Introduction

Unlike other cancers, hepatocellular carcinomas (HCCs) can be diagnosed radiologically without histological confirmation. Several international scientific organisations or societies, including the European Association for the Study of the Liver (EASL), the American Association for the Study of Liver Diseases (AASLD), the American College of Radiology (ACR) and the Organ Procurement and Transplantation Network/United Network for Organ Sharing (OPTN/UNOS), have proposed different criteria for the imaging diagnosis of HCC, in which tumour size, arterial phase hyperenhancement, washout appearance and capsule appearance are major imaging features [1,2,3,4].

In the imaging diagnosis of hepatic malignancy, the accuracy of magnetic resonance imaging (MRI) for HCC is known to be superior to that of computed tomography (CT), and liver MRI is widely used for pre-treatment evaluation of patients with HCC [5,6,7,8,9,10]. Currently, two types of contrast agents are used for liver MRI: extracellular contrast agent (ECA) and hepatocyte-specific or hepatobiliary contrast agent (HBA). While ECAs such as gadoterate meglumine (Gd-DOTA) are distributed only in the intravascular and extracellular spaces before undergoing renal excretion, HBAs such as gadoxetic acid disodium (Gd-EOB-DTPA) are initially distributed in the extracellular space and then taken up by hepatocytes. The maximum hepatocyte uptake of gadoxetic acid, called the hepatobiliary phase (HBP), occurs 15–20 min after injection. The high contrast in signal intensity between hypointense focal lesions and the hyperintense surrounding liver in the HBP increases the diagnostic sensitivity of HBA-enhanced MRI [11,12,13,14].

In our institution, liver MRI is performed using ECA or HBA at the physician’s discretion. The same protocol is used regardless of the type of contrast agent, and thus 15-min delayed-phase images are obtained for both ECA- and HBA-enhanced MRI. Additional acquisition of the 15-min delayed phase does not increase the scan time beyond approximately 20 s required for scanning one dynamic sequence, because we acquire T2-weighted and diffusion-weighted images between the late dynamic and the 15-min delayed phases.

In practice, we have observed that some hepatic lesions showing washout or capsule appearance only in the 15-min delayed phase of ECA-enhanced MRI were later confirmed as HCC. From these observations, we hypothesised that, if the imaging features observed only in the 15-min delayed phase are exclusively seen in HCC, including the 15-min delayed phase in the ECA-enhanced MRI protocol could help diagnose additional HCCs without making a false-positive diagnosis.

The purpose of this study was to examine the added value of the 15-min delayed phase of ECA-enhanced MRI for diagnosis of HCC in patients with chronic liver disease.

Materials and methods

Subjects

The institutional review board approved this retrospective study and waived the requirement for informed consent. Our database contained 2,188 patients with chronic liver disease who underwent liver MRI from January 2014 to December 2015. We excluded 2,008 patients who were previously treated for hepatic (n = 1,317) or other (n = 67) malignancies, who had no focal lesion other than definite benign lesions such as cysts or arterioportal shunts (n = 148), who underwent MRI using HBA (n = 339) or who were diagnosed with locally advanced or disseminated cancers (n = 141). A radiologist (H.S.H., with 3 years of experience in liver MRI) reviewed the MRI data of the remaining 176 patients and selected 304 suspicious focal hepatic lesions based on the prospectively written radiology reports. If a patient had more than five suspicious lesions, the five largest lesions with arterial phase hyperenhancement were selected. Among these 304 lesions, 171 observations were excluded from the analysis because their final diagnoses remained indeterminate (treated without pathological diagnosis [n = 77] or inconclusive follow-up [n = 94]; see ‘Reference standards’ section below). Our final study included 133 focal hepatic lesions from 103 patients (Fig. 1).

Fig. 1
figure 1

Flowchart illustrating subject selection. HBA hepatobiliary contrast agent, ECA extracellular contrast agent, HCC hepatocellular carcinoma

Full size image

Reference standards

Patients’ medical records and images were reviewed by a single investigator (H.S.H.) who determined the final diagnoses of the hepatic lesions. Histopathological diagnosis confirmed by surgery or biopsy was used as a reference standard. In cases for which histological diagnosis was not available, a hepatic lesion was diagnosed as HCC if it showed local or marginal recurrence after transarterial chemoembolisation or radiofrequency ablation, if it presented threshold growth (≥ 50 % diameter increase within 6 months or ≥ 100 % diameter increase over 6 months) [15, 16], or if it revealed new arterial phase hyperenhancement or washout appearance at follow-up. If a hepatic lesion disappeared or showed no interval change during a follow-up period of ≥ 18 months, it was considered benign [17].

