Hepatobiliary phase hypointensity on gadobenate dimeglumine-enhanced magnetic resonance imaging may improve the diagnosis of hepatocellular carcinoma

Abstract

Background

To determine the clinical value of hepatobiliary phase (HBP) hypointensity for noninvasive diagnosis of hepatocellular carcinoma (HCC).

Methods

A total of 246 high-risk patients with 263 selected nodules (126 HCCs, 137 non-HCCs) undergoing gadobenate dimeglumine (Gd-BOPTA)-enhanced magnetic resonance imaging (MRI) were included in the study. Imaging-based diagnoses of small (≤3 cm) and large (>3 cm) HCCs were made using the following 4 criteria: (I) non-rim arterial phase hyper-enhancement (APHE) plus hypointensity on the portal venous phase (PVP); (II) non-rim APHE plus hypointensity on the PVP and/or transitional phase (TP); (III) non-rim APHE plus hypointensity on the PVP and/or TP and/or HBP; (IV) criterion 3 plus non-LR-1/2/M. Based on typical imaging features, LR-1, LR-2, or LR-M (if definitely benign, probably benign, malignant but not HCC specific, respectively) were defined according to the Liver Imaging Reporting and Data System (LI-RADS). Sensitivities and specificities of imaging criteria were calculated and compared using McNemar’s test.

Results

Among the diagnostic criteria for small HCCs, criterion 3 and 4, which included HBP hypointensity, showed significantly higher sensitivities (96.4% and 94.6%, respectively) than criterion 1 (58.9%, P<0.001 for both). Moreover, criterion 4, which included HBP hypointensity and ancillary features, showed significantly higher specificity (94.7%) than criterion 3 (66.7%, P<0.001) and comparable specificity to criterion 1 (97.4%, P=0.375), achieving the highest accuracies (94.7%). The diagnostic performance of criterion 4 for large HCCs was similar to that for small HCCs.

Conclusions

HBP hypointensity acquired from Gd-BOPTA-MRI can improve sensitivity and maintain high specificity in the diagnosis of both small and large HCCs after excluding benignities or non-HCC malignancies according to characteristic imaging features.

Introduction

Liver cancer, with hepatocellular carcinoma (HCC) constituting 75–85% of its total incidence, was predicted to be the sixth most commonly diagnosed cancer and the fourth leading cause of cancer death worldwide in 2018, with about 841,000 new cases and 782,000 deaths annually (1). Unlike most solid cancers, HCC can be diagnosed and treated based solely on noninvasive imaging without biopsy confirmation using arterial phase hyperenhancement (APHE) and washout on the portal venous phase (PVP) or delayed phase (2,3). Recently, several meta-analyses have suggested that dynamic contrast-enhanced magnetic resonance imaging (MRI) may be more sensitive than dynamic contrast-enhanced computed tomography (CT) for HCC diagnosis (4-7). Furthermore, 2 meta-analyses have found that MRI paired with hepatobiliary contrast agents to be of higher sensitivity than when paired with extracellular agents, particularly in small HCCs (6,8). In contrast to conventional extracellular contrast agent, a hepatobiliary contrast agent is selectively taken up by functioning hepatocytes through specific organic anion transporting polypeptide (OATP) transporters located on the hepatocyte surface, allowing the acquisition of hepatobiliary phase (HBP) images (9-11).

As most HCCs (80–90%) demonstrate hypointensity in the HBP, this feature may contribute to the differentiation of HCC from benign hepatic nodules developed in chronic liver diseases (12,13). However, until now, there has been no established consensus regarding the value of HBP hypointensity during liver MRI. In East Asia, to achieve early detection of HCC, some guidelines attach importance to the use of HBP hypointensity, even as an alternative to the washout appearance of HCCs (14-16). Meanwhile, in the West, where transplantation is the major curative treatment for HCC developing in the cirrhotic liver, the practice guidelines suggest that washout should be determined on the PVP in hepatobiliary-contrast-enhanced MRI, rather than on the transitional phase (TP) or HBP, so as to obtain the highest specificity (17,18).

