ELSEVIER
-
clinical ŘADÍOLOGY
MRCR
Diagnostic efficacy of the Liver Imaging-Reporting and Data System (LI-RADS) with CT imaging in categorising small nodules (10-20 mm) detected in the cirrhotic liver at screening ultrasound
CrossMark
M. Abd Alkhalik Basha*, D. Abd El Aziz El Sammak, A.A. El Sammak
Department of Diagnostic Radiology, Zagazig University, Egypt
ARTICLE INFORMATION
Article history:
Received 22 February 2017 Received in revised form 23 May 2017 Accepted 26 May 2017
AIM: To estimate the diagnostic accuracy of the Liver Imaging-Reporting and Data System (LI-RADS) with computed tomography (CT) for diagnosing hepatic nodules (10-20 mm) detected in cirrhotic livers.
MATERIALS AND METHODS: Fifty-five patients with liver cirrhosis and a solitary nodule (10 -20 mm in diameter) detected via ultrasound surveillance, underwent hepatic CT and fine- needle biopsy. All the CT images were analysed and the lesions were categorised into five categories according to the LI-RADS.
RESULTS: Final diagnoses of the 55 nodules were as follows: 34 hepatocellular carcinomas (HCCs), one intrahepatic cholangiocarcinomas, one adrenocortical carcinoma metastasis, and 19 benign lesions. None (0%) of four LI-RADS category 1 lesions, two (22%) of nine category 2 lesions, seven (50%) of 14 category 3 lesions, two (67%) of three category 4 lesions, 22 (96%) of 23 category 5 lesions and one (50%) of two lesions classified as other malignancies was HCC. One category 5 lesion was adrenocortical carcinoma metastasis and one of two lesions categorised as other malignancies was intrahepatic cholangiocarcinoma. In patients with nodules detected at sur- veillance ultrasound, the best threshold for confident HCC diagnosis was more than LI-RADS category 3. The use of this threshold produced a sensitivity and specificity of 72.7% and 90%, respectively. So combining LI-RADS 4 and 5 categories for confident HCC diagnosis would improve accuracy and sensitivity with no significant impairment of specificity or positive predictive value.
CONCLUSION: LIRADS with CT provides a strong validity for the diagnosis of small hepatic nodules, and is very useful to improve the accuracy of CT reports.
@ 2017 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction
Hepatocellular carcinoma (HCC) is the most commonly occurring primary liver malignancy and the third most
common leading cause of cancer death worldwide.1 Pa- tients with liver cirrhosis have a high prevalence of HCC2; hence, the existing management strategy of the American Association for the Study of Liver Disease (AASLD) and the European Association for the Study of the Liver (EASL) ad- vises repeated imaging screening for patients with cirrhosis and a high risk of developing HCC.3
In patients with cirrhosis, small nodules of 10-20 mm detected at ultrasound during routine surveillance repre- sent a diagnostic challenge.4 Contrast-enhanced computed
* Guarantor and correspondent: M. Abd Alkhalik Basha, Department of Diagnostic Radiology, Zagazig University, Egypt. Tel .: +20 01099098900.
E-mail address: drmohammad_basha@yahoo.com (M. Abd Alkhalik Basha).
tomography (CT) and magnetic resonance imaging (MRI) are currently utilised as imaging techniques for the non- invasive diagnosis and screening of HCC. The extensive variability of findings and the utilisation of incompatible and inaccurate terminology by radiologists impair communication with referring physicians and can result in interpretation mistakes. In this scenario, using a stand- ardised terminology and criteria to interpret and report imaging examinations of the liver is an essential require- ment to reduce this dilemma.5
The American College of Radiology organised a com- mittee of radiologists that convened in 2008 to establish a comprehensive system for reporting and interpreting the findings of CT and MRI examinations of the liver in patients at risk of developing HCC. The resulting version 1.0 of the Liver Imaging-Reporting and Data System (LI-RADS) was accessible online in 2011 and updated in 2013 and 2014.6
The aims of the LI-RADS involve decreasing variability in lesion interpretation; increasing communication with cli- nicians; helping decision-making procedures (such as for transplantation, ablative therapy, or chemotherapy); decreasing errors of important information from CT and MRI reports by standardising report content and structure; and enhancing outcome monitoring, performance auditing, quality assurance, and research.7 The LI-RADS offers a comprehensive algorithm that classifies liver observations by their possibility of being HCC from definitely benign (LI- RADS category 1) to definitely HCC (LI-RADS category 5).8
This study aimed to validate the non-invasive diagnostic criteria proposed by the LI-RADS using CT to describe the nature of small nodules (10-20 mm) detected during ul- trasound screening in patients with cirrhosis.
Materials and methods
Patients
This study was conducted from September 2015 to January 2017 and included 55 consecutive adult patients (32 males and 23 females; age range 39-82 years; mean age 58.8±10.8 years). Patients underwent triphasic CT of the liver and fine-needle biopsy (the reference standard).
