AMERICAN ASSOCIATION FOR THE STUDY OFLIVERD I S E ASES

HEPATOLOGY, VOL. 67, NO. 1, 2018 Imaging for the Diagnosis of : A Systematic Review and Meta-analysis

Lewis R. Roberts ,1 Claude B. Sirlin,2 Feras Zaiem,3 Jehad Almasri,3 Larry J. Prokop,3 Julie K. Heimbach,1 M. Hassan Murad,3 and Khaled Mohammed3

Multiphasic computed tomography (CT) and magnetic resonance imaging (MRI) are both used for noninvasive diagnosis of hepatocellular carcinoma (HCC) in patients with cirrhosis. To determine if there is a relative diagnostic benefit of one over the other, we synthesized evidence regarding the relative performance of CT, extracellular contrast–enhanced MRI, and gadoxetate-enhanced MRI for diagnosis of HCC in patients with cirrhosis. We also assessed whether liver biopsy ver- sus follow-up with the same versus alternative imaging is best for CT-indeterminate or MRI-indeterminate liver nodules in patients with cirrhosis. We searched multiple databases from inception to April 27, 2016, for studies comparing CT with extracellular contrast–enhanced MRI or gadoxetate-enhanced MRI in adults with cirrhosis and suspected HCC. Two reviewers independently selected studies and extracted data. Of 33 included studies, 19 were comprehensive, while 14 reported sensitivity only. For all tumor sizes, the 19 comprehensive comparisons showed significantly higher sensitivity (0.82 versus 0.66) and lower negative likelihood ratio (0.20 versus 0.37) for MRI over CT. The specificities of MRI versus CT (0.91 versus 0.92) and the positive likelihood ratios (8.8 versus 8.1) were not different. All three modalities performed better for HCCs 2 cm. Performance was poor for HCCs <1 cm. No studies examined whether adults with cirrhosis and an indeterminate nodule are best evaluated using biopsy, repeated imaging, or alternative imaging. Concerns about publi- cation bias, inconsistent study results, increased risk of bias, and clinical factors precluded support for exclusive use of either gadoxetate-enhanced or extracellular contrast–enhanced MRI over CT. Conclusion: CT, extracellular contrast– enhanced MRI, or gadoxetate-enhanced MRI could not be definitively preferred for HCC diagnosis in patients with cir- rhosis; in patients with cirrhosis and an indeterminate mass, there were insufficient data comparing biopsy to repeat cross- sectional imaging or alternative imaging. (HEPATOLOGY 2018;67:401-421).

epatocellular carcinoma (HCC) is unique hepatic arterial, portal venous, and subsequent among malignancies in having tumor char- phases.(5) The differences in blood flow and extracellu- H acteristics on cross-sectional multiphasic lar volume between HCC tissues and non-neoplastic contrast computed tomography (CT) or magnetic res- cirrhotic liver tissue lead to hallmark imaging charac- onance imaging (MRI) that allow for a highly accurate teristics during the multiphasic flow of contrast, diagnosis of HCC without an invasive biopsy.(1-4) The including arterial phase hyperenhancement, subse- ability of cross-sectional imaging studies to reliably quent washout appearance, and capsule appearance.(6,7) detect and diagnose HCCs in the cirrhotic liver rests The pathophysiological underpinnings of arterial- primarily on characterizing the enhancement of a sus- phase hyperenhancement, washout appearance, and pected tumor relative to background liver in the capsule appearance are complex and reviewed

Abbreviations: CI, confidence interval; CT, computerized tomography; HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; NFS, nephrogenic systemic fibrosis.

Received January 10, 2017; accepted August 29, 2017. Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.29487/suppinfo. Supported by a contract from the American Association for the Study of Liver Diseases (to M.M.). Copyright VC 2017 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI 10.1002/hep.29487

Potential conflict of interest: Dr. Sirlin consults and is on the speakers’ bureau for Bayer. He received grants from GE and Siemen’s.

401 ROBERTS ET AL. HEPATOLOGY, January 2018 elsewhere.(8) Also of fundamental importance is pretest with some, but not all, of the hallmark features of probability: patients with cirrhosis have a high pretest HCC. The differential diagnosis for such nodules probability of HCC and a low pretest probability of includes HCC, non-HCC malignancy, and nonmalig- nonmalignant nodules that may resemble HCC at nant entities. Because imaging does not establish a imaging. In such patients, nodules with the hallmark specific diagnosis in such cases, prior clinical practice imaging features of HCC can be reliably diagnosed as guidelines by the American Association for the Study HCC.(9) of Liver Diseases recommended biopsy for all liver While the imaging features distinctive of HCC can lesions >1 cm initially detected by surveillance ultra- be observed both by multiphasic CT and MRI, MRI sound and interpreted as indeterminate by diagnostic offers a number of additional imaging sequences that call-back CT and MRI.(9,13) The evidence supporting can be helpful in HCC diagnosis, including T2- this recommendation was not provided, however. Due weighted sequences, diffusion-weighted imaging, and, to the limitations of biopsy(2,14) and the complexities in combination with the use of a partially extracellular of working up suspected HCC, alternative strategies and partially hepatocellular such as such as follow-up or alternative imaging may be prefer- gadoxetate disodium, the ability to distinguish even able in individual cases. relatively small and subtle lesions by hypointensity in We conducted this systematic review and meta- the hepatobiliary phase.(10-13) MRI has important analysis to synthesize the existing evidence about the diagnostic disadvantages, however, including greater comparative performance of multiphasic CT and MRI technical complexity, higher susceptibility to artifacts, with extracellular or gadoxetate contrast in the diagno- and less consistent image quality. In particular, MRI sis of HCC in patients with underlying cirrhosis. quality may be compromised in patients with difficulty While a number of systematic reviews and meta- breath-holding, trouble keeping still, or large-volume analyses have examined the performance of CT and/or ascites. Also, although noncontrast MRI may detect MRI in the diagnosis of HCC,(15-24) relatively few tumor nodules, such sequences rarely provide sufficient studies have examined these imaging modalities in specificity to enable noninvasive diagnosis of HCC. comparative studies in which CT was directly com- Thus, while noncontrast MRI may provide useful pared to either extracellular or partially extracellular information, it does not reliably permit definitive diag- and partially hepatocellular contrast agent MRI. A nosis and staging of HCC in most patients. For these number of the publications included studies that were reasons, the comparative diagnostic performance of not directly comparative,(15,17-19,21,22) and some did multiphasic CT and MRI in real-life practice remains not include more recent studies.(16,20) Two studies uncertain. published since our analyses were performed provide Another area of controversy is the optimal manage- some complementary information and are reviewed in ment of patients in whom CT or MRI detects a nodule the Discussion.(22,23) Also, we examined the available

ARTICLE INFORMATION:

From the 1Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine and Science, Rochester, MN; 2Liver Imaging Group, Department of Radiology, University of California San Diego, San Diego, CA; 3Evidence-Based Practice Center, Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, Mayo Clinic College of Medicine and Science, Rochester, MN. ADDRESS CORRESPONDENCE AND REPRINT REQUESTS TO:

Lewis R. Roberts, M.B. Ch.B., Ph.D. San Diego, CA Division of Gastroenterology and Hepatology E-mail: [email protected] Mayo Clinic College of Medicine and Science or Rochester, MN M. Hassan Murad, M.D., M.P.H. E-mail: [email protected] Evidence-Based Practice Center, Mayo Clinic Robert D. Tel: 11-507-284-4823 and Patricia E. Kern Center for the Science of Health Care or Delivery Claude B. Sirlin, M.D. Mayo Clinic College of Medicine and Science Liver Imaging Group, Department of Radiology Rochester, MN University of California San Diego E-mail: [email protected]

402 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL. evidence supporting biopsy or additional imaging for DATA EXTRACTION indeterminate lesions in patients with cirrhosis. We extracted the following variables from each study: study characteristics including primary author, Materials and Methods time period of study/year of publication and country of study; patient baseline characteristics including coun- We followed a predefined protocol developed by the try, age, lesion number, lesion size, alpha-fetoprotein HCC clinical practice guideline writing and systematic level, Child-Pugh score, and cause of cirrhosis; index review committees of the American Association for the and reference test characteristics; outcomes of interest. Study of Liver Diseases. We reported this systematic We extracted true-positive, false-positive, false- review according to the Preferred Reporting Items for negative, and true-negative values of the index tests Systematic Reviews and Meta-Analyses statement.(25) (MRI versus CT). Data extraction was done in duplicate. ELIGIBILITY CRITERIA Two questions were identified by the American METHODOLOGICAL QUALITY Association for the Study of Liver Diseases practice guidelines committee. Table 1 describes detailed inclu- AND RISK OF BIAS ASSESSMENT sion and exclusion criteria for the two questions of We used the Quality Assessment of Diagnostic interest. For question 1, we included studies that Accuracy Studies 2 tool(26) to assess the risk of bias enrolled adults diagnosed with cirrhosis and suspected and applicability of diagnostic accuracy studies. We HCC and compared multiphasic CT versus MRI with assessed the following domains: patient selection, and without extracellular contrast or gadoxetate diso- index test, reference standard, flow, and timing. Qual- dium. For question 2, we included studies that enrolled ity of evidence was evaluated using the Grading of adults with cirrhosis and indeterminate hepatic lesion Recommendations Assessment, Development, and and compared biopsy, repeated imaging, or alternative Evaluation approach.(27) imaging for the diagnostic evaluation.

