To prospectively compare the diagnostic yield of non-contrast MRI including diffusion-weighted imaging (DWI) and US as a surveillance tool for detecting HCCs in patients with cirrhosis at high risk of HCC.
METHOD AND MATERIALSA prospective surveillance study included 407 consecutive cirrhosis patients with an estimated annual risk of HCC > 5% who underwent one to three, biannual screening examinations with paired liver MRI and US between November 2011 and August 2014. The referral criteria for HCC on non-contrast MRI were defined as a nodule showing one of the following criteria: the presence of intralesion fat, mild to moderate hyperintensity on T2-weighted imaging, or hyperintensity on DWI at b-500 s/mm2. On US, a discrete focal lesion equal or more than 1cm in diameter or suspicious tumor thrombosis was referred for a further work-up for HCC. During image interpretation, radiologists were blinded to the results of the other imaging modality. The confirmation of HCC was based on the results of a histologic examination and/or typical CT images of HCC. Per-lesion sensitivity of HCC and per-exam specificity were compared between non-contrast MRI and US.
RESULTSIn 1100 screening rounds of paired MRI and US, 48 HCCs were diagnosed in 43 patients. Among the 48 HCCs, intralesional fat was detected in 4 HCCs (8.3%), 31 HCCs (64.6%) showed mild to moderate hyperintensity on T2-weighted imaging, and 37 HCC (77.1%) appeared hyperintensity on DWI. Among the 71 positive cases on US, 12 patients were confirmed to have HCC. Using the diagnostic criteria, per lesion sensitivity of non-contrast MRI was 81.3 % (39/48), which was significantly higher that of 25% (12/48) (P<0.001). Non-contrast MRI showed a significantly higher per-exam specificity that US [98.0% (1036/1057) vs. 94.4% (998/1057), P<0.001].
CONCLUSIONIn this prospective intraindividual comparison study, non-contrast MRI including DWI outperformed US as a surveillance test for HCC in patients with cirrhosis at high risk of HCC.
CLINICAL RELEVANCE/APPLICATIONIn patients with cirrhosis at high risk of HCC, the non-contrast MRI including DWI can be considered to an alternative surveillance tool to US.
To evaluate the diagnostic accuracy of prospectively applied Liver Imaging Reporting and Data System (LI-RADS, v2014) for hepatocellular carcinoma (HCC) on gadoxetic acid-enhanced liver MRI.
METHOD AND MATERIALSThis study included 528 hepatic observations from 268 patients who met the following criteria: 1) at high risk for HCC, 2) underwent gadoxetic acid-enhanced liver MRI in 2014 at our institution, 3) not previously treated for HCC, and 4) whose radiologic reports were prospectively reported using LI-RADS. Final diagnoses were determined histologically for all non-HCC malignancies (two cholangiocarcinomas [CCs] and four combined HCC-CCs), 110 of 122 HCCs, seven regenerative or dysplastic nodules, and one hemangioma. Fifty benign lesions and 12 HCCs were clinically diagnosed with follow-up imaging. The sensitivities and specificities of LI-RADS categories were calculated with 95% confidence intervals (CIs).
RESULTSNone of the nine (0%) LR-1, one of 22 (4.5%) LR-2, 14 of 35 (40.0%) LR-3, 19 of 27 (70.4%) LR-4, 73 of 75 (97.3%) LR-5, and all of the five (100%) LR-5V observations were HCCs. Two of 75 (2.7%) LR-5 observations were combined HCC-CCs. Of 13 LR-M lesions, only three (23.1%) were non-HCC malignancies and the remainder were HCCs. Three of six (50%) non-HCC malignancies were categorized as LR-4 or LR-5. When LR-5 and LR-5V were considered positive diagnosis for HCC, the sensitivity and specificity were 63.9% (95% CI, 54.8−72.4) and 96.9% (95% CI, 89.2−99.6), respectively. When LR-4, LR-5, and LR-5V were considered positive, the sensitivity and specificity were 79.5% (95% CI, 71.3−86.3) and 84.4% (95% CI, 73.1−92.2), respectively.
CONCLUSIONIn high-risk patients for HCC, LR-5 and LR-5V showed a high specificity for diagnosing HCC. However, a significant portion of non-HCC malignancies were categorized as LR-4 or LR-5 and a majority of LR-M lesions were HCCs, which suggests the necessity for modification of criteria for the LR-M category.