Inclusion criteria included patients with liver cirrhosis and patients with newly ultrasound-detected solitary small hepatic nodules of 10-20 mm in diameter. Exclusion criteria included patients with hepatic nodules at ultra- sound that did not appear at CT examination (two patients); small nodules <10 mm; patients who refused biopsy (11 patients); severe motion artefacts that rendered part of the dynamic phase of non-diagnostic quality (one patient); and indeterminate histopathology results (three patients).
The hospital’s institutional review board approved the study and informed consent was obtained from each pa- tient before starting the study.
CT
All images were obtained using a 128-section multi- detector CT system (Philips Healthcare) with a collimation
of 128×0.625, a section thickness and interval of 5 mm, a pitch of 0.993, a display field of view of 42 cm2, a matrix of 512×512, and a gantry rotation of 330 ms, 120 kVp, and 250 mAs. The scan started from the diaphragmatic dome down to the iliac crest or the lower margin of the right lobe of the liver. All CT examinations comprised four phases: an unenhanced phase, followed by late hepatic arterial, portal venous, and delayed phases. For the contrast-enhanced phases, patients were injected with non-ionic contrast material (Ultravist 370, Bayer Schering Pharma AG, Berlin, Germany) using a power injector through an antecubital venous access at a rate of 4 ml/s; a total of 90-120 ml (1.5 ml/kg of body weight) was injected. The hepatic arterial phase was triggered 18 seconds after contrast enhancement of the upper abdominal aorta to an attenuation threshold of 150 HU using bolus-tracking software. The portal venous and delayed phases were timed to start at 70 and 180 sec- onds after the start of the contrast medium injection, respectively.
Image analysis
All CT studies were performed within 2 weeks after the detection of nodules using ultrasound. Two radiologists with experience in hepatic imaging (A.A.E. and D.A.E., with 14 and 12 years of experience, respectively) read all CT images in consensus. Any discrepancies in interpretation were resolved by a third radiologist (M.A.B., with 15 years of experience in hepatic imaging). The radiologists were blinded to any clinical information and the results of biopsy, but used the LI-RADS version 2014 algorithm for the assignment of an LI-RADS category for each nodule.
The CT imaging was interpreted considering only the target nodules initially detected at ultrasound screening. The following items obtained at CT imaging were individ- ually evaluated for each target nodule: (1) the diameter: the largest dimension (outer edge to outer edge) of the nodules measured in millimetres; (2) the enhancement pattern in the arterial phase: the degree of contrast material uptake during the arterial phase. The nodules were evaluated in terms of enhancement in comparison to the background liver parenchyma and were classified at visual inspection as: (a) hyperenhancement: enhancement in the arterial phase that is definitely greater than that of the background liver parenchyma; or (b) hypo- or iso-enhancement: enhancement in the arterial phase that is less than or equal to that of the background liver parenchyma; (3) the washout appearance: visually evaluated temporal reduction in the enhancement relative to the background liver pa- renchyma from an earlier to a later phase, resulting in a portal venous phase or delayed phase hypo-enhancement; (4) appearance of the capsule: marginal rim of smooth hyperenhancement in the portal venous or delayed phases that is definitely thicker or more obvious than the rims surrounding the background nodules.
An LI-RADS category was assigned to all detected hepatic nodules using LI-RADS version 2014 algorithm.6 The LI- RADS categories are category 1 (LR1): definitely benign; category 2 (LR2): probably benign; category 3 (LR3):
intermediate probability of HCC; LI-RADS category 4 (LR4): probably HCC; and category 5 (LR5): definitely HCC and split into LR-5us, if there is both washout appearance and visibility as discrete nodules at antecedent ultrasound sur- veillance or LR-5V: tumour in vein; or LR-M: probable malignancy other than HCC.
Reference standard
The diagnoses of the lesions were established based on histological findings after biopsy. All biopsies were taken using a fine needle under ultrasound guidance. Two passes were performed. Aspirates were obtained using 10 ml sy- ringes fitted on 21-G needles. The contents of the needles were applied to glass slides. The slides were then processed and stained with haematoxylin and eosin. All specimens were reviewed by two histopathologists, and the results were obtained by consensus. Indeterminate cases were excluded from the study (three patients).
Statistical analysis
All statistical analyses were accomplished using SAS software (SAS system release 9.2) (SAS Institute Inc, Cary, NC, USA) and Microsoft Excel 2013. Student’s t-test was used to compare between patients with HCC nodules and patients with non-HCC nodules within the LI-RADS cate- gory groups. For each LI-RADS category, the accuracy, sensitivity, specificity, and positive and negative predictive values were reported with 95% confidence intervals. Receiver operating characteristic (ROC) curve analyses were used to reveal the cut-off values, the area under the curve (AUC), and the 95% confidence interval. A p-value ≤0.05 was considered statistically significant.