SEARCH STRATEGY STATISTICAL ANALYSIS A comprehensive search of several databases from We calculated measures of diagnostic test accuracy each database inception to April 27, 2016, in any lan- (sensitivity, specificity, likelihood ratios, and diagnostic guage was conducted. The databases included Ovid odds ratios) using a bivariate regression model, allow- Medline In-Process & Other Non-Indexed Citations, ing for correlation between sensitivity and specificity. 2 > Ovid MEDLINE, Ovid EMBASE, Ovid Cochrane I 50% suggests high heterogeneity. Summary re- Central Register of Controlled Trials, and Scopus. ceiver operating characteristic curves were also esti- The search strategy was designed and conducted by an mated. Statistical analyses were conducted using Stata experienced librarian with input from the primary version 13 (StataCorp, College Station, TX). We cal- investigators. Supporting Tables S1 and S2 demon- culated an interaction P value to estimate the statistical strate the detailed search strategy for questions 1 and significance between the accuracy measures of the two 2, respectively. index tests (MRI versus CT). A separate analysis of studies that only reported detection rates (without a 3 STUDY SELECTION 2 2 diagnostic table) was performed. In this subset of studies we pooled the detection rate with 95% confi- Using an online reference management system (Dis- dence intervals (CIs) using Jeffreys method. We tillerSR; Evidence Partners, Inc.), two reviewers inde- pooled log-transformed event rates with DerSimonian pendently screened the titles and abstracts for potential and Laird random-effect models. We assessed publica- eligibility. Full text versions of the included abstracts tion bias by examining funnel plot asymmetry and were retrieved and screened in duplicate. Disagree- Egger’s regression test. Subgroup and sensitivity analy- ments were harmonized by consensus and, if not pos- ses were done based on the cohort enrollment date sible by consensus, through arbitration by a third (<2,000 and 2,000), the size of the hepatic lesion reviewer. (<1 cm, 1-2 cm, and >2 cm), and the type of MRI

403 ROBERTS ET AL. HEPATOLOGY, January 2018

TABLE 1. Inclusion and Exclusion Criteria

Q1 Q2

Population Adults with cirrhosis and suspected HCC Adults with cirrhosis and an indeterminate nodule after contrast-enhanced CT or MR or CEUS Intervention versus comparison Diagnosis and staging of HCC with contrast- Repeat imaging after observation period 2-3 enhanced, multiphasic CT versus MR with and months versus biopsy versus repeat alternative without extracellular contrast or gadoxetate imaging technique disodium Outcomes Accuracy of identifying and staging HCC Survival, cost-effectiveness, stage at diagnosis, patient tolerance Study design Comparative studies Comparative studies Exclusions Noncomparative studies that included either MRI or Noncomparative studies, reviews, case reports, CT only; reviews, case reports, and studies with and studies with fewer than 5 patients fewer than 5 patients

Abbreviation: CEUS, contrast-enhanced ultrasonography. contrast material used (gadoxetate-enhanced versus bias for the remaining 14 studies that reported detec- extracellular contrast–enhanced). tion rate only was considered low to moderate (Fig. 2B). Detailed assessment of the methodological quality Results of the studies is provided in Table 3. Pooled Analysis of Diagnostic Accuracy Our search strategy identified 2,256 citations. After screening titles and abstracts, a total of 433 were Nineteen studies(3,28-45) compared the diagnostic deemed eligible for full text retrieval. We eventually accuracy of MRI versus CT in detecting HCC. Com- included 33 studies. Figure 1 demonstrates the study pared to CT, MRI with an extracellular agent, or MRI selection process. with gadoxetate disodium showed a significantly higher sensitivity (0.82; 95% CI, 0.75-0.87; I2 5 2 Q1: WHAT IS THE DIAGNOSTIC 80.74; versus 0.66; 95% CI, 0.60-0.72; I 5 72.53) ACCURACY OF MULTIPHASIC CT and lower negative likelihood ratio (0.20; 95% CI, 0.15-0.28; versus 0.37; 95% CI, 0.30-0.44) in diagno- VERSUS MULTIPHASIC MRI IN sis of HCC lesions. No significant difference was ADULTS WITH CIRRHOSIS AND found for the pooled specificity, positive likelihood SUSPECTED HCC? ratio, or diagnostic odds ratio. Figures 3 and 4 illus- trate sensitivity and specificity forest plots of CT and Nineteen studies(3,28-45) reported true-positive, MRI, respectively. Analysis of the receiver operating false-positive, false-negative, and true-negative values, characteristic area under the curve (Fig. 5) showed that while 14 studies(5,46-58) reported only detection rate accuracy of MRI for detection of HCC was 0.91 (95% (sensitivity rate). Table 2 describes the detailed base- CI, 0.89-0.93) compared to 0.80 (95% CI, 0.77-0.83) line characteristics of the included studies. for CT. Sensitivity analysis was done to exclude two cohorts(39,44) that enrolled patients before 2000; no Methodological Quality change in the results was found. Table 4 shows the of the Included Studies diagnostic accuracy of CT versus MRI for diagnosis of The overall risk of bias of the 19 studies (Fig. 2A) HCC. that reported all diagnostic accuracy values was low to Subgroup Analysis moderate. Almost 50% of the studies were judged to have low risk of bias in terms of patient selection, index Subgroup analysis was done based on the type of test, reference standard, flow, and timing. The majority MRI contrast material. Eight studies(28,31,33,34,38,41,43,45) of the studies were considered to have a low risk of bias evaluated gadoxetate-enhanced MRI versus CT, and 11 on applicability to clinical practice in terms of reference studies(3,29,30,32,35-37,39,40,42,44) evaluated extracellular standard, index test, and patient selection. The risk of contrast–enhanced MRI versus CT in detecting HCC.

404 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.

Pooled analysis demonstrated that both gadoxetate- 94.6; versus 0.48; 95% CI, 0.34-0.62; I2 5 52). No enhanced MRI and extracellular contrast–enhanced differences were found in the pooled specificity, nega- MRI provided significantly higher sensitivity and lower tive likelihood ratio, positive likelihood ratio, and diag- negative likelihood ratio than CT. No significant differ- nostic odds ratio. Also, no differences were identified ence was noticed in the pooled specificity, positive likeli- in pooled performance characteristics between MRI hood ratio, and diagnostic odds ratio. Figures 6 and 7 with an extracellular agent and CT for 1-2 cm HCCs illustrate the forest plots of sensitivity and specificity of or between MRI with gadoxetate and CT for <2cm gadolinium ethoxybenzyl contrast MRI and extracellular HCCs. Table 5 presents subgroup analysis of the diag- contrast only MRI versus CT, respectively. nostic accuracy of MRI versus CT for the detection of Subgroup analysis was done based on the size of the HCC. The sensitivity of contrast-enhanced CT versus hepatic focal lesion. MRI with an extracellular agent gadoxetate-enhanced MRI was 0.73 versus 0.87, with showed a significantly higher sensitivity compared to a P value of 0.02, while the sensitivity of contrast- CT for both hepatic lesions >2 cm (0.88; 95% CI, enhanced CT versus extracellular contrast MRI was 0.80-0.93; I2 5 70.5; versus 0.79; 95% CI, 0.70-0.86; 0.61 versus 0.75, with a P value of 0.006. The negative I2 5 88.2) and <1 cm (0.69; 95% CI, 0.54-0.81; I2 5 likelihood ratio for contrast-enhanced CT versus



RECORDS IDENTIFIED THROUGH DATABASE RECORDS IDENTIFIED THROUGH SEARCHING FOR QUESTION 1 DATABASE SEARCHING FOR QUESTION 2 (N = 870) (N = 1386) Idenficaon

Records screened (n = 2256)

Records excluded (n = 1823 ) Screening Full-text arcles assessed for eligibility (n = 433) Full-text arcles excluded (n = 400 )

Different populaon = 47

Different intervenon = 116

Different outcomes = 143 Studies included (n =33) Abstracts = 76 Review arcles and case reports = 10

Duplicates = 8

Q 1 Q 2

(33 studies) (0 studies) Included Studies Eligibility

FIG. 1. Flow diagram of study selection.