In fact, prior studies have suggested that HBP hypointensity is highly sensitive and specific for HCC when combining with non-rim APHE (19-21). In particular, Joo and Kim et al. showed that HBP hypointensity acquired from gadoxetic acid (Gd-EOB-DTPA)-enhanced MRI was an effective alternative to washout after exclusion of nodules suggestive of benignity or non-HCC malignancy according to characteristic imaging features, like hemangioma and intrahepatic cholangiocarcinoma. This procedure enabled a highly sensitive diagnosis with little loss in specificity for HCC (22,23). In these studies, however, 2 reference standards might have led to incorporation bias, especially for malignant lesions confirmed by imaging features. Moreover, whether HBP hypointensity on MRI using another hepatobiliary contrast agent like gadobenate dimeglumine (Gd-BOPTA) is helpful for small and/or large HCC diagnoses remains unknown. Thus, as an extension to Joo and Kim et al.’s study, we aimed to investigate the diagnostic performance of HBP hypointensity for the detection of small and large HCCs by using a more strict inclusion criteria in which all malignancies were confirmed pathologically.

Methods

Study participants

This single-center retrospective cohort study was approved by the Institutional Review Board, who determined the requirement for informed consent could be waived. Between January 2016 and January 2018, the research team searched the radiologic diagnoses of MRI scans with keywords “lesion”, “nodular”, “mass”, “tumor”, or “carcinoma” of “liver”, which, according to our abdominal radiologists, are all the most frequently used words in the reports on hepatic lesions found on the local picture archiving and communication system (PACS). In total, 6,031 abdominal gadobenate Gd-BOPTA)-enhanced MRI scans were extracted.

The electronic medical records and MR images were reviewed by 1 radiologist with 3 years of experience in the interpretation of abdominal MRI. In sum, 899 MR exams satisfied the following criteria: (I) hepatic nodules in patients with high-risk factors, namely histopathological or clinical diagnosis of liver cirrhosis or chronic hepatitis B viral infection according to the Liver Imaging Reporting and Data System (LI-RADS) 2018 (18); (II) hepatic lesions in patients without known extrahepatic cancer; (III) without any previous treatment for hepatic lesions, such as surgery, trans-arterial chemoembolization (TACE), or radiofrequency ablation; (IV) hepatic lesions confirmed by histopathology or benign lesions confirmed by characteristic imaging and follow-up of a minimum of 12 months. Subsequently, 636 MRI scans were gradually excluded due to the following criteria: (I) multiple MRI scans were excluded in favor of the first scan in a patient, as some benign lesions might have been followed up by undergoing more than one MRI scan (n=450); (II) suboptimal MR image quality for interpretation (n=7), or without HBP scan (n=95); (III) hepatic lesions measuring smaller than 0.5 cm (n=45); (IV) hepatic lesions with definite acute or subacute infection, which suggest the diagnosis of abscess (n=39). To avoid data clustering, for patients with multiple lesions, the radiologist selected the largest one if hepatic lesions had similar diagnoses. Finally, a total of 246 consecutive patients with 263 qualifying lesions comprising 170 small lesions and 93 large lesions were included for image analysis (Figure 1).

Figure 1.

Flowchart of inclusion and exclusion criteria for patients. HCC, hepatocellular carcinoma.

Reference standard

Of the 263 qualifying lesions, 126 HCCs were all confirmed pathologically (17 cases by biopsy, 109 cases by surgery), and the remaining 137 lesions were determined to be non-HCCs. Of the 137 non-HCC lesions, 16 malignant lesions (4 cases by biopsy, 12 cases by surgery) were all confirmed by histopathology including combined hepatocellular and cholangiocarcinomas (cHCC-CCs) (n=2), intrahepatic cholangiocarcinomas (ICCs) (n=10), metastases (n=2), and neuroendocrine neoplasms (n=2), while 121 non-HCC benign lesions were confirmed by histopathology (n=17) or by characteristic imaging findings and follow-up of a minimum of 12 months (n=104) (Table 1). After measuring the largest outer-edge-to-outer-edge dimension of the lesions, the radiologist divided them into small (≤3 cm) and large (>3 cm) lesions groups.

Table 1. Lesion confirmation of non-HCC benign lesions.