Results
A total of 55 patients were included in the study, and a small hepatic nodule10-20 mm was revealed in each pa- tient on ultrasound examination. The LI-RADS categories were designated in the 55 target nodules detected at CT. The main patient characteristics are summarised in Table 1.
The final diagnoses in the 55 patients were 34 HCCs (61.8%), one intrahepatic cholangiocarcinoma (1.8%), one metastasis from adrenocortical carcinoma (1.8%), and 19 benign lesions (34.6%). The final diagnosis was confirmed at histopathology examination in all nodules (Fig 1). CT im- aging features of all target nodules are summarised in Table 2.
LI-RADS categories
The LI-RADS categories of the nodules are summarised in Table 3. Four lesions (7.3%) were classified as LR-1, nine (16.4%) as LR-2, 14 (25.5%) as LR-3, three (5.5%) as LR-4, and 23 (41.8%) as LR-5. Two lesions (3.6%) were categorised as LR-M. All four lesions classified as LR-1 were benign. One of these benign lesions was a hemangioma (Fig 2). Two (22%) of the nine lesions classified as LR-2 were eventually diag- nosed as HCC. These lesions had diameters of 11 and 12 mm, respectively. Seven (50%) of the 14 lesions classified as LR-3 were HCCs. The LR-3 nodules eventually diagnosed as HCCs were larger (median size=16 mm [range, 12-18 mm] versus 12 mm [range, 10-14 mm]; p=0.0022) than the LR-3 non- HCC nodules. All lesions >15 mm (n=6) were found to be HCCs. Five of the seven HCC nodules showed arterial phase hyperenhancement without a washout or capsule appear- ance. The remaining two HCC nodules were hypo- and iso- enhancing in the arterial phase without a washout or capsule appearance (Fig 3). The main characteristics of the LR-3 lesions are summarised in Table 4. Three lesions were classified as LR-4, and two were HCCs. They were ≥20 mm and displayed arterial phase hyperenhancement without a washout or capsule appearance (Fig 4). Twenty-three le- sions were classified as LR-5 lesions, and all but one were HCC. The only false-positive result corresponded to a met- astatic nodule from adrenocortical carcinoma (Fig 5). Three LR-5 lesions were <20 mm with arterial phase hyper- enhancement, a washout appearance, and a capsule appearance. Seven LR5 lesions were ≥20 mm with arterial phase hyperenhancement and a washout appearance. Thirteen LR-5 lesions were <20 mm with arterial phase hyperenhancement and a washout appearance and were visible as discrete nodules at antecedent surveillance
| Characteristic | All patients | Patients with HCC | Patients with non-HCC nodules | p-Value | |
|---|---|---|---|---|---|
| No. of patients | 55 | 34 | 21 | ||
| Age (years) | Total | 58.8±10.8 (39-82) | 61.5±10.1 (39-82) | 54.4±9.2 (41-74) | 0.0168 |
| Male | 55.6±9.4 (39-82) | 57±9.7 (39-82) | 52.8±8.6 (41-71) | 0.3388 | |
| Female | 63.2±11.3 (44-81) | 68.6±9.6 (48-81) | 56.1±10 (44-74) | 0.0031 | |
| Sex of patients | Male | 32 | 21 | 11 | |
| Female | 23 | 13 | 10 | ||
| Cause of cirrhosis | Hepatitis C | 50 | 32 | 18 | |
| Hepatitis B | 2 | 1 | 1 | ||
| Primary biliary cirrhosis | 1 | 0 | 1 | ||
| Cirrhosis of unknown cause | 2 | 1 | 1 | ||
| Prothrombin time ratio (%) | 78.5 (33-103) | 76.5 (33-103) | 84 (56-102) | 0.051 | |
| a .- Fetoprotein level (ng/ml) | 7 (1-1163) | 12 (1-1163) | 4(1-168) | 0.061 | |
HCC, hepatocellular carcinoma.
40
35
34
30
No. of patients
25
20
19
15
10
5
1
1
0
HCC
Benign lesions
Cholangiocarcinoma
Metastasis
Series 1
34
19
1
1
Hepatic Nodules
HCC
Benign lesions
Cholangiocarcinoma
Metastasis
ultrasound. These 13 lesions were previously classified as LR-4A in LI-RADS version 2013, but in the latest LI-RADS version 2014,6 they are classified as LR-5us (Fig 6). The two lesions classified as LR-M were one intrahepatic chol- angiocarcinoma and one HCC that displayed atypical findings.
Diagnostic accuracy of LI-RADS for HCC diagnosis
The diagnostic performance of LI-RADS with CT for HCC diagnosis is summarised in Table 5. Considering only those lesions classified as LR-5 for confident HCC diagnosis of nodules detected during ultrasound surveillance in cirrhosis, the LI-RADS had an accuracy, sensitivity, speci- ficity, and positive and negative predictive value of 46.7%, 66.7%, 95.5%, 95.7%, and 65.6%, respectively. Considering LR- 4 and LR-5 together as definitive for HCC, the accuracy, sensitivity, specificity, and positive and negative predictive values were 70.6%, 75%, 91.3%, 92.3%, and 72.4%, respec- tively. Considering LR-3, LR-4, and LR-5 together as defini- tive for HCC, the accuracy, sensitivity, specificity, and positive and negative predictive values were 91.2%, 91.3%, 57.1%, 77.5%, and 80%, respectively.