405 406 TABLE 2. Characteristics of the Included Studies AL. ET ROBERTS

Study Design/ CT Imaging MRI Patients Male Age No. of Tumor Size AFP Level Child-Pugh Gold Cause of Reference Country Description fDescription (n) (n) (Years) Lesions (cm) (ng/mL) Score Standard Cirrhosis

Chen et al.(28) Retrospective/ Dynamic Gadoxetate 139 101 68 6 11 139 0.05-3 NR NR Pathology (115) NR Japan contrast- disodium– Benign imaging fea- enhanced CT enhanced MRI tures with >1 year follow- up (24) Golfieri et al.(30) Prospective/ Quadruple-phase Dynamic 3D MRI 63 53 63.3 123 1-3 NR A 46 Transplant (10) HBV 5 22, Italy MDCT with performed following B13 Resection (6) HCV 5 32 contrast administration of C4 Liver (8) biopsy Alcoholic 5 9 gadopentetate 2-year follow-up (9) dimeglumine Granito et al.(46) Prospective/ Quadruple-phase Gadoxetate 33 25 70 48 1.8 7 A5 22 Radiology HCV 5 19 Italy MDCT using a disodium– (48-84) (0.1-3) (1.8-375) A6 6 Biopsy HBV 5 6 helical CT enhanced MRI B7 4 Cryptogenic 5 4 scanner, B8 1 NASH 5 2 contrast- Alcohol 5 1P enhanced BC 5 1 Hassan, et al.(32) Retrospective/ MDCT triphasic MRI with contrast 61 37 46.5 95 0.52 NR NR Biopsy NR Kuwait imaging with () (19-74) (0.04-1) Radiological and nonionic iodinated clinical follow-up contrast agent Haradome et al.(31) Retrospective/ MDCT scanner with Gd-EOB-DTPA- 75 60 54.7 86 1.74 6 0.6 NR A 48 Pathologic HCV (23), multicenter 16 detector rows; all enhanced (42-67) B5 specimens: HBV (14), patients received MRI C1 surgical HCV1HBV (4), alco- intravenous nonionic resection holic (10), and cryp- contrast medium (19) or fine togenic (3) needle biopsy (41) Hayashida et al.(47) Retrospective/ A multidetector row Dynamic MRI 38 23 72 NR <3 NR NR Surgical Ethanol 5 3 Japan CT scanner with four with (41-83) resection (6) HBV 5 5 detector rows, triple- chelate Percutaneous HCV 5 26 phase contrast- needle biopsy (11) HBV1HCV 5 1 enhanced dynamic Combination of Autoimmune 5 2 CT with nonionic clinical and Cryptogenic 5 1 contrast medium radiological criteria (21) Hidaka et al.(33) Retrospective/ A 64-MDCT Contrast-enhanced 11 NR NR 17 <3 9.6 9 Histopathological HBV 5 3, EAOOY aur 2018 January HEPATOLOGY, Japan scanner with sequences with (3.2-506) (6-13) analysis HCV 5 7, contrast Gd-EOB-DTPA Others 5 1 Hori et al.(48) Japan Helical CT with MRI with 50 38 65 125 <1 5 41 NR NR Radiological HBV 5 5, contrast gadopentetate (42-81) 1-2 5 43 HCV 5 44, medium dimeglumine 2-3 5 24 non-B, non-C 5 1 (Gd-DTPA) > 5 324 Inoue et al.(49) Retrospective/ MDCT scanning MRI with 66 42 66 86 2.1 NR A 53 Pathological HCV-related chronic Japan using 64 Gd-EOB-DTPA (54-84) (0.6-4.0) B13 diagnosis hepatitis (16), channels with C0 HCV-related cirrhosis nonionic contrast (30), material HBV-related chronic hepatitis (9), HBV-related cirrhosis (11) EAOOY o.6,N.1 08RBRSE AL. ET ROBERTS 2018 1, No. 67, Vol. HEPATOLOGY,

TABLE 2. Continued

Study Design/ CT Imaging MRI Patients Male Age No. of Tumor Size AFP Level Child-Pugh Gold Cause of Reference Country Description fDescription (n) (n) (Years) Lesions (cm) (ng/mL) Score Standard Cirrhosis Jung et al.(50) Prospective/ Biphasic contrast- Nonenhanced and 40 29 66.4 41 5.5 (1.5-12) NR A 14 Pathological NR Germany enhanced spiral Gd-EOB-DTPA- (36-82) B7 diagnosis CT enhanced MRI C1, and unknown 5 9 Kasai et al.(34) Retrospective/ Dynamic enhanced Gd-EOB-DTPA MRI 47 35 65.4 6 9.1 112 NR NR NR All available NR Japan CT images with clinical information nonionic contrast (including US, CT, material MRI findings; laboratory data; histopathological findings; and radiologic follow-up examinations) Kawada et al.(51) Retrospective/ A 64-channel MDCT Dynamic MRI with 13 10 67 15 <2 95.3 A12 Pathological HCV 5 8, Japan with a nonionic Gd-EOB-DTPA (51-77) (2-437) B1 diagnosis by HBV 5 1, US-guided Alcoholic 5 1, agent biopsy non-B, non-C 5 3 Khalili et al.(35) Retrospective/ 64-detector CT scan MRI with gadobenate 84 53 58 (22-79) 101 1-2 NR NR Biopsy HBV 5 42 Canada with precontrast, dimeglumine Growth of HCV 5 28, arterial phase lesion on CT or MRI Others 5 14 (20s after trigger), on follow-up over portal venous 18 months phase (60s), and Recurrenceor delayed metastasis after (equilibrium) treatment phase (180s) Leoni et al.(36) Italy Helical multidetector Superparamagnetic 60 52 65.2 6 10 1 5 46 1.8 (1-3) 11 A40 Biopsy HCV 5 33 quadruple-phase iron oxide MR and 2 5 13 (2-2,849) B18 HBV 5 18 CT with contrast gadolinium MR 3 5 1 C2 HBV1HCV 5 1 Alchol 5 6 Cryptogenic 5 2 Libbrecht et al.(37) Retrospective/ Dynamic helical CT MRI with 49 32 53.4 6 11.6 CT 5 33 CT 5 27.5 6 NR NR Biopsy HBV 5 9 Belgium with contrast dimeglumine US 5 77 10.6 HCV 5 11 gadopentetate MRI 5 20 MRI 5 Cholestatic liver dis- or meglumine 26.0 6 11.5 ease 5 8 gadoterate Alcohol 5 12 Other or combination 5 9 407 408 AL. ET ROBERTS