Non-HCC benign lesions	Confirmed by imaging and follow-up	Number of lesions
Hemangioma	Classic enhancement pattern on dynamic contrast-enhanced MRI, peripheral nodular enhancement in AP followed by progressive centripetal filling in the delay phase, and marked hyperintensity on T2-weighted MR imaging, without interval change during follow-up	35
APS	Isointensity on dynamic MRI other than AP images, isointensity on the HBP, and no change or a decrease in size during follow-up	28
FNH or FNH-like nodules	APHE plus iso- or hypersignal intensity during the portal-venous phase, at least partial hyperintensity compared to the liver parenchyma in the hepatobiliary phase, and no interval change	9
RN-DN	Hyperintensity on T1-weighted imaging and isointensity or hypointensity on T2-weighted imaging, isointensity in AP image or APHE without discernible “washout”, and no interval change during follow-up	32

HCC, hepatocellular carcinoma; AP, arterial phase; APS, arterioportal shunt; FNH, focal nodular hyperplasia; RN, regenerative nodule; DN, dysplastic nodule; AP, arterial phase; APHE, arterial phase hyperenhancement; HBP, hepatobiliary phase.

Image acquisition

MR images were acquired by the 3.0T MR imaging system (Magnetom Verio; Siemens Healthcare, Erlangen, Germany) equipped with an 8-channel body phased-array coil. The routine MR sequences included a breath-hold in- and opposed-phase T1-weighted gradient-echo sequence, a respiratory-triggered T2-weighted fat-saturated turbo spin echo sequence, and diffusion-weighted imaging (DWI) performed with respiratory triggering by using a single-shot echoplanar imaging sequence (b values of 50 and 800 sec/mm²). Dynamic MR images were acquired before and after contrast agent administration. A weight-based dose of 0.1mmol/kg of gadobenate dimeglumine (Multihance, Bracco Imaging S.p.A) was intravenously administered by a using power injector at 2 mL/s, and subsequently followed by a 20-mL saline flush. Dynamic images, including unenhanced, arterial phase (AP) (20–25 seconds), PVP (60–70 seconds), TP (2 minutes), and delayed HBP (80–100 minutes), were obtained using a T1-weighted three-dimensional volumetric interpolated breath-hold sequences with the same parameters.

Imaging analysis

All MR images were reviewed independently by another 2 experienced abdominal radiologists (with 30 and 4 years of experience in the interpretation of liver MRI) who were blinded to the clinical data and the final diagnosis of the lesions. For each target lesion, each radiologist evaluated the following parameters in both the small and large hepatic lesion groups: (I) non-rim APHE, referring to signal intensity (SI) of whole or partial lesions, was unequivocally greater than the surrounding liver parenchyma with non-rim appearance in the AP; (II) for non-rim APHE nodules, hypointensity on the PVP, TP, and HBP was qualitatively defined when any part of the nodule showing arterial hyper-enhancement presented a corresponding hypointense area relative to the surrounding liver; (III) the largest outer-edge-to-outer-edge dimension of an observation was measured in a clearly visible phase and appropriate plane; (IV) lesions of typical imaging features suggesting benignity were designated as LR-1 or LR-2 according to LI-RADS (18,24,25), whereas LR-M was determined if the lesion exhibited any of rim APHE, peripheral “washout”, delayed central enhancement, targetoid TP or HBP appearance, or targetoid restriction, which indicated non-HCC malignancy (Figure 2).

Figure 2.

A histopathologically confirmed intrahepatic cholangiocarcinoma (ICC) in a 44-year-old male patient. A hepatic lesion located in the left lobe of the liver shows mild-to-moderate hyperintensity on T2-weighted imaging and targetoid restriction (black ring) on an apparent diffusion coefficient (ADC) map (A,B). On pre- (C) and post-contrast (D,E,F,G) enhanced images, the nodule demonstrates rim arterial phase hyper-enhancement (D), peripheral “washout” on portal venous phase (arrow), delayed central enhancement (triangle) on transitional phase, and targetoid HBP appearance (white ring). The pathological result (H) is poorly differentiated intrahepatic cholangiocarcinoma (magnification, 10×20).

In addition to the results of image review, the 2 radiologists applied 4 different imaging criteria for HCC diagnosis which were as follows: (I) non-rim APHE plus hypointensity on PVP; (II) non-rim APHE plus hypointensity on PVP and/or TP; (III) non-rim APHE plus hypointensity on PVP and/or TP and/or HBP; (IV) non-rim APHE, hypointensity on PVP and/or TP and/or HBP plus non-LR-1/2/M. When any interobserver disagreement arose, the reviewers reevaluated the images together and reached a consensus.