ROC analyses
ROC analyses of the diagnostic performance of the LI- RADS with CT for a definite diagnosis of HCC (Fig 7a)
revealed an AUC of 0.883 (95% CI: 0.765-0.955, p< 0.0001) with a best threshold for confident HCC diagnosis of >LR-3. The application of this threshold resulted in a sensitivity of 72.7% (95% CI: 54.5-86.7), a specificity of 90% (95% CI: 68.3-98.8), and a likelihood ratio of 7.27 (95% CI: 1.9-27.5).
The LI-RADS showed a sensitivity and specificity of 100% and 20%, respectively, in probably benign nodules (LR-2); 93.9% and 55%, respectively, in indeterminate nodules; and 66.7% and 95%, respectively, in HCC nodules (Table 6).
Analysis of the LR-3 diameters (Fig 7b) yielded an ideal cut-off of >14 mm. The application of this threshold is associated with a sensitivity of 85.71% (95% CI: 42.1-99.6), a specificity of 100% (95% CI: 54.1-100), and a likelihood ratio of 10.5 (95% CI: 1.5-73.2). Thus, if >14 mm is chosen as an additional upgrade criterion for LR-3 nodules, the overall performance improved (AUC=0.929, 95% CI: 0.647-0.999, p<0.0001).
When considering the arterial phase hyperenhancement in the diagnosis of HCC, it revealed an AUC of 0.712 (95% CI: 0.574-0.826, p=0.0008) with a sensitivity of 85.29% (95% CI: 68.9-95), a specificity of 57.14% (95% CI: 34-78.2), and a likelihood ratio of 1.99 (95% CI: 1.2-3.3).
When considering the washout appearance in the diag- nosis of HCC, it revealed an AUC of 0.791 (95% CI: 0.660-0.889, p<0.0001) with a sensitivity of 67.65% (95% CI: 49.5-82.6), a specificity of 90.48% (95% CI: 69.6-98.8), and a likelihood ratio of 7.10 (95% CI: 1.9-27.1). Thus, a
| Feature | All nodules (n=55) | HCCs (n =34) | Non-HCC (n=21) | p-Value | |
|---|---|---|---|---|---|
| Diameter (mm) | 10-15 | 27 (49%) | 9 (26.5%) | 18 (85.7%) | 0.7238 |
| 16-20 | 28 (51%) | 25 (73.5%) | 3 (14.3%) | 0.2961 | |
| Arterial phase hyperenhancement | 38 (69.3%) | 29 (85.3%) | 9 (42.9%) | ||
| Washout | 25 (45.5%) | 23 (67.6%) | 2 (9.5%) | ||
| Capsule | 3 (5.5%) | 3 (8.8%) | 0 | ||
HCC, hepatocellular carcinoma.
| LI-RADS category | Total no. of lesions | Final diagnosis | ||
|---|---|---|---|---|
| HCC lesions | Non-HCC malignant lesions | Benign lesions | ||
| LR-1 | 4 (7.3%) | 0 | 0 | 4 (100%) |
| LR-2 | 9 (16.4%) | 2 (22%) | 0 | 7 (78%) |
| LR-3 | 14 (25.5%) | 7 (50%) | 0 | 7 (50%) |
| LR-4 | 3 (5.5%) | 2 (67%) | 0 | 1 (12.5%) |
| LR-5 | 23 (41.8%) | 22 (96%) | 1 (4%) | 0 |
| LR-M | 2 (3.6%) | 1 (50%) | 1 (50%) | 0 |
| Total no. of lesions | 55 | 34 (61.8%) | 2 (3.6%) | 19 (34.6%) |
washout appearance has a higher specificity but a lower sensitivity than arterial phase hyperenhancement in the diagnosis of HCC.
Discussion
Confident diagnosis of cancer is a critical step before starting treatment. In the majority of malignant diseases, the diagnosis depends mainly on biopsy, but with the improvement of imaging techniques and specific tumour markers, the diagnosis of malignancy can be established without the need for pathological confirmation. These non-
invasive diagnostic criteria are relevant for the manage- ment of patients with suspicion of HCC when the detected nodule is small.º
There is a high prevalence of HCC, which has increased more in recent years; however, there is a great dilemma in the diagnosis and management of HCC. Many clinicians think that any focal lesions in the cirrhotic liver are considered HCC and must be treated immediately by inva- sive procedures. For this reason, many protocols have been designated for the diagnosis and management of HCC so that patients may avoid unnecessary invasive procedures and detection of HCC at a very early stage can be achieved. The LI-RADS is the most recent.