TABLE 2. Continued

Study Design/ CT Imaging MRI Patients Male Age No. of Tumor Size AFP Level Child-Pugh Gold Cause of Reference Country Description fDescription (n) (n) (Years) Lesions (cm) (ng/mL) Score Standard Cirrhosis Maiwald et al.(38) Prospective/ Multiphase-64-slice 3-Tesla MRI with 50 42 60.6 NR NR NR A 27 Biopsy Alcoholic 5 26, Germany contrast- Gd-EOB-DTPA (29-84) B16 HBV 5 2, enhanced CT C7 HCV 5 3, Hemochromatosis 5 3, Budd-Chiari- syndrome 5 1, NASH 5 1, Cryptogenic 5 14 Di Martino et al.(29) Prospective/ Multiphasic CT MRI with gadobenate 140 104 66 254 >2 cm, NR A 71 Pathological NR Italy using a dimeglumine (23-82) 1-2 cm, <1cm B43 findings or 64-slice C26 substantial growth at 12-month follow-up Mita et al.(52) Retrospective/ Helical CT with MRI with 29 13 70.5 6 7.96 34 <2cm <20 5 21 NR Pathological HBV 5 1, Japan nonionic Gd-EOB-DTPA <21 5 8 diagnosis HCV 5 24, contrast medium Alcohol 5 4 Onishi et al.(53) Retrospective/ Multiphasic CT MRI with 31 28 70.2 73 <1.0 5 28 NR A 23 Partial surgical HBV 5 4, Japan performed using gadoxetate (52-82) 1-2 5 32 B8 resection (12) HCV 5 21, 8-channel CT disodium >2 5 13 C1 Biopsy (4) Alcoholic 5 4, (n 5 9) or Liver NASH 5 1 64-channel CT transplantation (15) (n 5 22) with and nonionic contrast medium Park et al.(54) Retrospective/ A contrast-enhanced MRI with 55 44 55 67 <3cm NR A53 Pathologically Patients without cir- South Korea multiphasic gadoxetate (28-73) B2 confirmed rhosis 5 17 helical CT, a disodium (HBV 5 13, 16-row MDCT HCV 5 2, scanner in 6 Unknown risk cases, a 64-row factor 5 2) MDCT scanner in Patients with 49 cases, and a cirrhosis 5 38 2018 January HEPATOLOGY, 128-row MDCT (HBV 5 36, scanner in 12 HCV 5 1, cases Alcoholic 5 1) Pitton et al.(55) Cohort/ Triphasic contrast- MRI with dynamic 28 25 67.0 6 10.8 NR MRI 5 1.5 NR 24 / 4 Biopsy Ethanol 5 13 Germany enhanced with a contrast- (0.5-14) HBV 5 2 64-row MDCT gadolinium-DTPA CT 5 1.2 HCV 5 7 (0.4-12.9) Cryptogenic 5 6 Puig et al.(39) Italy MDCT MRI with 50 NR NR 83 NR NR NR NR NR gadolinium EAOOY o.6,N.1 08RBRSE AL. ET ROBERTS 2018 1, No. 67, Vol. HEPATOLOGY,

TABLE 2. Continued

Study Design/ CT Imaging MRI Patients Male Age No. of Tumor Size AFP Level Child-Pugh Gold Cause of Reference Country Description fDescription (n) (n) (Years) Lesions (cm) (ng/mL) Score Standard Cirrhosis Rode et al.(40) Prospective/ Noncontrast MRI with 43 30 51 (27-65) 59 1.24 NR NR Pathological HBV 5 4 France spiral CT gadolinium HCV 5 19 Alcoholism 5 14 Biliary disease 5 2 Autoimmune hepatitis 5 1 Wilson disease 5 1 Cryptogenic 5 1 Sangiovanni et al.(3) Prospective/ A 64-row MDCT Dynamic MRI with 64 47 65 (44-80) 67 <1 5 2 11 A63 Fine-needle NR Italy with iodinated gadolinium 1-2 5 55 (1-2,156) B1 biopsy contrast medium >2 5 2 Sano et al.(41) Retrospective/ A 16–detector row Gadoxetate 64 47 67 6 9.3 108 0.4-2 NR A 54 Pathologic NR Japan dynamic contrast disodium– B10 diagnosis material–enhanced enhanced MRI CT Serste et al.(42) Case-only, Multidetector, Dynamic MRI 74 58 60 56 1-2 8 (1-413) NR Biopsy HCV 5 33 observational multiphasic CT (38-88) HBV 5 20 study/France before and after No cirrhosis 5 13 contrast medium Sugimoto et al.(56) Retrospective/ Helical with Dynamic 27 13 71.5 6 5.99 27 < 2 Hypovascular NR Biopsy HBV 5 1, Japan precontrast and contrast-enhanced HCC 5 7.7 6 5.4 HCV 5 19, postcontrast with MRI and Hypervascular Alcohol 5 6 nonionic contrast enhanced HCC 5 53.7 6 Non-HBV/HCV 5 1 medium (Gd-EOB-DTPA) MRI 118.9 Sun et al.(43) South Korea 8-, 16-, and 64- Gadoxetate 69 56 55.8 97 HCC 5 1.37 6 NR NR Histopathological HBV 5 56 MDCT scanners disodium– (39-73) 0.41 confirmation (16) HCV 5 6 with no contrast enhanced MRI Arterial Combination of CT, Alcohol enhancing angiographic abuse 5 3 pseudolesions 5 findings, lipiodol CT, Unknown 5 4 1.09 6 0.26 and AFP levels (28) Toyota et al.(57) Retrospective/ Contrast-enhanced Gd-EOB dual 50 32 68.8 57 2.24 (0.6-8.7) NR A 49 Histopathology HCV 5 26 Japan MDCT, a 64- gradient-recalled (50-89) B1 HBV 5 18 detector echo (GRE) MRI C0 NASH 5 2 or a 128- Alcoholic 5 1 detector CT Cryptogenic hepatitis 5 3 Tsurusaki et al.(58) Prospective/ A 64–detector Gadoxetate 54 39 63 83 NR NR A 53 Histopathological HBV 5 12 Japan row MDCT unit with disodium– (35-84) B1 examination, HCV 5 34 Non-B, contrast material enhanced MRI non-C 5 8 409 ROBERTS ET AL. HEPATOLOGY, January 2018

gadoxetate-enhanced MRI was 0.28 versus 0.13, with

186 a P value of 0.01, while the sensitivity of contrast- 66 5 5 251 9 enhanced CT versus extracellular contrast MRI was Cirrhosis Cause of 5 5 0.45 versus 0.29, with a P value of 0.002. HBV and/or HCV Liver cirrhosis PBC Alcoholic fibrosis Analysis of Studies That Reported Detection Rate Only Gold Standard Fourteen studies(5,46-58) compared MRI versus CT follow-up findings were used as the gold standard if the lesion was not confirmed histologically and reported only detection/sensitivity rate. Pooled analysis (Fig. 8) showed that detection/sensitivity rate Score

Child-Pugh and heterogeneity of CT, extracellular contrast– enhanced MRI, and gadoxetate-enhanced MRI were 0.70 (95% CI, 0.63-0.77; I2 5 80.5%; P < 0.001; 0.79; 95% CI, 0.67-0.93; I2 5 93.3%; P < 0.001; and (ng/mL) AFP Level 0.86; 95% CI, 0.81-0.92; I2 5 48.9%; P 5 0.057, respectively). Stratified analyses based on the size of hepatic lesions and the degree of HCC differentiation CV, hepatitis C virus; MDCT, multidetector CT; NASH, nonalcoholic steatohepa-

(cm) are shown in Supporting Figs. S9-S13. Tumor Size Publication Bias 61 NR NR NR Angiographic and NR 1.9 (0.5-3) NR NRVisual Biopsy inspection NR of the funnel plot suggests possible No. of Lesions publication bias (Supporting Fig. S14), which is con- sistent with the results of Egger’s regression test (P 5 Continued 0.04). Age (Years) (29-84) (48-83) Quality of Evidence TABLE 2.

28 67 The quality of evidence (i.e., certainty in the esti- (n) Male mates) is considered low to moderate, downgraded due to the methodological limitations of the included stud- 39 NR NR NR NR NR NR NR NR the (n) ies and possible publication bias. 50 (42 Patients analysis) included in Q2: ADULTS WITH CIRRHOSIS AND AN INDETERMINATE MRI

fDescription HEPATIC NODULE UNDERGO A disodium– enhanced MRI enhanced dynamic or gadopentetate dime- glumine MRI Gadoxetate Dynamic MRI 512 385 54.6 Gadolinium- BIOPSY, REPEATED IMAGING, OR ALTERNATIVE IMAGING FOR THE DIAGNOSTIC EVALUATION Description CT Imaging contrast contrast material No studies were found eligible to answer this Noncontrast spiral CT, then spiral CT with contrast Triphasic CT with population, intervention, comparison, and outcomes question. Country Japan China Study Design/ Retrospective/ Prospective/ Japan Helical CT with Discussion (5) (44)

(45) MAIN FINDINGS We identified 33 studies in this systematic review Ueda et al. Xing et al. Yamashita et al. Reference Abbreviations: AFP, alpha-fetoprotein; Gd-EOB-DTPA,titis; gadolinium-ethoxybenzyl-diethylenetriamine NR, penta-acetic not acid reported; HBV, PBC, hepatitis primary B biliary virus; cirrhosis; H US, ultrasound. which evaluated the performance of multiphasic

410 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.

TABLE 3. Risk of Bias Assessment Risk of Bias Applicability Are there Are there concerns that Could the Could the concerns that Are there the target conduct or reference the included concerns that condition as Could the interpretation standard, its Could the patients and the index defined by the selection of of the index conduct, or its patient flow setting do not test, its conduct, reference patients have test have interpretation have match the or interpretation standard does introduced introduced have introduced introduced review differ from the not match the Reference bias? bias? bias? bias? question? review question? question?