Statistical analysis

The degree of agreement between the two reviewers was assessed by using Cohen k statistics as follows: excellent (k=0.81–1.0), good (k=0.61–0.80), moderate (k=0.41–0.60), fair (k=0.21–0.40), and poor (k=0–0.20). Sensitivity, specificity, and accuracy were calculated for each imaging diagnostic criterion of HCC in small and large lesions group, respectively. Thereafter, sensitivities and specificities of the imaging criteria were compared each other using McNemar’s test. The diagnostic performance of each criterion was also compared by using ROC analysis. Statistical analyses were performed using SPSS software version 25.0. A P value of less than 0.05 was considered to indicate statistical significance.

Results

Patient characteristics

A total of 263 nodules composed of 126 HCCs (mean size, 5.3 cm; range, 0.8–21.1 cm) and 137 non-HCCs (mean size, 2.0 cm; range, 0.5-17.2 cm) were found in 246 patients. The clinical characteristics of the 246 patients are summarized in Table 2. There were 202 males (82.1%) and 44 females (17.9%), with a mean age of 50.0 years (range, 20–83 years). Hepatitis B was the predominant risk factor (n=232, 94.3%), followed by alcoholism (n=6, 2.5%). Elevated serum alpha-fetoprotein (>20 ng/mL) was found in 76 patients. There were 56 HCCs (mean size, 2.0±0.57 cm) and 114 non-HCCs (mean size, 1.2±0.61 cm) in the small hepatic lesion group, while 70 HCCs (mean size, 7.9±4.15 cm) and 23 non-HCCs (mean size, 6.2±3.2 cm) were found in the large lesion group.

Table 2. Clinical characteristics.

Variables	Total (n=246)
Mean age (y)	50.0±11.6
Sex, n (%)
Male	202 (82.1)
Female	44 (17.9)
Risk factors, n (%)
Hepatitis B	232 (94.3)
Hepatitis C	4 (1.6)
Alcoholism	6 (2.5)
Cryptogenic cirrhosis	4 (1.6)
AFP level (ng/mL)
>20	76 (30.9)
≤20	170 (69.1)
Number of nodules, n (%)
1	231 (93.9)
2	13 (5.3)
3	2 (0.8)
Mean size of nodules (cm)
HCC	5.3
Non-HCC	2.0

Data are mean ± standard deviation or number (percentage). AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma.

Diagnostic performance of imaging criteria for small hepatic lesions

Interobserver agreement was good to excellent for all imaging features (k=0.611–0.852), especially for criterion 1 (k=0.852) of small hepatic lesions. The sensitivity, specificity, and accuracy of each imaging criterion for small lesions were calculated and are listed in Table 3. Of the 114 non-HCCs, only 3 lesions showed non-rim APHE and hypointensity on PVP (criterion 1), including 1 cHCC-CC and 2 dysplastic nodules (DN). Most nodules (33/36) which met criterion 1 were HCCs (Figure 3). Thus, imaging criterion 1 provided the highest specificity (97.4%) but the lowest sensitivity (58.9%). Although most small HCCs (55/56) showed non-rim APHE, 22 of them showed no “washout” in PVP, and 9 HCCs showed hyper or isointensity in TP, while these lesions all showed HBP hypointensity expect for 1 HCC (Figure 4). Therefore, non-rim APHE plus hypointensity on PVP and/or TP and/or HBP (criterion 3) showed significantly higher sensitivity (96.4%) than criterion 2 (non-rim APHE plus hypointensity on PVP and/or TP) (82.1%, P=0.008) and criterion 1 (P<0.001), but had lower specificity (66.4%) than both criterion 1 (P<0.001) and criterion 2 (97.4%, P<0.001). As many hemangiomas and ICCs can demonstrate HBP hypointensity, after excluding classical hallmarks of hemangioma/ICC designated as LR-1, LR-2, or LR-M, criterion 4 (non-rim APHE plus hypointensity on PVP and/or TP and/or HBP plus non-LR-1/2/M) showed significantly higher specificity (94.7%) than criterion 3 (P<0.001) and comparable specificity to criterion 1 (P=0.375), achieving the highest accuracies (94.7%). Figure 5 presents the receiver operating characteristic (ROC) curves of the 4 imaging criteria. In the diagnosis of small HCC, criterion 4 showed excellent discriminative power with area under curve (AUC) of 0.934, with criteria 1–3 demonstrating AUCs of 0.781, 0.898, and 0.815, respectively.