The LI-RADS is a standardised reporting system for liver observations in patients at risk for developing HCC. It has become a broadly accepted tool to enhance communication among radiologists and between radiologists and clinicians.10 10
The present study focused on patients who had a small (10-20 mm) solitary nodule in a cirrhotic liver detected during routine ultrasound surveillance. Larger (>2 cm) nodules were excluded as they are easily diagnosed and smaller (<1 cm) nodules were also excluded as they are difficult to identify using non-invasive imaging examina- tions. Despite continuous developments in imaging tech- nology, imaging detection and description of small hepatic nodules remains limited; moreover, localisation and histo- pathology is challenging, which confuses the evaluation of
(a)
(b)
(c)
(d)
1
1
(a)
(b)
1
1
(c)
(d)
diagnostic accuracy of imaging for such lesions. For these reasons, the LI-RADS and most other systems do not permit the non-invasive diagnosis of HCC for <1 cm lesions.11 The LI-RADS was used in the present study, even though it is not yet universally accepted, because it is the most recent comprehensive system for the diagnosis of HCC and is a dynamic system that continues to be developed and updated with the progress in knowledge and technology. CT was used instead of MRI because of the former’s prevalent availability, lower cost, and relative absence of operator- related variability.
In the present study, the diagnosis was based on histo- pathology results in 55 patients with cirrhosis in whom a solitary nodule of 10-20 mm was detected during ultra- sound screening. In addition to 34 HCC nodules, 19 benign nodules, one metastatic nodule from adrenocortical carci- noma, and one nodule of cholangiocarcinoma were found.
The prevalence of HCC in the present study was 61.8% (34/55), and this result similar to the results of Forner
| Feature | All nodules (n=14) | HCCs (n=7) | Non-HCC (n=7) |
|---|---|---|---|
| Diameter (mm) 10-15 | 8 (57.1%) | 1 (14.3%) | 7 (100%) |
| 16-20 | 6 (42.9) | 6 (85.7%) | 0 |
| Arterial phase hyperenhancement | 7 (53.8%) | 5 (71%) | 2 (50%) |
| Washout | 0 | 0 | 0 |
| Capsule | 0 | 0 | 0 |
HCC, hepatocellular carcinoma.
et al.,12 who showed that the prevalence of HCC was 67.4% (60/89), but higher than the result of Becker et al.,13 who found that 55 of 104 (52.9%) liver observations were HCCs.
A single case of cholangiocarcinoma was found; a neoplasm that is not as common as HCC, but might also develop in a cirrhotic liver,14 and this case was correctly categorised as LR-M.
Nineteen benign lesions (34.6%) were diagnosed, indi- cating a high prevalence of benign lesions in the cirrhotic liver. This result is higher than the result of Darnell et al.,15 who reported 27 benign lesions (20.3%), but lower than result of Becker et al.,13 who reported 49 benign lesions (47.1%).
In the present study, the most commonly visualised major feature was arterial phase hyperenhancement fol- lowed by a washout appearance where 29 (85.3%) of 34 HCC nodules displayed arterial phase hyperenhancement, 23 (67.6%) displayed a washout appearance, and three (8.8%) displayed a capsule appearance. The washout appearance has a higher specificity but a lower sensitivity than arterial phase hyperenhancement (90.48% versus 57.14% and 67.65% versus 85.29%, respectively) in the diagnosis of HCC. These results are similar to the results of Ehman et al.,16 who re- ported arterial phase hyperenhancement in 86%, a washout appearance in 82%, and a capsule appearance in 27% of lesions.
Using the latest LI-RADS version, the present study demonstrated that if both LR-4 and LR-5 are combined as definitely indicating HCC, the specificity is minimally decreased (91.3%), but the accuracy and sensitivity are
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
(e)
substantially increased (70.6% and 75%, respectively). The ROC analyses revealed an AUC of 0.883 (p<0.0001) with a best threshold >LR-3 for confident HCC diagnosis. The application of this threshold resulted in a sensitivity of 72.7% and a specificity of 90%. These results are similar to the results of previous studies15-17; thus, in agreement with these studies, both LR-4 and LR-5 should be combined for the diagnosis of HCC because if LR-5 alone is considered to be conclusive for HCC diagnosis, the current LI-RADS version would miss several HCCs.
The present study revealed two (22%) of nine lesions classified as LR-2 were eventually diagnosed as HCC. This result is similar to the results of Darnell et al.,15 who re- ported two (25%) of eight lesions classified as LR-2 were HCC, but this result is different from the result of Zhenhui et al.,17 who reported 48 lesions in LR-1 and LR-2 that were
all benign. This percentage is relatively high; therefore, an additional active diagnostic work-up is strongly recom- mended to reduce this percentage.
The present study revealed seven (50%) of 14 lesions classified as LR-3 were HCCs. This incidence is similar to the results of Sangiovanni et al.18 and Leoni et al.,19 but this high incidence of HCC in LR-3 lesions is different from that of a recent retrospective study by Choi et al.,20 where only four (6%) of 69 lesions were categorised as probable or definitive HCCs after 1 year of follow-up; however, this study had several pitfalls.