CLINICAL RELEVANCE/APPLICATIONLR-M category of LI-RADS v2014 needs improvement to better exclude non-HCC malignancies in high risk patients for HCC and to decrease the misdiagnosis rate of HCCs as non-HCC malignancies.
Hepatocellular carcinoma (HCC) is the most common primary malignant tumor of the liver and is the third leading cancer-related cause of death. Major risk factors are identified including cirrhosis caused by viral hepatitis B and C infection, and hereditary hemochromatosis. According to European guidelines patients at risk for HCC should be surveyed with abdominal ultrasound (US) every 6 months.
METHOD AND MATERIALSIn our tertiary referral institute, we selected all patients between October 2005-October 2010 from our HCC database. Inclusion criteria were: pathological or histochemical confirmed HCC, abdominal ultrasound and contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) within three months. Number and size of the detected lesions were compared.
RESULTSIn 88 patients included, 83 HCC lesions were described at US evaluation compared to 181 lesions detected using CT or MRI. Lesions found on US had a mean diameter of 60mm (range 5-135mm, median 45mm), on cross-sectional imaging the mean diameter was 56mm (range 5-160mm, median 45mm). In 30 of the 88 (34%) patients US was negative, while cross-sectional imaging detected HCC lesions. In 18 of those 30 cases all lesions were smaller than 25mm (60%). Of all the patients (n=25) with lesions smaller than 25mm, 18 ultrasound examinations were false negative (72%).
CONCLUSIONSurveillance for HCC in high-risk patients using US is inadequate. A significant number of HCC lesions are missed using US surveillance, especially small lesions, resulting in a false negative examination in a third of the cases.
CLINICAL RELEVANCE/APPLICATIONThis warrants the question if surveillance for HCC should be performed with advanced cross sectional imaging modalities.
Liver Imaging Reporting and Data System (LI-RADS, LR) uses major features (arterial phase hyperenhancement [APHE], “washout” [WO], “capsule”, diameter, threshold growth [TG]) to codify probability of an observation being hepatocellular carcinoma. Inclusion of TG in the LI-RADS algorithm was based on expert opinion rather than scientific evidence. The goal of this study was to assess the effect of TG on LR categorization.
METHOD AND MATERIALSAll MR and CT reports created using a standardized LR v2014 template at one tertiary care center between 4/15–2/16 were identified. For each LR3, LR4, and LR5 reported observation, the presence of every LR major feature was recorded retrospectively. Two LR categories were then assigned: one using all LR-v2014 major features and one using a revised system that disregards TG (LR-revised). Categories assigned using LR-v2014 and LR-revised were compared. Some analyses were repeated excluding patients without prior exams (TG not applicable).
RESULTS136 patients (85 [62%] male, mean age 62 [±10] years) with 297 observations (median diameter 13mm, IQR 9-20mm) were included. Of 297 observations, 204 (69%) had APHE, 186 (63%) had WO, 49 (16%) had “capsule” and 40 (14%) had TG. Of 40 observations with TG, 26 (65%) were new observations ≥10mm, 8 (20%) had diameter increase ≥50% in ≤6 months and 6 (15%) had diameter increase ≥100% in >6 months. LR-v2014 categories were LR3 in 131/297 (44%), LR4 in 87/297 (29%) and LR5 in 79/297 (27%). LR-revised categories were LR3 in 147/297 (50%), LR4 in 78/297 (26%) and LR5 in 72/297 (24%). Assigned categories were discrepant in 22/297 (7%) observations. 7/79 (9%) observations categorized LR5 with LR-v2014 were recategorized LR4 with LR-revised. Of these, all 7 were 10-19 mm and had APHE; 5/7 (71%) were new observations and 2/7 (29%) had diameter increase ≥50% in ≤6 months. When excluding 70 observations without prior exams, 22/227 (10%) categories were discrepant; 7/50 (14%) observations categorized LR5 with LR-v2014 were recategorized LR4 with LR-revised.
CONCLUSIONTG affects LI-RADS category in the minority of cases. Disregarding TG causes a small but meaningful proportion of LR5 observations to be downgraded to LR4.
CLINICAL RELEVANCE/APPLICATIONRemoving threshold growth as a major feature to simplify LI-RADS algorithm would not affect final LI-RADS category in most cases but would cause downgrading of a meaningful proportion of LR5 observations.