Table 3. Diagnostic performance of different imaging criteria of small hepatocellular carcinoma.

Imaging criteria	Sensitivity					Specificity				Accuracy (%)
	Value (%)	P1	P2	P3		Value (%)	P1	P2	P3
Non-rim APHE + hypointensity on PVP	58.9 (33/56)	–	–	–		97.4 (111/114)	–	–	–	84.7 (144/170)
Non-rim APHE + hypointensity on PVP and/or TP	82.1 (46/56)	<0.001	–	–		97.4 (111/114)	>0.999	–	–	92.4 (157/170)
Non-rim APHE + hypointensity on PVP and/or TP and/or HBP	96.4 (54/56)	<0.001	0.008	–		66.7 (76/114)	<0.001	<0.001	–	76.5 (130/170)
Non-rim APHE + hypointensity on PVP and/or TP and/or HBP + non-LR-1/2/M	94.6 (53/56)	<0.001	0.016	>0.999		94.7 (108/114)	0.375	0.375	<0.001	94.7 (161/170)

Percentages were calculated according to the numbers in parentheses. P¹, P², and P³ values mean the comparison of sensitivity or specificity between criteria 1, 2, 3, and other criteria, respectively. APHE, arterial phase hyperenhancement; PVP, portal venous phase; TP, transitional phase; HBP, hepatobiliary phase. LR-1/2/M was designated in accordance to LI-RASD 2018.

Figure 3.

A moderately differentiated hepatocellular carcinoma (HCC) in a 42-year-old male patient with chronic hepatitis B. There is a well-defined nodule showing mild-moderate hyperintensity on T2-weighted imaging and hyperintensity on diffusion-weighted imaging with corresponding hypointensity on the apparent diffusion coefficient (ADC) map (A,B,C). On non-enhanced (D) and contrast-enhanced images, the lesion shows non-rim arterial phase hyper-enhancement (E) and hypointensity relative to the liver parenchyma in the portal venous phase (F), which met criterion 1 and can be correctly diagnosed as an HCC. On the subsequent transitional phase and hepatobiliary phase images (G,H), the nodular lesion shows hypointensity with “enhancing capsule” (arrow) and also satisfies the other imaging criteria.

Figure 4.

A nodule located in hepatic segment V shows hypointensity on non-enhanced T1-weighted imaging (A), arterial phase hyper-enhancement (B) and persistent enhancement in the portal venous phase (C) and transitional phase (D). However, about 90 minutes after contrast administration, this lesion shows hypointensity in the hepatobiliary phase (E). The appearance of this lesion does not meet criteria 1 and 2 but does meet criteria 3 and 4. The histologic result is a moderately differentiated hepatocellular carcinoma (magnification, 10×20) (F) in a 51-year-old male.

Figure 5.

The receiver operating characteristic (ROC) curve and area under the curve (AUC) of 4 imaging criteria: (A) for diagnosis of small hepatocellular carcinomas (HCCs); (B) for diagnosis of large HCCs.

Diagnostic performance of imaging criteria for large hepatic lesions

The result of interobserver agreement was good to excellent for all imaging features (k=0.702–0.899), especially for criterion 1 (k=0.899) of the large hepatic lesions group. A portion of large HCCs (15/70) showed hypointensity in the AP (non-APHE) in this group (Figure 6). Eight HCCs were determined as LR-M according to their imaging characteristics (Figure 7). As shown in Table 4, using criterion 1 (non-rim APHE and “washout” on PVP) for large HCC diagnosis led to the lowest sensitivity (62.9%, 44/70). All large HCCs demonstrated hypointensity in the HBP. Criterion 3 and 4, which included HBP hypointensity, presented significantly higher sensitivities (78.6% and 75.7%, respectively) than criterion 1. In comparison to criterion 3 (34.8%), criterion 4 (91.3%), including HBP hypointensity and non-LR-1/2/M, demonstrated significantly higher specificity (P<0.001). Of particular note, the specificity of criterion 4 was the same as criterion 1 (91.3%, P>0.999). Criterion 4 also showed the highest accuracy among these imaging criteria (79.6%). As shown in Figure 5, for diagnosis of large HCC, criterion 4 presented the best diagnostic performance with AUC of 0.813, followed by criteria 1–3 with AUCs of 0.771, 0.770, and 0.567, respectively. The results above were similar to those of the small lesion group.