One of the seven LR-3 nodules <15 mm was ultimately diagnosed as HCC. By contrast, all LR-3 lesions >15 mm (n=6) were eventually diagnosed as HCC. Consistent with previously published studies,13,21 the present results showed that a cut-off value of 15 mm early HCC to be
| Diagnostic Performance | Diagnostic criteria | |||||||
|---|---|---|---|---|---|---|---|---|
| LR-4 | LR-5 | LR-4 + LR-5 | LR-3 + LR-4 + LR-5 | |||||
| % | 95% CI | % | 95% CI | % | 95% CI | % | 95% CI | |
| Sensitivity | 15.4 | 1.9-45.5 | 66.7 | 48.2-82 | 75 | 56.6-88.5 | 91.2 | 76.3-98.1 |
| Specificity | 95.5 | 77.2-99.9 | 95.5 | 77.2-99.9 | 91.3 | 72-98.9 | 57.1 | 34-78.1 |
| AUC | 0.55 | 0.38-0.72 | 0.81 | 0.68-0.90 | 0.83 | 0.71-0.92 | 0.74 | 0.61-0.85 |
| Positive likelihood ratio | 3.38 | 0.34-33.8 | 14.7 | 2.1-101 | 8.6 | 2.3-32.9 | 2.1 | 1.3-3.5 |
| Negative likelihood ratio | 0.89 | 0.69-1.14 | 0.35 | 0.2-0.6 | 0.27 | 0.15-0.51 | 0.15 | 0.05-0.48 |
| Disease prevalence | 37.1 | 21.5-55.1 | 60 | 45.9-73 | 58.2 | 44.1-71.4 | 61.8 | 47.7-74.6 |
| Positive predictive value | 66.7 | 9.4-99.2 | 95.7 | 78.1-99.9 | 92.3 | 74.9-99.1 | 77.5 | 61.6-89.2 |
| Negative predictive value | 65.6 | 46.8-81.4 | 65.6 | 46.8-81.4 | 72.4 | 52.7-87.3 | 80 | 51.9-95.7 |
AUC, area under the receiver operating characteristic curve; CI, confidence interval.
differentiated from benign hepatocellular nodules but was in contrast to the study of Darnell et al.,15 which suggested that the cut-off should be better set at 10 mm.
An interesting finding in the present study is that a small number of lesions were classified as LR-4 (3); this number is low in comparison to previous studies15,17 because the ex- istence of a prior positive ultrasound result was incorpo- rated to grade LR-4 nodules as conclusive for HCC according to the most recent LI-RADS version 2014.6 These nodules were categorised as LR-4A in the previous LI-RADS version. Hence, any case with a ultrasound-detected nodule that showed arterial phase hyperenhancement and a washout appearance was considered to be a LR-5 lesion. This inclu- sion of ultrasound in the LI-RADS algorithm increased the sensitivity for the diagnosis of HCC and become congruent with the recent literature,22 where most 1-2 cm lesions with arterial phase hyperenhancement and a single
additional major imaging feature are categorised as LR-4; however, they can be categorised as LR-5us if they were visible as discrete nodules at antecedent surveillance US.
The present study is compatible with the research of Choi et al.,20 where the present results showed that most non-hyperenhancing nodules in cirrhotic patients were benign. Only four of 27 non-hyperenhancing nodules were found to be LR-5. This in contrast to the study of Park et al.,23 which reported that many well-differentiated HCCs, particularly early HCCs, do not display arterial enhancement in the early phases of dynamic imaging owing to inadequate neovascularity. Therefore, for the evaluation of non- hyperenhancing nodules and to reach the correct diag- nosis of HCC, MRI is necessary and may be superior to CT for the identification of ancillary features, such as mild to moderate T2 hyperintensity, a capsule appearance, the presence of nodule-in-nodule appearance, and an interval
LI-RADS
LR3 diameter
100
100
90
90
80
Sensitivity: 85.7
Specificity: 100.0
80
Criterion: >14
70
70
Sensitivity: 72.7
Sensitivity
Sensitivity
60
Specificity: 90.0
60
Criterion:>3
50
50
40
40
30
30
20
20
10
10
0
0
0 10 20 30 40 50 60 7
70
80
90
100
0
10 20 30 40 50 60 70 80 90 100
100-Specificity
100-Specificity
(a)
(b)
| Criterion | Sensitivity | 95% CI | Specificity | 95% CI | +LR | 95% CI | -LR | 95% CI |
|---|---|---|---|---|---|---|---|---|
| ≥1 | 100 | 89.4-100 | 0 | 0-16.8 | 1 | 1-1 | ||
| >1 | 100 | 89.4-100 | 20 | 5.7-43.7 | 1.25 | 1-1.6 | 0 | |
| >2 | 93.94 | 79.8-99.3 | 55 | 31.5-76.9 | 2.09 | 1.3-3.4 | 0.11 | 0.03-0.4 |
| >3 | 72.73 | 54.5-86.7 | 90 | 68.3-98.8 | 7.27 | 1.9-27.5 | 0.30 | 0.2-0.5 |
| >4 | 66.67 | 48.2-82 | 95 | 75.1-99.9 | 13.33 | 1.9-91.5 | 0.35 | 0.2-0.6 |
| >5 | 0 | 0.0-10.6 | 100 | 83.2-100 | 1 | 1-1 |
CI, confidence interval; +LR, positive log ratio; - LR, negative log ratio.
