0.63±0.19

2/6 (33.3%) 9.60%±10.48% 2/6 (33.3%) 2.00±1.36 4/6 (66.7%) 1404±404.9 0.67±0.20

.52 .06 .25 .13 .10 .37 .16

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

fibrosis. We believe this discrepant observation again probably reflects the combination of aforementioned limitations of the study. The morphologic liver evaluation showed a progressive significant increase in liver surface irregularity, a progressive significant liver volume reduction and a progressive significant modified C-RL ratio increase, in correlation with worsening of histopathologic fibrosis stage, which concurs with the development of cirrhosis in the setting of NASH [5,34,35]. The mean modified C-RL ratio for precirrhotic NASH patients was 0.49±0.17, which is similar to that reported by Oliva et al. [34] of 0.43 in NASH patients evaluated by CT. Interestingly, the mean modified C-RL ratio for cirrhotic patients in our study was 0.69±0.15, which is below the cutoff value of 0.90 as reported by Awaya et al. [23] to achieve high accuracy in the diagnosis of cirrhosis. Our belief is that this reflects the limitation of using the modified C-RL ratio to determine the presence of cirrhosis as various forms of cirrhosis may result in varying patterns of morphologic change. For illustration, a prior report on autoimmune hepatitis did not observe a prominent modified C-RL ratio [20]. Our results showed a relatively low incidence of hypervascular nodules in NASH patients, with only one patient having an HCC (3.3%). This was in the range of reported incidences (0–27%) for HCC development in NASH patients [4,36]. The evaluation of ancillary features showed a high incidence of portal hypertension findings, which were higher than the incidence of histopathologically proven cirrhosis. This may reflect that portal hypertension arises in a precirrhotic state. The incidence of portal vein thrombosis was similar to prior reports [37]. Lymphadenopathy however, had a very low incidence in our study, which is lower than observed in other etiologies such as autoimmune hepatitis [20], viral hepatitis [38], primary sclerosing cholangitis [39], and primary biliary cirrhosis [40]. A recent report described a higher lymphadenopathy incidence in NASH patients (77.8%) [34]. This may reflect that our population included both precirrhotic and cirrhotic patients, while they included precirrhotic NASH patients only. A larger prospective study would better define the correlation of MRI and histopathological findings with the MELD score and whether early patchy enhancement correlates with hepatocellular necrosis and inflammation. It is, however, interesting that prior MRI reports on primary sclerosing cholangitis [39] and autoimmune hepatitis [20] also did not show good correlation with MELD. We believe that this may not invalidate the importance of MRI findings but may suggest that MRI provides another measure of the extent of liver diseases that emphasizes morphology and extent of fibrosis. The findings of our study do not suggest that MRI may replace liver biopsy in the evaluation of NASH. However, it shows that MRI has a strong potential for the evaluation of NASH, particularly with the contribution of new MRI

985

techniques [6]. Three-point Dixon, IDEAL or MR spectroscopy techniques may quantify the fat fraction in the liver better [6,41–44]. While MR spectroscopy may be used in the detection of inflammation [6], MR elastography, diffusion weighted MRI and perfusion techniques may be employed to determine the extent of fibrosis [6,45–47]. However, further research is necessary for the establishment of roles of these new techniques in the evaluation of NASH. The limitations of this study included its small sample size and retrospective nature. Another limitation was the untargeted acquisition of histopathologic specimens with a needle biopsy in 66.7% of the patients, which might result in substantial sampling error in terms of degree of steatosis and severity of fibrosis and inflammation. Another limitation of the study was the timing of MRI studies in relation to histopathological evaluation, which was not standardized to a short interval, reflecting the retrospective nature of the study. This was also a limitation for a number of studies in the literature as well [13,14,32]. We believe this probably contributed to the absence of correlation between patchy liver enhancement on MRI and necroinflammatory activity by histopathology. Another limitation was the ambiguity error of using in-phase and out-of-phase imaging to quantify fat [14]. This leads to the underestimation of degree of steatosis when the fat fraction is greater than the water fraction in the liver [14]. Another limitation of the hepatic steatosis evaluation used in our study is that the subjective evaluation of parenchymal signal and the objective measurement of signal intensities on in-phase and out-of-phase images can be affected by iron deposition and do not reflect the fat content correctly. However, this was not a problem in our study because none of the patients had clinical, imaging and histopathologic findings for iron deposition. The other limitation was the acquisition of thicker slices on the hepatic arterial dominant phase with non–fat-suppressed SGE compared to fat-suppressed 3D-GE sequences. SGE was used because of its higher image quality and soft tissue contrast resolution on our relatively old MR systems; therefore, we do not believe that the use of SGE sequence had substantial effect on our evaluation. Additionally, fatsuppression techniques were not used with SGE sequence to increase the number of slices acquired, and fat-suppression effects were evaluated on other fat-suppressed sequences. However, fat-suppressed 3D-GE sequences should be used for all phases of postcontrast imaging on the state of art MR systems. In conclusion, MRI findings of liver steatosis and fibrosis in NASH showed moderate correlations with histopathologic grades of steatosis and stages of fibrosis, but MRI findings of NASH did not demonstrate any significant correlations with MELD score. With the contribution of new MRI techniques, MRI may be used as a noninvasive and comprehensive diagnostic method in the assessment of NASH; however, further research is needed to establish the role of these new MRI techniques.