Chen et al.(28) High Low Low Low Low Low low Golfieri et al.(30) Low Unclear High High Low Low low Granito et al.(46) Low Low Unclear Unclear Low Low low Haradome High Low Low Low Low Low unclear et al.(31) Hassan et al.(32) Unclear Unclear High High High High unclear Hayashida Low Unclear High Unclear Low Low low et al.(47) Hidaka et al.(33) High Unclear Low Unclear Low Low low Hori et al.(48) Low Unclear High High Low Unclear unclear Inoue et al.(49) High Unclear Low Low Low Low low Jung et al.(50) High Low Low Low Low Low low Kasai et al.(34) Low Unclear High High Low Low low Kawada et al.(51) Unclear Unclear Low Low High Low low Khalili et al.(35) Low Low Low Low Low Low low Leoni et al.(36) Low Low Unclear Unclear Low Low low Libbrecht et al.(37) High Unclear Low Low Low Low low Maiwald et al.(38) Low Low High High Low Low low Di Martino Low Unclear Low Low Low Low low et al.(29) Mita et al.(52) High Unclear Low Low Low Low low Onishi et al.(53) High Low High High Low Low unclear Park et al.(54) High Low Low Low Low Low low Pitton et al.(55) Low Unclear High Unclear Low Low low Puig et al.(39) Low Low Unclear Unclear Low Low low Rode et al.(40) Low Unclear Low Low Low Low low Sangiovanni Low Low Low Low Low Low low et al.(3) Sano et al.(41) High Low Low Low Low Low low Serste et al.(42) High Low Low Low Low Low low Sugimoto et al.(56) High Unclear Low Low Low Low low Sun et al.(43) Low Low Low Low Low Low low Toyota et al.(57) High Low Low Low Low Low unclear Tsurusaki Low High Low Low Low Low low et al.(58) Ueda et al.(44) High Unclear High High Low Low unclear Xing et al.(45) Unclear Unclear Unclear Unclear Unclear Unclear unclear Yamashita et al.(5) Low Unclear Low High Low Low high cross-sectional contrast CT in comparison to con- a significantly lower negative likelihood ratio (0.20 trast MRI performed with extracellular contrast versus 0.37) for MRI over CT. However, the specif- agent or with the partially extracellular/partially icity was 0.91 for MRI versus 0.92 for CT, and the hepatocellular agent gadoxetate. Nineteen of the positive likelihood ratios of 8.8 for MRI versus 8.1 studies assessed all parameters for determining the for CT were not different (Table 4). Overall, the diagnostic accuracy of the tests, while 14 reported summary receiver operating characteristic curves only the detection rate or sensitivity of the tests. showedahigherareaunderthecurveof0.91for Pooled analysis of the 19 studies that compared the MRI versus 0.80 for CT. The results were similar diagnostic accuracy of MRI to CT showed a signifi- when considering only the 17 studies started in the cantly higher sensitivity (0.82 versus 0.66) as well as year 2000 or later (Table 4).

411 ROBERTS ET AL. HEPATOLOGY, January 2018



FIG. 2. The Quality Assessment of Diagnostic Accuracy Studies 2 results showing methodological quality of the studies included.



In subgroup analyses by type of MRI contrast mate- studies comparing CT to gadolinium ethoxybenzyl rial, similar observations were made with an absolute contrast, potentially suggestive of an unknown system- 14% increase in sensitivity for gadoxetate-enhanced atic bias in these studies. MRI and for extracellular contrast–enhanced MRI In subgroup analyses by lesion size, MRI showed higher over CT (Table 5). There was an unexplained better per-lesion sensitivity for <1 cm HCCs and for >2cm performance of cross-sectional imaging overall in the HCCs but not for 1-2 cm HCCs or for <2cmHCCs.

412 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.

TABLE 4. Accuracy of Contrast-Enhanced CT Versus MRI for Diagnosis of HCC Sensitivity Specificity 1 Likeli- – Likeli- Diagnostic Studies (95% CI) (95% CI) hood Ratio hood Ratio Odds Ratio Modality (n) I2 (%) P I2 (%) P (95% CI) P (95% CI) P (95% CI) P

All studies irrespective of cohort year Contrast- 19 0.66 0.0003 0.92 0.83 8.1 0.86 0.37 0.001 22 0.24 enhanced CT (0.60-0.72) (0.84-0.96) (4.1-16.2) (0.30-0.44) (10-50) I2 5 72.53 I2 5 86.74 MRI with and without 19 0.82 0.91 8.8 0.20 43 contrast (0.75-0.87) (0.82-0.95) (4.6-16.9) (0.15-0.28) (20-92) I2 5 72.90 I2 5 89.81 All cohorts started in the year 2000 or later Contrast- 17 0.69 0.002 0.94 0.82 11.9 0.96 0.32 0.01 37 0.3 enhanced CT (0.63-0.76) (0.87-0.98) (5.1-27.7) (0.26-0.40) (15-90) I2 5 73.9 I2 5 88.93 MRI 17 0.84 0.93 12.3 0.17 73 (0.77-0.90) (0.84-0.97) (5.1-29.5) (0.11-0.25) (29-181) I2 5 86.18

Moreover, although the sensitivity of MRI for <1cm Our results are similar to those of recent systematic HCCs was higher compared to CT (0.69 versus 0.49), reviews and meta-analyses, which found that multi- specificity at a trend level was lower (0.46 versus 0.69). phasic MRI is more sensitive than CT, while



FIG. 3. Sensitivity and specificity forest plots of contrast-enhanced CT studies for diagnosis of HCC.



413 ROBERTS ET AL. HEPATOLOGY, January 2018



FIG. 4. Sensitivity and specificity forest plots of MRI studies for diagnosis of HCC.





FIG. 5. Summary receiver operating characteristic curves showing performance of CT and MRI for diagnosis of HCC. Abbreviations: AUC, area under the curve; SROC, summary receiver operating characteristic.



414 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.



FIG. 6. Sensitivity and specificity forest plots from the eight studies comparing contrast-enhanced CT versus gadoxetate-enhanced MRI in diagnosis of HCC.





FIG. 7. Sensitivity and specificity forest plots from the 11 studies comparing contrast-enhanced CT versus extracellular contrast– enhanced MRI in diagnosis of HCC.



415 416 TABLE 5. Accuracy of Contrast-Enhanced CT Versus MRI in Diagnosis of HCC (Subgroup Analysis) AL. ET ROBERTS Specificity Sensitivity (95% CI) (95% CI) 1 Likelihood Ratio – Likelihood Ratio Diagnostic Odds Ratio Variable Modality Studies (n) I2 (%) P I2 (%) P (95% CI) P (95% CI) P (95% CI) P

Subgroup analysis based on type of MRI contrast material Gadoxetate contrast Contrast-enhanced CT 8 0.73 0.02 0.96 0.47 18.0 0.77 0.28 0.01 65 0.41 only MRI (0.64-0.81) (0.90-0.98) (7.2-45.4) (0.20-0.38) (23-179) I2 5 76.35 I2 5 80.31 Gadoxetate MRI 8 0.87 0.94 15.3 0.13 115 (0.79-0.93) (0.90-0.97) (8.3-28.3) (0.08-0.22) (47-278) I2 5 78.12 I2 5 60.07 Extracellular contrast Contrast-enhanced CT 11 0.61 0.006 0.87 0.91 4.5 0.73 0.45 0.002 10 0.31 only MRI (0.54-0.67) (0.73-0.94) (2.2-9.3) (0.38-0.54) (4-23) I2 5 57.77 I2 5 84.84 Extracellular contrast MRI 11 0.75 0.86 5.5 0.29 19 (0.67-0.82) (0.68-0.95) (2.3-13.1) (0.23-0.36) (8-45) I2 5 73.67 I(2) 5 90.04 Subgroup analysis based on size of hepatic lesion 1-2 cm Contrast CT 6 0.64 0.15 0.88 0.78 6.2 0.89 0.45 0.47 13 0.78 (0.58-0.70) (0.82-0.92) (1.83-20.86) (0.39-0.54) (4.8-39.6) I2 5 61.79 I2 5 90.89 Extracellular MRI 6 0.70 0.87 5.5 0.39 17 (0.64-0.75) (0.81-0.91) (1.6-18.3) (0.27-0.55) (5.1-62.4) I2 5 80.4 I2 5 92.1 >2 cm Contrast CT 3 0.79 0.09 0.9 0.71 6.46 0.99 0.26 0.5 25.79 0.47 (0.70-0.86) (0.76 - 0.97) (2.72-15.32) (0.07-0.95) (2.8-237.96) I2 5 88.2 I2 5 0 I2 5 0 I2 5 88.8 I2 5 65.1 Extracellular MRI 3 0.88 0.87 6.48 0.15 64.66 (0.80-0.93) (0.73-0.96) (2.98-14.07) (0.05-0.5) (19.13-218.55) I2 5 70.5 I2 5 0 I2 5 0 I2 5 78.3 I2 5 0 <2 cm Contrast CT 2 0.68 0.31 0.98 0.9 21.50 0.9 0.35 0.3 57.46 0.73 (0.55-0.79) (0.90-1) (4.32-106.9) (0.23-0.55) (9.89-333.97)