follow-up.24,25 In addition, O’Malley et al.26 reported that most arterial phase-enhancing nodules 10-20 mm in size are not HCCs.
When comparing the present study using CT to the research of Darnell et al.15 using MRI, it was found that there was no significant statistical difference regarding the diag- nostic performance of the LI-RADS with CT and the LI-RADS with MRI in the categorisation of small hepatic nodules (p=0.4600). This agrees with the results of Lin et al.,27 which proved that the use of four-phase dynamic CT and MRI demonstrate similar diagnostic accuracy for HCC in tu- mours of 1-2 cm (p=0.587) and with the results of Rao,28 which proved that CT and MRI have a sensitivity of 60-94% and 58.5-93%, respectively, for nodules >1 cm; however, this is in contrast to the previous study of Lee et al.,29 which reported a higher sensitivity for liver MRI than liver CT in a study of 78 patients, and also in contrast to the most recent studies8,30 that reported considerable discordance between CT and MRI in LI-RADS categorisation, mainly because of the superiority of MRI for detecting the presence of imaging features or lesions themselves. The absence of significant differences between the present study using CT and that of Darnell et al.15 using MRI as the result of fewer cases and the exclusion of nodules <10 mm and nodules that did not appear on CT. Thus, to compare between CT and MRI in the LI-RADS categorisation of he- patic nodules, a separate study is required using CT and MRI together.
The present study had some limitations. First, patients whose nodules did not appear on CT imaging were excluded. Second, a small number of patients was included as many refused biopsy. Third, nodules <10 mm were excluded. Finally, the LI-RADS is still under modification.
In conclusion, the combination of LI-RADS with CT offers a powerful tool for the diagnosis of HCC and is valuable for improving the accuracy of CT reports and facilitating decision-making. If ultrasound screening in high-risk pa- tients with cirrhosis shows a new nodule that is 10-20 mm in diameter, the categorisation of this nodule as LR-1 at subsequent CT imaging is highly accurate and sensitive for definitely benign lesions (100%), and the categorisation of this nodule as LR-4 or LR-5 is highly specific for HCC (95.5%) but less sensitive (67%). A major proportion of LR-2 (22%) and LR-3 (50%) nodules were HCCs. Thus, utilising LI-RADS with CT for the diagnosis of small nodules in cirrhotic pa- tients requires an effective diagnostic work-up and
advanced modifications to diagnose and treat HCC at a very early stage.
Acknowledgements
The authors thank all staff members and colleagues in the Radiology Department-Zagazig University for their helpful cooperation and all the study participants for their patience and support.
References
1. Purysko AS, Remer EM, Coppa CP, et al. LI-RADS: a case-based review of the new categorization of liver findings in patients with end-stage liver disease. RadioGraphics 2012 Nov 1;32(7):1977-95.
2. Llovet JM, Burroughs A, Bruix J. Hepatocellular carcinoma. Lancet 2003 Dec 6;362(9399):1907-17.
3. European Association for the Study of the Liver. EASL-EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012 Apr 30;56(4):908-43.
4. Brancatelli G, Baron RL, Peterson MS, et al. Helical CT screening for hepato- cellular carcinoma in patients with cirrhosis: frequency and causes of false- positive interpretation. AJR Am J Roentgenol 2003 Apr; 180(4):1007-14.
5. Vega B, Pineda DL, Martin EE, et al. A Practical case based review of LIRADS. In ECR 2014. Poster C-1311. http://dx.doi.org/10.1594/ecr2014/C- 1311.
6. American College of Radiology. Liver Imaging-Reporting And Data System. Published March 2011. Updated 2013. Available at: http://www.acr.org. Accessed July 2014.
7. American College of Radiology. Quality and safety resources: Liver Imag- ing-Reporting and Data System. Available at: http://www.acr.org/ QualitySafety/Resources/LIRADS. Accessed 22 April 2012.
8. Corwin MT, Fananapazir G, Jin M, et al. Differences in liver imaging and reporting data system categorization between MRI and CT. AJR Am J Roentgenol 2016 Feb;206(2):307-12.
9. Kojiro M, Roskams T. Early hepatocellular carcinoma and dysplastic nodules. Semin Liver Dis 2005;25(2):133-42.
10. American College of Radiology. Liver Imaging-Reporting and Data System, version 2014-C. Available at: http://www.acr.org/QualitySafety/ Resources/LIRADS/. Accessed: December 2015.