986

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987

References [1] Wieckowska A, Feldstein AE. Diagnosis of nonalcoholic fatty liver disease: invasive versus noninvasive. Semin Liver Dis 2008;28: 386–95. [2] Bahl M, Qayyum A, Westphalen AC, et al. Liver steatosis: investigation of opposed-phase T1-weighted liver MR signal intensity loss and visceral fat measurement as biomarkers. Radiology 2008;249:160–6. [3] Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002;346: 1221–31. [4] Iannaccone R, Piacentini F, Murakami T, et al. Hepatocellular carcinoma in patients with nonalcoholic fatty liver disease: helical CT and MR imaging findings with clinical-pathologic comparison. Radiology 2007;243:422–30. [5] Sanyal AJ, American Gastroenterological Association. AGA technical review on nonalcoholic fatty liver disease. Gastroenterology 2002;123: 1705–25. [6] Lall CG, Aisen AM, Bansal N, Sandrasegaran K. Nonalcoholic fatty liver disease. AJR 2008;190:9931002. [7] Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-Tetri BA, Bacon BR. Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94:2467–74. [8] Lee SW, Park SH, Kim KW, et al. Unenhanced CT for assessment of macrovesicular hepatic steatosis in living liver donors: comparison of visual grading with liver attenuation index. Radiology 2007;244: 479–85. [9] Park SH, Kim PN, Kim KW, et al. Macrovesicular hepatic steatosis in living liver donors: use of CT for quantitative and qualitative assessment. Radiology 2006;239:105–12. [10] Limamond P, Raman SS, Lassman C, et al. Macrovesicular hepatic steatosis in living related liver donors: correlation between CT and histologic findings. Radiology 2004;230:276–80. [11] Strauss S, Gavish E, Gottlieb P, Katsnelson L. Interobserver and intraobserver variability in the sonographic assessment of fatty liver. AJR 2007;189:W320–3. [12] Mehta SR, Thomas EL, Bell JD, Johnston DG, Taylor-Robinson SD. Non-invasive means of measuring hepatic fat content. World J Gastroenterol 2008;14:3476–83. [13] Qayyum A, Goh SJ, Kakar S, Yeh BM, Merriman RB, Coakley FV. Accuracy of liver fat quantification at MR imaging: comparison of outof-phase gradient-echo and fat-saturated fast spin-echo techniques — initial experience. Radiology 2005;237:507–11. [14] Kim SH, Lee JM, Han JK, et al. Hepatic macrosteatosis: predicting appropriateness of liver donation by using MR imaging — correlation with histopathologic findings. Radiology 2006;240:116–29. [15] Hamer OW, Aguirre DA, Casola G, Lavine JE, Woenckhaus M, Sirlin CB. Fatty liver: imaging patterns and pitfalls. Radiographics 2006;26: 1637–53. [16] Hamer OW, Aguirre DA, Casola G, Sirlin CB. Imaging features of perivascular fatty infiltration of the liver: initial observations. Radiology 2005;237:159–69. [17] Yoshimitsu K, Kuroda Y, Nakamuta M, et al. Noninvasive estimation of hepatic steatosis using plain CT vs. chemical-shift MR imaging: significance for living donors. J Magn Reson Imaging 2008;28:678–84. [18] Ramalho M, Altun E, Heredia V, Zapparoli M, Semelka RC. Liver MR imaging: 1.5T versus 3T. Magn Reson Imaging Clin N Am 2007;15: 321–47. [19] Semelka RC, Chung JJ, Hussain SM, Marcos HB, Woosley JT. Chronic hepatitis: correlation of early patchy and late linear enhancement patterns on gadolinium-enhanced MR images with histopathology initial experience. J Magn Reson Imaging 2001;13:385–91. [20] Bilaj F, Hyslop WB, Rivero H, et al. MR imaging findings in autoimmune hepatitis: correlation with clinical staging. Radiology 2005;236:896–902. [21] Kanematsu M, Danet MI, Leonardou P, et al. Early heterogeneous enhancement of the liver: magnetic resonance imaging findings and clinical significance. J Magn Reson Imaging 2004;20:242–9.