I2 5 23.2 I2 5 13.3 I2 5 0 I2 5 29.7 I2 5 0 2018 January HEPATOLOGY, Gadoxetate MRI 2 0.76 0.96 20.39 0.25 86.5 (0.67-0.84) (0.87-0.99) (5.2-80.01) (0.16-0.40) (17.91-417.99) I2 5 0 I2 5 0 I2 5 0 I2 5 0 I2 5 0 <1 cm Contrast CT 2 0.48 0.049 0.69 0.08 1.54 0.55 0.77 0.72 2.05 0.8 (0.34-0.62) (0.51-0.83) (0.88-2.70) (0.55-1.08 (0.84-5.04) I2 5 52 I2 5 0 I2 5 0 I2 5 0 I2 5 0 Extracellular MRI 2 0.69 0.46 1.27 0.63 2.3 (0.54-0.81) (0.29-0.63) (0.96-1.68) (0.22-1.77) (0.80-6.55) I2 5 94.6 I2 5 84.3 I2 5 0 I2 5 61.3 I2 5 0 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.

maintaining similar specificity.(18-23) The one meta- P 0.9 analysis published since our analysis was performed included studies in which there was no direct compari- son between CT and MRI and is therefore complemen- tary to ours.(23) The major distinguishing feature of this 14 (2-104) (1-137) study was the inclusion of only those studies that (95% CI) directly compared CT with MRI, thus reducing poten-

Diagnostic Odds Ratio tially substantial biases in studies without direct perfor- mance comparisons. In particular, the reported accuracy P

0.37 14 of imaging for HCC diagnosis depends on multiple fac- tors that may vary across studies, including population characteristics (e.g., severity of cirrhosis, degree of portal hypertension, and frequency of ) and applied ref- 0.31

(95% CI) erence standard (e.g., follow-up imaging, biopsy, hepa- (0.14-0.69) (0.30-0.72) tectomy pathology). Restricting our analysis to articles – Likelihood Ratio reporting within-study performance helps mitigate the P 0.7 0.47 confounding effects of these variables. Other distin- guishing features of our study compared to some meta- analyses(15-17,20,21,23) were the inclusion of CT, extracel-

4.4 lular agent MRI, and gadoxetate-MRI and the subanal- (1-20.5) (1-43.6)

(95% CI) yses by contrast material and lesion size. Likelihood Ratio 1 STRENGTHS AND LIMITATIONS P 0.7 6.6 While multiphasic MRI appears to be marginally more Continued 92

81.5 sensitive than CT in the pooled analysis of comparative (%) 5 0.83 5 2

I 2 I studies, examination of the data suggested possible publi- 2 I (95% CI) Specificity (0.43-0.97) (0.60-0.99) cation bias, and the quality of the evidence was considered

TABLE 5. to be low to moderate, with particular concerns including P

0.2 0.91 the methodological limitations of the studies and incon- sistencies across studies. Consequently, the differences in pooled diagnostic performance are considered insufficient 61.6 86.28

(%) to definitively recommend MRI over CT. Key factors 5 2 I 5 2 I 2 driving this conclusion include the overall low quality of I (0.50-0.89) (0.44-0.70) the evidence, practical concerns about generalizability to Sensitivity (95% CI) nonacademic settings, and recognition that multiple fac- tors beyond diagnostic accuracy inform the optimal selec- tion of imaging modalities in individual patients. Compared to multiphasic CT, multiphasic MRI has the advantages of providing greater soft tissue contrast, more detailed evaluation of nodule and background liver tissue characteristics, and absence of exposure to ionizing radiation. However, MRI also has important disadvan- tages, including greater cost, higher technical complex- ity, longer scan times, increased tendency to artifact, and MRI 4 0.74 Contrast CT 4 0.58 less consistent image quality due to patient factors such

< as difficulty with breath-holding, difficulty holding still, 2cm

< or high-volume ascites. MRI also has a larger variety of

1 contraindications, and—particularly outside the United States—substantially less availability, leading to longer 1cm Combination of Variable Modality Studies (n) wait times for imaging.

417 ROBERTS ET AL. HEPATOLOGY, January 2018



FIG. 8. Sensitivity/detection rate of contrast-enhanced CT, gadoxetate-enhanced MRI, and extracellular contrast–enhanced MRI in detection of HCC from the 14 studies reporting only sensitivity/detection rates.



In contrast, CT is more readily available and faster consequences including the incompletely understood and, due to more wide open scanner bores, less likely to long-term effects of exposure to ionizing radiation provoke claustrophobia but has the shortcoming of (CT) and gadolinium deposition (MRI). Because exposing patients to radiation. Both CT and MRI those risks are difficult to quantify and predict, individ- require intravenous access and contrast agents, the use ual patient preferences should be considered. Addi- of which may be problematic in patients with acute kid- tionally, MRI with gadolinium-based agents may be ney injury or chronic renal failure. preferable in patients with mild to moderate chronic The choice between CT and MRI also depends on kidney disease due to the reduced risk of contrast- patient safety preferences as both modalities are associ- induced nephrotoxicity, while patients with end-stage ated with potential downstream adverse health kidney disease on dialysis should probably undergo CT

418 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL. with iodinated agents to prevent the rare development may be compromised. In such patients, extracellu- of nephrogenic systemic fibrosis (NSF). Contrast- lar contrast agent MRI or CT may be preferable. enhanced imaging should be rescheduled for patients 4. Contrast agent contraindications: CT and MRI con- with acute kidney injury if possible. In regard to the trast agents may be contraindicated in some risk of NSF from use of MRI contrast agents in patients due to safety concerns, while other patients patients with chronic renal failure, while there have may refuse contrast agents or have inadequate intra- been some publications suggesting that newer MRI venous access. In such patients, noncontrast MRI contrast agents may be associated with increased risk may be best. Contrast-enhanced ultrasound is of NSF, there is insufficient evidence from well- another option in patients in whom CT or MRI is controlled studies to address this question. A prospec- contraindicated for safety considerations; but this tive, multicenter, nonrandomized phase 4 study of modality has only recently been approved in the gadoxetate disodium including patients with moderate United States, and the requisite expertise for its (n 5 193) and severe (n 5 85) renal impairment did application is not yet widespread. not raise any clinically significant safety concern, and Finally the choice may depend on the preferred sensi- no NSF cases were observed. tivity/specificity trade-off: 1. Gadoxetate-MRI is more sensitive for <2cm HCCs than CT or extracellular contrast agent MRI. CLINICAL AND RESEARCH Thus, gadoxetate-MRI may be the preferred modal- IMPLICATIONS ity in clinical practice paradigms (such as in Asia) that emphasize aggressive locoregional treatment or This systematic review confirms the utility of cross- excision of small HCC lesions. sectional multiphasic imaging for noninvasive diagno- 2. Extracellular contrast agent MRI and CT are > sis of HCCs 2 cm in size but also shows that perfor- marginally more specific for <2 cm HCCs than mance of both imaging modalities begins to degrade gadoxetate-MRI. Thus, the former two modalities substantially below a lesion size of 2 cm and that both may be preferred in clinical practice paradigms modalities have only modest accuracy below a lesion (such as in the United States) that emphasize liver size of 1 cm. Based on the limitations of the available transplantation for treating early-stage HCC with- evidence, no definitive recommendation can be made out requiring biopsy confirmation. for systematic use of gadoxetate-enhanced MRI or extracellular contrast–enhanced MRI over CT. How- The results of this systematic review highlight the ever, numerous other factors may guide the choice need for rigorously designed and executed compara- between modalities, but these were not assessed by our tive studies that can reliably answer the question of the relative performance of multiphasic CT versus meta-analysis due to lack of reporting on the relevant extracellular contrast–enhanced or gadoxetate- variables. A complete discussion is beyond the scope of enhanced MRI and comparison of gadoxetate- this article. A few common factors are described below: enhanced MRI versus extracellular contrast–enhanced 1. Patients with ascites: Ascites can introduce severe MRI. An additional question related to imaging for artifacts on MRI, especially at 3 tesla. Such HCC is whether adults with cirrhosis and an indeter- patients may be better off with CT. If MRI is minate hepatic nodule are best served by diagnostic pursued, 1.5T scanning should be considered. evaluation using biopsy, repeated imaging, or alterna- 2. Poor breath-holders: Poor breath-holders may be tive imaging. As there was a complete lack of applica- better off with CT for the same reason. Robust ble studies identified by the search strategy, this is a free-breathing sequences are under development critical area for future research. and someday may eliminate this problem, but such sequences are not yet widely available REFERENCES 3. Liver decompensation and/or severe iron overload: Gadoxetate disodium uptake by the liver tends to 1) Mitsuzaki K, Yamashita Y, Ogata I, Nishiharu T, Urata J, Takahashi M. Multiple-phase helical CT of the liver for detect- be reduced in patients with decompensated liver ing small hepatomas in patients with liver cirrhosis: contrast- disease. In patients with severe iron overload, the injection protocol and optimal timing. AJR Am J Roentgenol uptake may be preserved, but the ability of the 1996;167:753-757. agent to enhance the liver and characterize lesions