11. Tang A, Valasek MA, Sirlin CB. Update on the Liver Imaging Reporting and Data System: what the pathologist needs to know. Adv Anat Pathol 2015 Sep 1;22(5):314-22.
12. Forner A, Vilana R, Ayuso C, et al. Diagnosis of hepatic nodules 20 mm or smaller in cirrhosis: prospective validation of the noninvasive diagnostic criteria for hepatocellular carcinoma. Hepatology 2008 Jan 1;47(1):97-104.
13. Becker AS, Barth BK, Marquez PH, et al. Increased interreader agreement in diagnosis of hepatocellular carcinoma using an adapted LI-RADS al- gorithm. Eur J Radiol 2017 Jan 31;86:33-40.
14. Malhi H, Gores GJ. Cholangiocarcinoma: modern advances in under- standing a deadly old disease. J Hepatol 2006 Dec 31;45(6):856-67.
15. Darnell A, Forner A, Rimola J, et al. Liver imaging reporting and data system with MR imaging: evaluation in nodules 20 mm or smaller
detected in cirrhosis at screening US. Radiology 2015 Feb 5;275(3):698-707.
16. Ehman EC, Behr SC, Umetsu SE, et al. Rate of observation and inter- observer agreement for LI-RADS major features at CT and MRI in 184 pathology proven hepatocellular carcinomas. Abdom Radiol 2016 May 1;41(5):963-9.
17. Zhenhui LI, Dafu Z, Depei GAO, et al. Application of LI-RADS classification criteria for diagnosis of primary hepatocellular carcinoma. J Clin Radiol 2016;3:384-8.
18. Sangiovanni A, Manini MA, Iavarone M, et al. The diagnostic and eco- nomic impact of contrast imaging techniques in the diagnosis of small hepatocellular carcinoma in cirrhosis. Gut 2010 May 1;59(5):638-44.
19. Leoni S, Piscaglia F, Golfieri R, et al. The impact of vascular and nonvascular findings on the noninvasive diagnosis of small hepatocel- lular carcinoma based on the EASL and AASLD criteria. Am J Gastro- enterol 2010 Mar 1;105(3):599-609.
20. Choi JY, Cho HC, Sun M, et al. Indeterminate observations (liver imaging reporting and data system category 3) on MRI in the cirrhotic liver: fate and clinical implications. AJR Am J Roentgenol 2013 Nov;201(5):993-1001.
21. Fischer MA, Brehmer K, Svensson A, et al. Renal versus splenic maximum slope based perfusion CT modelling in patients with portal-hyperten- sion. Eur Radiol 2016 Nov 1;26(11):4030-6.
22. An C, Rakhmonova G, Choi JY, et al. Liver imaging reporting and data system (LI-RADS) version 2014: understanding and application of the diagnostic algorithm. Clin Mol Hepatol 2016 Jun;22(2):296.
23. Park YN, Kim MJ. Hepatocarcinogenesis: imaging-pathologic correla- tion. Abdom Imaging 2011 Jun 1;36(3):232-43.
24. Khan AS, Hussain HK, Johnson TD, et al. Value of delayed hypointensity and delayed enhancing rim in magnetic resonance imaging diagnosis of small hepatocellular carcinoma in the cirrhotic liver. J Magn Reson Im- aging 2010 Aug 1;32(2):360-6.
25. Kim MJ. Current limitations and potential breakthroughs for the early diagnosis of hepatocellular carcinoma. Gut Liver 2011 Mar 1;5(1):15-21.
26. O’malley ME, Takayama Y, Sherman M. Outcome of small (10-20 mm) arterial phase-enhancing nodules seen on triphasic liver CT in patients with cirrhosis or chronic liver disease. Am J Gastroenterol 2005 Jul 1;100(7):1523-8.
27. Lin MT, Wang CC, Cheng YF, et al. Comprehensive comparison of multiple-detector computed tomography and dynamic magnetic reso- nance imaging in the diagnosis of hepatocellular carcinoma with vary- ing degrees of fibrosis. PloS One 2016 Nov 9;11(11):e0166157.
28. Rao PN. Nodule in liver: investigations, differential diagnosis, and follow-up. J Clin Exper Hepatol 2014 Aug 31;4:S57-62.
29. Lee DH, Kim SH, Lee JM, et al. Diagnostic performance of multidetector row computed tomography, superparamagnetic iron oxide-enhanced magnetic resonance imaging, and dual-contrast magnetic resonance imaging in predicting the appropriateness of a transplant recipient based on Milan criteria: correlation with histopathological findings. Invest Radiol 2009;44(6):311-21.
30. Zhang YD, Zhu FP, Xu X, et al. Liver Imaging Reporting and Data System: substantial discordance between CT and MR for imaging classification of hepatic nodules. Acad Radiol 2016 Mar 31;23(3):344-52.