Figure 6.

A moderately differentiated hepatocellular carcinoma (HCC) in a 66-year-old male patient with chronic hepatitis B. A mass showing moderate hyperintensity on T2-weighted imaging (A), and hypointensity in the arterial phase (C) [non-arterial phase hyper-enhancement (APHE)] is seen in the right lobe of the liver. This lesion also shows heterogeneous hypointensity relative to the liver parenchyma in pre-contrast image (B), the portal venous phase (D), transitional phase (E), and hepatobiliary phase (F) images. Criteria 1–4, which include APHE, will lead to a missed diagnosis of HCC.

Figure 7.

A histopathologically confirmed poorly differentiated hepatocellular carcinoma in a 71-year-old male patient with chronic hepatitis B. A mass in left lobe of the liver shows mild-to-moderate T2 hyperintensity (A), restricted diffusion (B,C), hypointensity on pre-contrast image (D), rim arterial phase hyperenhancement (E) (arrow), delayed central enhancement (F,G) (triangle), and hepatobiliary phase hypointensity (H). These imaging features would lead to incorrect designation as LR-M, suggesting intrahepatic cholangiocarcinoma.

Table 4. Diagnostic performance of different imaging criteria of large hepatocellular carcinoma.

Imaging criteria	Sensitivity					Specificity				Accuracy (%)
	Value (%)	P1	P2	P3		Value (%)	P1	P2	P3
Non-rim APHE + hypointensity on PVP	62.9 (44/70)	–	–	–		91.3 (21/23)	–	–	–	69.7 (65/93)
Non-rim APHE + hypointensity on PVP and/or TP	71.4 (50/70)	0.031	–	–		82.6 (19/23)	0.500	–	–	74.2 (69/93)
Non-rim APHE + hypointensity on PVP and/or TP and/or HBP	78.6 (55/70)	0.001	0.063	–		34.8 (8/23)	<0.001	0.001	–	67.7 (63/93)
Non-rim APHE + hypointensity on PVP and/or TP and/or HBP + non-LR-1/2/M	75.7 (53/70)	0.004	0.250	0.500		91.3 (21/23)	>0.999	0.500	<0.001	79.6 (74/93)

Percentages were calculated according to numbers in parentheses. P¹, P², and P³ values mean the comparison of sensitivity or specificity between criteria 1, 2, 3, and other criteria, respectively. APHE, arterial phase hyperenhancement; PVP, portal venous phase; TP, transitional phase; HBP, hepatobiliary phase. LR-1/2/M was designated in accordance to LI-RASD 2018.

Discussion

This study analyzed the diagnostic performance of different imaging criteria for small and large HCCs, with the results being similar between the small and large hepatic lesion groups. Non-rim APHE and “washout” in PVP (criterion 1) have been suggested as a non-invasive diagnostic criteria of HCC due to their high specificity (2,3,17). In this study, criterion 1 provided the highest specificity at 97.4% of small HCCs and 91.3% of large HCCs, but had the lowest sensitivity. Criteria that included HBP hypointensity (criterion 3 and 4) were shown to provide a significantly higher sensitivity than those criteria that did not. Other studies have similarly reported that HBP hypointensity exhibits high sensitivity in the diagnosis of HCC (6,8,26), though our results showed more moderately high sensitivity (78.6%) for criterion 3 (non-rim APHE plus hypointensity on PVP and/or TP and/or HBP) in the large lesion group, which might have been caused by a relatively large number of non-APHE HCCs (15/70).

In the literature, the specificities of HBP hypointensity in the diagnosis of HCC have been variable (20,27,28). Hemangioma, intrahepatic cholangiocarcinoma, and combined hepatocellular-cholangiocarcinoma, which may present HBP hypointensity, may lead to false-positive diagnoses solely on the basis of imaging criteria that include HBP hypointensity causing lower specificity. Therefore, criterion 4, which excluded observations of LR-1, LR-2, and LR-M, showed significantly higher specificity than criterion 3 and demonstrated comparable specificity to criterion 1 without significant difference. Our results were in line with recent studies (22,23).