[22] Fishbein M, Castro F, Cheruku S, et al. Hepatic MRI for fat quantitation: its relationship to fat morphology, diagnosis, and ultrasound. J Clin Gastroenterol 2005;39:619–25. [23] Awaya H, Mitchell DG, Kamishima T, Holland G, Ito K, Matsumoto T. Cirrhosis: modified caudate-right lobe ratio. Radiology 2002;224: 769–74. [24] Dodd III GD, Baron RL, Oliver III JH, Federle MP. Spectrum of imaging findings of the liver in end-stage cirrhosis: Part II, focal abnormalities. AJR Am J Roentgenol 1999;173:1185–92. [25] Dodd III GD, Baron RL, Oliver III JH, Federle MP. Spectrum of imaging findings of the liver in end-stage cirrhosis: part I, gross morphology and diffuse abnormalities. AJR Am J Roentgenol 1999; 173:1031–6. [26] Ito K, Mitchell DG. Hepatic morphologic changes in cirrhosis: MR imaging findings. Abdom Imaging 2000;25:456–61. [27] Krinsky GA, Lee VS. MR imaging of cirrhotic nodules. Abdom Imaging 2000;25:471–82. [28] Kremers WK, van IM, Kim WR, et al. MELD score as a predictor of pretransplant and posttransplant survival in OPTN/UNOS status 1 patients. Hepatology 2004;39:764–9. [29] Ratziu V, Charlotte F, Heurtier A, et al. Sampling variability of liver biopsy in nonalcoholic fatty liver disease. Gastroenterology 2005;128: 1898–906. [30] Sheela H, Seela S, Caldwell C, Boyer JL, Jain D. Liver biopsy: evolving role in the new millennium. J Clin Gastroenterol 2005;39: 603–10. [31] Kato H, Kanematsu M, Zhang X, et al. Computer-aided dianosis of hepatic fibrosis: preliminary evaluation of MRI texture analysis using the finite difference method and an artificial neural network. AJR 2007;189:117–22. [32] Aguirre DA, Behling CA, Alpert E, Hassanein TI, Sirlin CB. Liver fibrosis: noninvasive diagnosis with double contrast material-enhanced MR imaging. Radiology 2006;239:425–37. [33] Adams LA, Sanderson S, Lindor KD, Angulo P. The histological course of nonalcoholic fatty liver disease: a longitudinal study of 103 patients with sequential liver biopsies. J Hepatol 2005;42:132–8. [34] Oliva MR, Mortele KJ, Segatto E, et al. Computed tomography features of nonalcoholic steatohepatitis with histopathologic correlation. J Comput Assist Tomogr 2006;30:37–43. [35] Liou I, Kowdley KV. Natural history of nonalcoholic steatohepatitis. J Clin Gastroenterol 2006;40:S11–6. [36] Hui JM, Kench JG, Chitturi S, et al. Long-term outcomes of cirrhosis in nonalcoholic steatohepatitis compared with hepatitis C. Hepatology 2003;38:420–7. [37] Shah TU, Semelka RC, Voultsinos V, et al. Accuracy of magnetic resonance imaging for preoperative detection of portal vein thrombosis in liver transplant candidates. Liver Transpl 2006;12:1682–8. [38] Zhang XM, Mitchell DG, Shi H, et al. Chronic hepatitis C activity: correlation with lymphadenopathy on MR imaging. AJR Am J Roentgenol 2002;179:417–22. [39] Bader TR, Beavers KL, Semelka RC. MR imaging features of primary sclerosing cholangitis: patterns of cirrhosis in relationship to clinical severity of disease. Radiology 2003;226:675–85. [40] Blachar A, Federle MP, Brancatelli G. Primary biliary cirrhosis: clinical, pathologic, and helical CT findings in 53 patients. Radiology 2001;220:329–36. [41] Hussain HK, Chenevert TL, Londy FJ, et al. Hepatic fat fraction: MR imaging for quantitative measurement and display - early experience. Radiology 2005;237:1048–55. [42] O'Reagan DP, Callaghan MF, Wylezinska-Arridge M, et al. Liver fat content and T2⁎: simultaneous measurement by using breath-hold multiecho MR imaging at 3.0 T — feasibility. Radiology 2008;247: 550–7. [43] Kim H, Taksali SE, Dufour S, et al. Comparative MR study of hepatic fat quantification using single-voxel proton spectroscopy, two-point dixon and three-point IDEAL. Magn Reson Med 2008;59: 521–7.

J. Elias Jr. et al. / Magnetic Resonance Imaging 27 (2009) 976–987 [44] Costa DN, Pedrosa I, McKenzie C, Reeder SB, Rofsky NM. Body MRI using IDEAL. AJR Am J Roentgenol 2008;190: 1076–84. [45] Taouli B, Tolia AJ, Losada M, et al. Diffusion-weighted MRI for quantification of liver fibrosis: preliminary experience. AJR 2007;189: 799–806.

987

[46] Hagiwara M, Rusinek H, Lee VS, et al. Advanced liver fibrosis: diagnosis with 3D whole-liver perfusion MR imaging — initial experience. Radiology 2008;246:926–34. [47] Huwart L, Sempoux C, Salameh N, et al. Liver fibrosis: noninvasive assessment with MR elastography versus aspartate aminotransferaseto-platelet ratio index. Radiology 2007;245:458–66.