To evaluate magnetic resonance elastography (MRE) as a means for predicting the development of hepatocellular carcinoma (HCC) in chronic liver disease.
METHOD AND MATERIALSWe reviewed data from 161 patients with chronic liver disease with the following inclusion criteria: had 2 MRE examinations between days 365–1424 with >12 month interval, no prior history of or development of HCC between the two exams, and available laboratory results. Liver stiffness was classified as low (<3 kPa), moderate (3–4.7 kPa), and high (>4.7 kPa). The classification for change in stiffness between the two MREs is as follows: high on both or high on the first and moderate on the second (group A, n=61), low on both (group C, n=39), and other combinations (group B, n=61). We used Cox analyses and Kaplan-Meyer methods to determine the risk of developing HCC.
RESULTSForty-seven patients (29.2%) developed HCC during the follow-up period (45.9% (28/61) of group A, 27.9% (17/61) of group B; 5.1% (2/39) of group C). There was a significant difference in the disease-free survival rates between groups A (54.9%), B (73.9%), and C (87.6%) at 3-years (p=0.0001). The independent risk factors for development of HCC included: belonging to Group A (hazard ratio [HR] versus group C=6.0, P=0.0028; versus group B=2.16, P=0.0268), age (HR = 1.04, P=0.0154), and alanine aminotransferase level (HR=1.02, P =0.0196).
CONCLUSIONResults from MRE can stratify the risk of developing HCC during the follow-up of patients with chronic liver disease.
CLINICAL RELEVANCE/APPLICATIONPatients with chronic liver disease with high liver stiffness (>4.7 kPa) on their first MR elastography are at high risk for developing an HCC, regardless of the results of their second MR elastography. Thus, they should have meticulous follow-up and be screened for HCC development.
To analyze the value of different parameters derived from CT texture analysis (CTTA) image data in hepatocellular carcinoma (HCC) for prediction of response and response evaluation to transarterial chemoembolization (TACE).
METHOD AND MATERIALSThe study group consisted of 56 HCC in 28 patients (27 male; mean age 67.2±10.4) who underwent CECT examinations before and after TACE therapy. The study was approved by the local ethic committee. Mean time between the two CT examinations was 39.93 ± 62.21 days. Standard of reference was perfusion-CT of the liver additionally to arterial and portal-venous post-contrast phases. Patients were assigned into subgroups: no response-NR (n=9), partial response-PR (n=34) and complete response-CR (n=13). CTTA parameters were: heterogeneity/intensity/average/deviation/skewness/contrast of NGTDM (Neighborhood Grey-Tone Difference Matrix). For each parameter mean, entropy and uniformity were calculated.Blood flow(BF), blood volume(BV), arterial liver perfusion(ALP), portal-venous perfusion(PVP) and hepatic perfusion index(HPI) were calculated in the pre- and post-TACE settings by liver perfusion-CT.
RESULTSPatients with CR showed higher tumor perfusion parameters before TACE and a significant decrease in BF/BV/ALP/HPI after TACE (p=0.002/0.002/0.002/0.003). Patients with PR showed similar results, but only in responding tumor parts. ROC analysis of CTTA parameters yielded predictive cut-off values for CR in the arterial CECT-phase (sensitivity/specificity) for mean intensity (88.9%/69.8%), mean average (88.9%/69.8%) and skewness (90.0%/58.1%) In the portal-venous CECT-phase similar for uniformity of heterogeneity, uniformity of skewness and mean contrast (92.3%/81.8%/92.3%/54.5%/92.3%/95.2%). Significant correlations were registered between changes in mean heterogeneity and BF (p=0.004, r=-0.815), BV (p=0.002, r=-0.851) and ALP (p=0.002, r=-0.851) in the arterial phase in CR and PR.
CONCLUSIONSignificant correlations exist between CTTA parameters and those derived from perfusion-CT both in the pre- and post- TACE setting with predictive value for TACE outcome.
CLINICAL RELEVANCE/APPLICATIONPrediction of response to local therapy (TACE) by means of CTTA can be implemented for choosing the best treatment strategy.Improved response monitoring by CTTA is beneficial for optimal patient management and could be a substitute for more sophisticated imaging techniques like perfusion-CT or MRI.