MRI

MRI was conducted using three 3.0-T systems (MAGNETOM Trio, A Tim System, Siemens Medical Solutions, Erlangen, Germany; Intera Achieva, Koninklijke Philips N.V., Best, The Netherlands; Ingenia, Koninklijke Philips N.V.). All images were obtained in the transverse plane with a field of view of 44 × 33 cm or 40 × 30 cm depending on the patient’s body size. After localiser images were taken, two-dimensional, dual-echo T1-weighted gradient-recalled echo images were acquired, with a slice thickness, intersection gap and repetition time (TR) of 7 mm, 0.7 mm and 150–192 ms, respectively. The echo times (TEs) for in-phase and opposed-phase images were 2.3–2.5 ms and 1.1–1.2 ms, respectively.

Pre- and post-contrast dynamic images were acquired using a three-dimensional gradient echo sequence with a section thickness, TR and TE of 2–4 mm, 2.5–4.5 ms and 0.9–2.2 ms, respectively. For dynamic imaging, an extracellular gadolinium-based contrast agent (Dotarem®, gadoterate meglumine, Guerbet, Aulnay-sous-Bois, France; Magnevist®, gadopentetate dimeglumin, Bayer Schering Pharma AG, Berlin, Germany) was injected at a dose of 0.2 ml/kg of body weight, followed by 20 ml of 0.9 % saline at an injection rate of 2 ml/s. The arterial phase began 3–5 s after peak contrast-enhancement of the abdominal aorta. The time-to-peak aorta enhancement was determined using a test bolus technique with 1 ml contrast agent, or by obtaining bolus-tracking images. Images of three subsequent dynamic phases were acquired after arterial phase imaging: early portal phase, late portal phase and late dynamic phase with a breath-hold; the scan time of each phase was 18–24 s. The start times for scanning the arterial, early portal, late portal and late dynamic phases were 25–30 s, 65–75 s, 95–105 s and 135–150 s after contrast injection, respectively. In addition, the far delayed-phase images were acquired at a fixed time delay of 15 min after contrast injection using the same imaging sequence as dynamic T1-weighted imaging.

During the interval between the late dynamic and 15-min delayed-phase, T2-weighted images were acquired by multi-shot and single-shot turbo spin echo sequences using a navigator-triggered technique, with a section thickness, gap, TR and TE of 5–7 mm, 1 mm, 1,589–3,250 ms and 70–96 ms, respectively. Diffusion-weighted images were acquired using a navigator-triggered technique at b-values of 50, 400 and 800 s/mm2.

Image analysis

All MRI images were retrospectively reviewed on a picture archiving and communication system (PACS; Centricity, Version 2.0, GE Healthcare, Barrington, IL, USA). A radiologist (H.S.H.) reviewed the pathology reports, surgical records and MRI data, and then recorded the location (hepatic segment), size and corresponding image number of each hepatic lesion. With this information, two board-certified abdominal radiologists (C.A. and M.J.K., with 6 and 23 years of experience in liver MRI, respectively) unaware of patients’ clinical history or final diagnosis independently analysed the MRI images. For each hepatic lesion, they determined the presence or absence of arterial phase hyperenhancement, washout appearance and capsule appearance as defined by the Liver Imaging Reporting and Data System (LI-RADS) [3]. They also recorded imaging features observed in the 15-min delayed phase, including washout and capsule appearances. In addition, without referring to the 15-min delayed-phase images, they categorised each hepatic lesion according to the LI-RADS diagnostic algorithm proposed by the ACR [3]: LR-1 (definitely benign), LR-2 (probably benign), LR-3 (indeterminate), LR-4 (probable HCC), LR-5 (definite HCC), LR-5V (definite tumour in vein) or LR-M (probably malignant, but not specific for HCC).

Based on the results of the image analysis, an investigator who did not participate in image analysis (S.E.L.) categorised each hepatic lesion as HCC or non-HCC according to the EASL criteria, in which HCC is diagnosed if a hepatic lesion > 1 cm displays both arterial phase hyperenhancement and washout appearance [2], with and without using the imaging features found in the 15-min delayed phase. She also examined whether the hepatic lesions that had been categorised as LR-3 or LR-4 by the reviewers could be upgraded to LR-5 by considering the imaging features observed in the 15-min delayed phase. For imaging diagnosis using the LI-RADS, lesions categorised as LR-5 or LR-5V were considered positive for HCC.

Statistical analysis

The sensitivity, specificity and accuracy of the diagnostic criteria were calculated and compared using generalised estimating equations. Interobserver agreement was presented by Cohen’s kappa coefficient. A kappa statistic of 0.8–1.0 was considered excellent agreement, 0.6–0.79 was good agreement, 0.40–0.59 was moderate agreement, 0.2–0.39 was fair agreement and 0–0.19 was poor agreement. Two-sided P-values < 0.05 were considered statistically significant. All statistical analyses were performed using SAS 9.2 (SAS Institute, Inc., Cary, NC, USA).