In sum, HBP hypointensity plays a pivotal role in the diagnosis of both small and large HCC in high-risk patients. Compared with the traditional imaging criterion (AP hyperenhancement and washout on the PVP or delayed phase) (2,3), it can significantly improve the sensitivity of diagnosis with little loss in specificity for HCC. However, HBP hypointensity, which favors malignancy but not HCC in particular, is currently confined to being an ancillary feature in LI-RADS. A recent article has pointed out the pitfalls and problems of the LI-RADS that should be addressed (29), which include some ancillary features like HBP hypointensity factoring more prominently in the diagnosis of HCC. It is anticipated that HBP hypointensity may be elevated to being a major feature with the appropriate combination of complimentary features.

Diagnostic and management strategies for HCC vary geographically. In East Asia, where the incidence of HCC is the highest (1), early detection of HCC is crucial, and high sensitivity for HCC diagnosis is valued even at the cost of an “acceptable” decrease in specificity (30). The high sensitivity of HBP hypointensity should be considered for the diagnosis of HCC in these regions. The newly proposed Korea practice guidelines suggest that on multiphase MRI paired with a hepatobiliary contrast agent, washout can be considered present not only during the PVP or delayed phase, but also during the HBP (16). By comparison, in the West, where deceased donor liver transplantation is a major treatment option for HCC, high specificity for HCC diagnosis is of value to ensure appropriate organ allocation (31). For this reason, criterion 1, which provided the highest specificity but at an expense of low sensitivity, can be reasonably used in this region.

Prior studies reported that approximately 6–15% of HCCs demonstrate iso- or hyperintensity (“paradoxical uptake”) on the HBP (12,32-34). However, there were rare HCCs (1/126) that exhibited iso- or hyperintensity on HBP in the current study. This discordance may be attributed to different study cohorts and using different contrast agents. Though studies have worked on differentiation of intrahepatic cholangiocarcinoma from HCC (35,36), it was still difficult to achieve in some cases. In this study, 8 HCCs were erroneously determined as LR-M based on their imaging characteristics.

HBP was captured during gadobenate dimeglumine (Gd-BOPTA)-enhanced MRI in this study. Published reports on the use of Gd-BOPTA are much less common than those on the use of gadoxetic acid (Gd-EOB-DTPA). One reason may be that its HBP is late (1–2 hours) compared with that of Gd-EOB-DTPA, which requires a second MR scan. Another reason may be that approximately 50% of the dose of gadoxetic acid (Gd-EOB-DTPA) is taken up by the hepatocytes and excreted into the bile ducts, while about 5% of Gd-BOPTA is excreted in the biliary system, which may result in a relatively weak liver SI and biliary tree enhancement (37). In our institution, Gd-BOPTA is conventionally used and HBP is obtained by a second MR scan about 90 minutes after contrast agent administration. Although there are few publications comparing the 2 contrast agents (38,39), our study demonstrated the definitive power of HBP acquired from Gd-BOPTA in the diagnosis of HCC. According to the European Society of Gastrointestinal and Abdominal Radiology (ESGAR) consensus statement on liver-specific contrast agents (2016), there are no data indicating the diagnostic superiority of one agent over the other (40).

This study had other limitations which might have contributed to selection bias, including its retrospective and single-center nature and the possibility that patients with lesions that were missed by the initial interpreting radiologist were overlooked for this study. However, in our institution, liver MR exams were performed conventionally for high-risk patients or after detections of suspicious nodules during ultrasound surveillance; moreover, images were all interpreted by abdominal radiologists with 10 years of experience or more, which might have minimized the rate of missed diagnosis. Second, 2 different reference standards were used in this study, which might have led to misclassification bias. However, to prevent this, we used strict reference standards where all the selected malignant lesions were confirmed pathologically and partial benign lesions were determined by follow-up of a minimum of 12 months. Finally, we did not analyze non-APHE hepatic lesions that could have been seen in some large HCCs, as the optimal criterion may not suitable for all high-risk hepatic lesions in HCC diagnosis.

In conclusion, after excluding nodules considered to be benignities or non-HCC malignancies according to characteristic imaging features, HBP hypointensity captured during Gd-BOPTA-MRI can improve sensitivity while maintaining high specificity in the diagnosis of both small and large HCCs for high-risk patients.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This single-center retrospective cohort study was approved by the Institutional Review Board (No. [2016]118 and No. [2019]283), who determined the requirement for informed consent could be waived.