419 ROBERTS ET AL. HEPATOLOGY, January 2018

2) Forner A, Vilana R, Ayuso C, Bianchi L, Sole M, Ayuso JR, 17) Chen L, Zhang L, Liang M, Bao J, Zhang J, Xia Y, et al. Mag- et al. Diagnosis of hepatic nodules 20 mm or smaller in cirrhosis: netic resonance imaging with disodium for the prospective validation of the noninvasive diagnostic criteria for detection of hepatocellular carcinoma: a meta-analysis of 18 stud- hepatocellular carcinoma. HEPATOLOGY 2008;47:97-104. ies. Acad Radiol 2014;21:1603-1613. 3) Sangiovanni A, Manini MA, Iavarone M, Romeo R, Forzenigo 18) Lee YJ, Lee JM, Lee JS, Lee HY, Park BH, Kim YH, et al. LV, Fraquelli M, et al. The diagnostic and economic impact of Hepatocellular carcinoma: diagnostic performance of multidetec- contrast imaging techniques in the diagnosis of small hepatocel- tor CT and MR imaging—a systematic review and meta-analysis. lular carcinoma in cirrhosis. Gut 2010;59:638-644. Radiology 2015;275:97-109. 4) Manini MA, Sangiovanni A, Fornari F, Piscaglia F, Biolato M, 19) Chou R, Cuevas C, Fu R, Devine B, Wasson N, Ginsburg A, Fanigliulo L, et al. Clinical and economical impact of 2010 et al. Imaging techniques for the diagnosis of hepatocellular car- AASLD guidelines for the diagnosis of hepatocellular carcinoma. cinoma: a systematic review and meta-analysis. Ann Intern Med J Hepatol 2014;60:995-1001. 2015;162:697-711. 5) Yamashita Y, Mitsuzaki K, Yi T, Ogata I, Nishiharu T, Urata J, et al. 20) Ye F, Liu J, Ouyang H. Gadolinium ethoxybenzyl diethylenetri- Small hepatocellular carcinoma in patients with chronic liver damage: amine pentaacetic acid (Gd-EOB-DTPA)–enhanced magnetic prospective comparison of detection with dynamic MR imaging and resonance imaging and multidetector-row computed tomography helical CT of the whole liver. Radiology 1996;200:79-84. for the diagnosis of hepatocellular carcinoma: a systematic review 6) Stevens WR, Johnson CD, Stephens DH, Batts KP. CT find- and meta-analysis. Medicine (Baltimore) 2015;94:e1157. ings in hepatocellular carcinoma: correlation of tumor characteris- 21) Li X, Li C, Wang R, Ren J, Yang J, Zhang Y. Combined appli- tics with causative factors, tumor size, and histologic tumor cation of gadoxetic acid disodium-enhanced magnetic resonance grade. Radiology 1994;191:531-537. imaging (MRI) and diffusion-weighted imaging (DWI) in the 7) Choi JY, Lee JM, Sirlin CB. CT and MR imaging diagnosis diagnosis of chronic liver disease–induced hepatocellular carci- and staging of hepatocellular carcinoma: part I. Development, noma: a meta-analysis. PLoS One 2015;10:e0144247. growth, and spread: key pathologic and imaging aspects. Radiol- 22) Hanna RF, Miloushev VZ, Tang A, Finklestone LA, Brejt SZ, ogy 2014;272:635-654. Sandhu RS, et al. Comparative 13-year meta-analysis of the sen- 8) Choi JY, Lee JM, Sirlin CB. CT and MR imaging diagnosis sitivity and positive predictive value of ultrasound, CT, and MRI and staging of hepatocellular carcinoma: part II. Extracellular for detecting hepatocellular carcinoma. Abdom Radiol (NY) agents, hepatobiliary agents, and ancillary imaging features. Radi- 2016;41:71-90. ology 2014;273:30-50. 23) Guo J, Seo Y, Ren S, Hong S, Lee D, Kim S, et al. Diagnostic per- 9) Bruix J, Sherman M; American Association for the Study of formance of contrast-enhanced multidetector computed tomography Liver Diseases. Management of hepatocellular carcinoma: an and gadoxetic acid disodium–enhanced magnetic resonance imaging update. HEPATOLOGY 2011;53:1020-1022. in detecting hepatocellular carcinoma: direct comparison and a meta- 10) Piana G, Trinquart L, Meskine N, Barrau V, Beers BV, Vilgrain analysis. Abdom Radiol (NY) 2016;41:1960-1972. V. New MR imaging criteria with a diffusion-weighted sequence 24) Floriani I, D’Onofrio M, Rulli E, Chen MH, Li R, Musicco L. for the diagnosis of hepatocellular carcinoma in chronic liver dis- Performance of imaging modalities in the diagnosis of hepatocel- eases. J Hepatol 2011;55:126-132. lular carcinoma: a systematic review and meta-analysis. Ultra- 11) Raman SS, Leary C, Bluemke DA, Amendola M, Sahani D, schall Med 2013;34:454-462. McTavish JD, et al. Improved characterization of focal liver 25) Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. lesions with liver-specific gadoxetic acid disodium-enhanced Preferred reporting items for systematic reviews and meta-analy- magnetic resonance imaging: a multicenter phase 3 clinical trial. ses: the PRISMA statement. PLoS Med 2009;6:e1000097. J Comput Assist Tomogr 2010;34:163-172. 26) Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, 12) Tsuboyama T, Onishi H, Kim T, Akita H, Hori M, Tatsumi Reitsma JB, et al. QUADAS-2: a revised tool for the quality M, et al. Hepatocellular carcinoma: hepatocyte-selective enhance- assessment of diagnostic accuracy studies. Ann Intern Med 2011; ment at gadoxetic acid-enhanced MR imaging—correlation with 155:529-536. expression of sinusoidal and canalicular transporters and bile 27) Murad MH, Montori VM, Ioannidis JP, Jaeschke R, Devereaux accumulation. Radiology 2010;255:824-833. PJ, Prasad K, et al. How to read a systematic review and meta- 13) Ichikawa T, Saito K, Yoshioka N, Tanimoto A, Gokan T, analysis and apply the results to patient care: users’ guides to the Takehara Y, et al. Detection and characterization of focal liver medical literature. JAMA 2014;312:171-179. lesions: a Japanese phase III, multicenter comparison between 28) Chen N, Motosugi U, Morisaka H, Ichikawa S, Sano K, gadoxetic acid disodium–enhanced magnetic resonance imaging Ichikawa T, et al. Added value of a gadoxetic acid-enhanced and contrast-enhanced computed tomography predominantly in hepatocyte-phase image to the LI-RADS system for diagnosing patients with hepatocellular carcinoma and chronic liver disease. hepatocellular carcinoma. Magn Reson Med Sci 2016;15:49-59. Invest Radiol 2010;45:133-141. 29) Di Martino M, De Filippis G, De Santis A, Geiger D, Del 14) Silva MA, Hegab B, Hyde C, Guo B, Buckels JA, Mirza DF. Monte M, Lombardo CV, et al. Hepatocellular carcinoma in cir- Needle track seeding following biopsy of liver lesions in the diag- rhotic patients: prospective comparison of US, CT and MR nosis of hepatocellular : a systematic review and meta-anal- imaging. Eur Radiol 2013;23:887-896. ysis. Gut 2008;57:1592-1596. 30) Golfieri R, Marini E, Bazzocchi A, Fusco F, Trevisani F, Sama 15) Colli A, Fraquelli M, Casazza G, Massironi S, Colucci A, Conte D, C, et al. Small (