Results

Baseline characteristics

Our study population of 103 patients consisted of 77 men and 26 women; the median age was 61 years (range, 42–81 years). Hepatitis B virus was the most common cause of chronic liver disease (81.6 %), and most patients (88.3 %) were in Child-Pugh class A (Table 1). Of 133 hepatic observations in 103 patients, 107 (80.5 %) were HCCs, three (2.3 %) were other malignancies and 23 (17.3 %) were benign. Of the 107 HCCs, 85 (79.4 %) were pathologically diagnosed after hepatic resection (n = 66), liver transplantation (n = 18) or percutaneous biopsy (n = 1). The remaining 22 HCCs (20.6 %) were diagnosed according to the clinical criteria; 18 presented marginal recurrence at immediate follow-up after transarterial chemoembolisation or radiofrequency ablation, and four showed threshold growth with the radiological hallmark of HCC (i.e. arterial phase hyperenhancement and washout appearance) at follow-up. The three other malignancies were pathologically diagnosed as intrahepatic cholangiocarcinoma, combined HCC-cholangiocarcinoma and combined HCC-neuroendocrine carcinoma after hepatic resection. Of the 23 benign lesions, two (8.7 %) were pathologically confirmed to be regenerative and high-grade dysplastic nodules.

Table 1 Characteristics of 103 patients with 133 hepatic lesions
Full size table

Inter-reader agreement of diagnosis was good regardless of the diagnostic criteria; kappa coefficients were 0.750 and 0.748 for the EASL and LI-RADS criteria, respectively.

Imaging features observed only in the 15-min delayed phase

Reviewer 1 found washout appearance in three lesions and capsule appearance in three other lesions only in the 15-min delayed phase (Figs. 2 and 3). Reviewer 2 found washout appearance in one lesion, capsule appearance in two lesions, and both capsule and washout appearances in three lesions only in the 15-min delayed phase. All lesions that showed additional washout or capsule appearance in the 15-min delayed phase were confirmed as HCC. For one intrahepatic cholangiocarcinoma, both reviewers concluded that delayed phase central enhancement and peripheral washout appearance, imaging features favouring other hepatic malignancy over HCC, were unequivocally present only in the 15-min delayed phase (Fig. 4). None of the benign lesions or other malignancies showed additional washout or capsule appearance in the 15-min delayed phase. The frequencies of washout and capsule appearances observed in hepatic lesions with and without 15-min delayed-phase images are presented in Table 2.

Fig. 2
figure 2

A hepatocellular carcinoma in a 59-year-old man with B-viral liver cirrhosis; our reviewers concluded that capsule appearance is unequivocally observed only in the 15-min delayed phase of extracellular contrast agent-enhanced MRI. (a) A 2.4-cm mass in the right posterior liver shows arterial phase hyperenhancement. (b) No capsule is seen in the portal or late dynamic phases. (c) The mass displays smooth, ring-like enhancement (arrow) in the 15-minute delayed phase. (d) Microscopic examination after resection shows a fibrous capsule (FC) between the tumour (T) and the normal parenchyma (P) (haematoxylin-eosin stain; original magnification, ×20)

Full size image
Fig. 3
figure 3

A hepatocellular carcinoma in a 69-year-old women with B-viral liver cirrhosis; our reviewers concluded that capsule and washout appearances are present only in the 15-min delayed phase of extracellular contrast agent-enhanced MRI. (a) A 3.3-cm mass in the hepatic dome shows diffuse arterial phase hyperenhancement. (b and c) No capsule appearance is seen in the portal (b) or late dynamic (c) phases. (d) Capsule appearance (arrows) is present in the 15-min delayed phase, and one reviewer also concluded that washout appearance is present in this phase

Full size image
Fig. 4
figure 4

An intrahepatic cholangiocarcinoma in a 67-year-old man with B-viral liver cirrhosis; imaging features are prominently visible in the 15-min delayed phase of extracellular contrast agent-enhanced MRI. (a) A 4.5-cm mass with irregular rim-like arterial phase hyperenhancement is seen in the right posterior liver. (b and c) No specific imaging findings other than subtle progressive contrast enhancement are noted in the portal (b) and late dynamic (c) phases. (d) In the 15-min delayed-phase, peripheral washout appearance (white arrows) and delayed-phase central enhancement (black arrow), features that favour other hepatic malignancy over hepatocellular carcinoma, are visible

Full size image
Table 2 Frequencies of washout and capsule appearances on extracellular contrast agent-enhanced MRI with and without the 15-min delayed phase
Full size table

Impact of the 15-min delayed phase on diagnostic performance

Using the EASL criteria, washout appearance observed only in the 15-min delayed phase resulted in three and four additional true-positive HCC diagnoses for Reviewers 1 and 2, respectively (Table 3). Consequently, the diagnostic sensitivity and accuracy improved by 2.8 % (P = 0.072) and 2.2 % (P = 0.080) for Reviewer 1, and by 3.7 % (P = 0.041) and 3.6 % (P = 0.042) for Reviewer 2, respectively. Using the LI-RADS criteria, one of six HCCs whose capsule appearances were only seen in the 15-min delayed phase was upgraded from LR-4 to LR-5 for Reviewer 1, minimally increasing the diagnostic sensitivity (P = 0.316) and accuracy (P = 0.315); three HCCs were upgraded from LR-3 to LR-4. For Reviewer 2, washout or capsule appearance seen only in the 15-min delayed phase upgraded four HCCs from LR-4 to LR-5, increasing the sensitivity and accuracy by 3.7 % (P = 0.043) and 3.1 % (P = 0.042), respectively; one HCC was upgraded from LR-3 to LR-4 (Table 3).

Table 3 Diagnostic performance of extracellular contrast agent-enhanced MRI with and without the 15-min delayed phase according to EASL and LI-RADS criteria
Full size table

There was no additional false-positive case due to imaging features observed in the 15-min delayed phase, irrespective of the diagnostic criteria or the reviewer. Thus, the diagnosis of benign or other malignant lesions was not affected, and the diagnostic specificity did not change due to the additional use of the 15-min delayed phase (Table 3).

Discussion

In the present study, approximately 5 % of HCCs manifested washout or capsule appearance only in the 15-min delayed phase; some resulted in additional true-positive diagnoses or upgrades in the LI-RADS category, without additional false-positive diagnoses. We believe this diagnostic benefit is sufficient to compensate for the additional scan time of 18–24 s needed for acquisition of 15-min delayed-phase images in ECA-enhanced MRI.

Although the underlying mechanism of washout appearance in HCC is not yet fully understood, it may be partly explained by the combination of various effects, including venous drainage of contrast media, background liver enhancement due to the retention of contrast media within fibrotic parenchyma, tumoral hypercellularity with corresponding reduction in extracellular volume, and intrinsic hypoattenuation or hypointensity [18, 19]. Capsule appearance, which is defined as a peripheral rim of smooth hyperenhancement in the portal venous or delayed phase of dynamic contrast-enhanced imaging, reflects the retention of contrast media within peritumoral fibrosis and prominent sinusoids surrounding a tumour [20]. Progressive concentric enhancement and delayed central enhancement, which are features favouring other malignancy such as cholangiocarcinoma over HCC, are due to the retention of contrast media in central stromal fibrosis [21].

From the mechanisms explained above, it can be inferred that a certain amount of time is required after contrast injection for these imaging features to be unequivocally exhibited. Although the time required may vary depending on the characteristics of each lesion and the surrounding liver parenchyma, the imaging features likely become more apparent as time passes after contrast injection. In the early years after the introduction of multi-detector CT for HCC diagnosis, several groups reported that some HCCs showed a washout appearance only in the 3- or 5-min delayed phase of ECA-enhanced CT, and that the diagnostic sensitivity and accuracy could be improved by adding delayed-phase imaging to the biphasic CT [22,23,24,25,26]. Afterwards, some investigators demonstrated that imaging features observed in the 10-min or longer delayed phase might be helpful in distinguishing HCC from other hepatic lesions [27, 28]. Other investigators reported that imaging features of intrahepatic cholangiocarcinoma could be better detected 10 min after contrast injection [29, 30], like the case we have experienced in this study.

Previously, it was not practical to include a long delayed phase in imaging protocols, because it would mean waiting until the delayed phase, doing nothing. Currently, however, liver MRI routinely includes T2-weighted and diffusion-weighted imaging, which can be acquired between the late dynamic (i.e. 3-min delayed) phase and 15-min delayed phase (or hepatobiliary phase for HBA-enhanced MRI) without compromising the diagnostic performance [14, 31,32,33]. Therefore, in ECA-enhanced MRI, we believe that the benefits of acquiring 15-min delayed images outweigh its costs.

A major limitation of this study was its retrospective nature. We excluded 171 observations due to the lack of pathological diagnosis or inconclusive follow-up during subject selection. Furthermore, some HCC and benign lesions were diagnosed using the non-pathological criteria, because the current guidelines permit non-invasive imaging diagnosis of HCC in at-risk patients. In addition, the reviewers were aware they were investigating the possible effects of a 15-min delayed phase on diagnostic accuracy. All these potential biases arising from the retrospective nature of this study might have led to the overestimation of diagnostic accuracy of ECA-enhanced MRI applying the 15-min delayed phase.

In conclusion, the additional use of 15-min delayed-phase images may enable diagnosis of more HCCs using ECA-enhanced MRI, as some HCCs display typical imaging features later than in the conventional delay time of 3 min after contrast injection.