420 HEPATOLOGY, Vol. 67, No. 1, 2018 ROBERTS ET AL.

32) Hassan A, Al-Ajami R, Dashti K, Abdoelmoneum M. Sixty- 47) Hayashida M, Ito K, Fujita T, Shimizu A, Sasaki K, Tanabe M, four multi-slice computed tomography and magnetic resonance et al. Small hepatocellular carcinomas in cirrhosis: differences in imaging in evaluation of hepatic focal lesions. Egyptian Journal contrast enhancement effects between helical CT and MR imag- of Radiology and Nuclear Medicine 2011;42:101-110. ing during multiphasic dynamic imaging. Magn Reson Imaging 33) Hidaka M, Takatsuki M, Okudaira S, Soyama A, Muraoka I, 2008;26:65-71. Tanaka T, et al. The expression of transporter OATP2/OATP8 48) Hori M, Murakami T, Oi H, Kim T, Takahashi S, Matsushita M, decreases in undetectable hepatocellular carcinoma by Gd-EOB- et al. Sensitivity in detection of hypervascular hepatocellular carcinoma MRI in the explanted cirrhotic liver. Hepatol Int 2013;7:655-661. by helical CT with intra-arterial injection of contrast medium, and by 34) Kasai R, Hasebe T, Takada N, Inaoka T, Hiruta N, Terada H. helical CT and MR imaging with intravenous injection of contrast Comparison of diagnostic efficacy of Gd-EOB-DTPA enhanced medium. Acta Radiol 1998;39:144-151. MRI and dynamic contrast-enhanced multislice CT in hepato- 49) Inoue T, Kudo M, Komuta M, Hayaishi S, Ueda T, Takita M, cellular carcinoma. Journal of the Medical Society of Toho Uni- et al. Assessment of Gd-EOB-DTPA–enhanced MRI for HCC versity 2012;59:279-289. and dysplastic nodules and comparison of detection sensitivity 35) Khalili K, Kim TK, Jang H-J, Haider MA, Khan L, Guindi M, versus MDCT. J Gastroenterol 2012;47:1036-1047. et al. Optimization of imaging diagnosis of 1-2 cm hepatocellular 50) Jung G, Breuer J, Poll LW, Koch JA, Balzer T, Chang S, et al. carcinoma: an analysis of diagnostic performance and resource Imaging characteristics of hepatocellular carcinoma using the utilization. J Hepatol 2011;54:723-728. hepatobiliary contrast agent Gd-EOB-DTPA. Acta Radiol 2006; 36) Leoni S, Piscaglia F, Golfieri R, Camaggi V, Vidili G, Pini P, et al. 47:15-23. The impact of vascular and nonvascular findings on the noninvasive 51) Kawada N, Ohkawa K, Tanaka S, Matsunaga T, Uehara H, diagnosis of small hepatocellular carcinoma based on the EASL and Ioka T, et al. Improved diagnosis of well-differentiated hepato- AASLD criteria. Am J Gastroenterol 2010;105:599-609. cellular carcinoma with gadolinium ethoxybenzyl diethylene tria- 37) Libbrecht L, Bielen D, Verslype C, Vanbeckevoort D, Pirenne J, mine pentaacetic acid–enhanced magnetic resonance imaging and Nevens F, et al. Focal lesions in cirrhotic explant livers: patho- Sonazoid contrast-enhanced ultrasonography. Hepatol Res 2010; logical evaluation and accuracy of pretransplantation imaging 40:930-936. examinations. Liver Transpl 2002;8:749-761. 52) Mita K, Kim SR, Kudo M, Imoto S, Nakajima T, Ando K, 38) Maiwald B, Lobsien D, Kahn T, Stumpp P. Is 3-tesla Gd- et al. Diagnostic sensitivity of imaging modalities for hepatocellu- EOB-DTPA-enhanced MRI with diffusion-weighted imaging lar carcinoma smaller than 2 cm. World J Gastroenterol 2010;16: superior to 64-slice contrast-enhanced CT for the diagnosis of 4187-4192. hepatocellular carcinoma? PLoS One 2014;9:e111935. 53) Onishi H, Kim T, Imai Y, Hori M, Nagano H, Nakaya Y, et al. 39) Puig J, Martin J, Donoso L, Falco J, Rue M. Comparison Hypervascular hepatocellular carcinomas: detection with gadoxe- between biphasic helical CT and dynamic gadolinium–enhanced tate disodium–enhanced MR imaging and multiphasic multide- MR in the detection and characterization of focal hepatic lesions tector CT. Eur Radiol 2012;22:845-854. in cirrhotic patients. Radiologia 1997;39:617-623. 54) Park VY, Choi JY, Chung YE, Kim H, Park MS, Lim JS, et al. 40) Rode A, Bancel B, Douek P, Chevallier M, Vilgrain V, Picaud G, Dynamic enhancement pattern of HCC smaller than 3 cm in et al. Small nodule detection in cirrhotic livers: evaluation with US, diameter on gadoxetic acid–enhanced MRI: comparison with spiral CT, and MRI and correlation with pathologic examination of multiphasic MDCT. Liver Int 2014;34:1593-1602. explanted liver. J Comput Assist Tomogr 2001;25:327-336. 55) Pitton MB, Kloeckner R, Herber S, Otto G, Kreitner KF, 41) Sano K, Ichikawa T, Motosugi U, Sou H, Muhi AM, Matsuda M, Dueber C. MRI versus 64-row MDCT for diagnosis of hepato- et al. Imaging study of early hepatocellular carcinoma: usefulness of cellular carcinoma. World J Gastroenterol 2009;15:6044-6051. gadoxetic acid-enhanced MR imaging. Radiology 2011;261:834-844. 56) Sugimoto K, Kim SR, Imoto S, Tohyama M, Kim SK, 42) Serste T, Barrau V, Ozenne V, Vullierme MP, Bedossa P, Matsuoka T, et al. Characteristics of hypovascular versus hyper- Farges O, et al. Accuracy and disagreement of computed tomog- vascular well-differentiated hepatocellular carcinoma smaller than raphy and magnetic resonance imaging for the diagnosis of small 2 cm—focus on tumor size, markers and imaging detectability. hepatocellular carcinoma and dysplastic nodules: role of biopsy. Dig Dis 2015;33:721-727. HEPATOLOGY 2012;55:800-806. 57) Toyota N, Nakamura Y, Hieda M, Akiyama N, Terada H, Matsuura 43) Sun HY, Lee JM, Shin CI, Lee DH, Moon SK, Kim KW, N, et al. Diagnostic capability of gadoxetate disodium–enhanced liver et al. Gadoxetic acid–enhanced magnetic resonance imaging for MRI for diagnosis of hepatocellular carcinoma: comparison with differentiating small hepatocellular carcinomas (<2 cm in diame- multi-detector CT. Hiroshima J Med Sci 2013;62:55-61. ter) from arterial enhancing pseudolesions: special emphasis on 58) Tsurusaki M, Sofue K, Isoda H, Okada M, Kitajima K, hepatobiliary phase imaging. Invest Radiol 2010;45:96-103. Murakami T. Comparison of gadoxetic acid–enhanced magnetic 44) Ueda K, Kitagawa K, Kadoya M, Matsui O, Takashima T, resonance imaging and contrast-enhanced computed tomography Yamahana T. Detection of hypervascular hepatocellular carci- with histopathological examinations for the identification of noma by using spiral volumetric CT: comparison of US and MR hepatocellular carcinoma: a multicenter phase III study. imaging. Abdom Imaging 1995;20:547-553. J Gastroenterol 2016;51:71-79. 45) Xing GS, Wang S, Ouyang H, Ma XH, Zhou CW. Compari- son of CT and dynamic-enhancement MRI for the diagnosis of hepatocellular carcinoma. [in Chinese] Chinese Journal of Medi- cal Imaging Technology 2010;26:1-4. 46) Granito A, Galassi M, Piscaglia F, Romanini L, Lucidi V, Renzulli M, et al. Impact of gadoxetic acid (Gd-EOB-DTPA)– enhanced magnetic resonance on the non-invasive diagnosis of Supporting Information small hepatocellular carcinoma: a prospective study. Aliment Pharmacol Ther 2013;37:355-363. Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.29487/suppinfo.

421