ABSTRACT
-
Background/Aims
Hepatocellular carcinoma (HCC) frequently recurs after curative treatment, posing challenges to long-term survival. Although contrast-enhanced multiphasic computed tomography (CECT) is commonly used for detecting recurrence, it is associated with risks such as radiation exposure and contrast agent reactions. This study aimed to compare the diagnostic performance of non-contrast magnetic resonance imaging (NC-MRI) with CECT for detecting recurrent HCC.
-
Methods
In this prospective multicenter intra-individual head-to-head comparison trial (study identifier: NCT05690451, KCT0006395), participants who had undergone curative treatment for HCC and remained recurrence-free for over two years were enrolled. Each participant underwent three follow-up imaging sessions at 2–6-month intervals using both CECT and NC-MRI. The primary outcome was the detection accuracy of each modality, analyzed using the generalized estimating equation analysis. Secondary outcomes included sensitivity and specificity.
-
Results
The study included 203 participants with a total of 528 paired imaging sessions, identifying recurrent HCC in 22 cases (10.8%). Among these, 21 cases involved intrahepatic recurrence with a median tumor size of 1.3 cm, and one case had aortocaval lymph node metastasis. NC-MRI achieved a detection accuracy of 96.6% (196/203), higher than CECT’s 91.6% (186/203) (P=0.006). NC-MRI also showed greater sensitivity (77.3% [17/22] vs. 36.4% [8/22]; P=0.012), while specificity was comparable between NC-MRI and CECT (98.9% [179/181] vs. 98.3% [178/181]; P=0.999).
-
Conclusions
NC-MRI demonstrated higher sensitivity and accuracy compared to CECT in detecting recurrent HCC in patients who had been disease-free for over two years following curative treatment, indicating its potential as a preferred imaging modality for this purpose.
-
Keywords: Hepatocellular carcinoma; Recurrence; Contrast enhanced CT; Non-contrast abbreviated MRI
Study Highlights
• NC-MRI provided higher accuracy in detecting recurrent HCC than CT (96.6% vs. 91.6%; P=0.006).
• The sensitivity in detecting recurrent HCC of NC-MRI is higher than CECT (77.3% vs. 36.4%; P=0.012).
• NC-MRI offers several merits over CECT including no radiation exposure and no contrast agent-related risk.
• NC-MRI might emerge as the preferred detection modality for detecting recurrence of HCC following curative treatment.
Graphical Abstract
INTRODUCTION
Hepatocellular carcinoma (HCC) is the most prevalent primary malignant liver cancer and ranks sixth among cancer-related causes of death worldwide [
1,
2]. International guidelines recommend biannual ultrasound surveillance in high-risk populations to detect HCC early, enabling timely curative treatment such as surgical resection, liver transplantation or local ablation, which significantly improves overall survival [
3-
6]. Despite these efforts, long-term survival rates following curative treatments for HCC remain unsatisfactory due to high recurrence rates [
7,
8]. Recurrences of HCC are categorized as early (within 2 years) and late (more than 2 years) based on their timing post-treatment [
7,
9]. Early recurrences typically stem from occult metastases of the initial HCC, while late recurrences often originate from
de novo secondary primary HCC in the remnant liver after treatment [
10,
11]. Even after prolonged periods of recurrence-free status, there remains a considerable risk of developing
de novo secondary HCC [
9,
12]. Therefore, current HCC management guidelines emphasize continuous follow-up imaging for recurrence detection after curative treatment [
3-
6,
13].
Although continuous imaging follow-up is essential for managing HCC, current guidelines do not clearly define the optimal imaging modality, timing, or duration for detecting recurrence after curative treatment [
3,
14,
15]. These guidelines generally recommend contrast-enhanced computed tomography (CECT) and/or magnetic resonance imaging (MRI) for recurrence detection, with intervals ranging from 3 to 6 months [
3,
6,
14,
15]. However, these methods have several important limitations. CT involves cumulative radiation exposure, while MRI is associated with high costs and limited availability. Additionally, repeated use of contrast agents carries risks such as nephrotoxicity and allergic reactions. Non-contrast abbreviated MRI (NC-MRI) has emerged as a potential alternative, utilizing selected MRI sequences that are essential for HCC detection without contrast injection to expedite scanning. Park et al. [
16] demonstrated NCMRI’s superior detection of HCC compared to ultrasound in high-risk populations undergoing primary surveillance. Recent prospective studies have also shown that NC-MRI offers significantly improved detection of HCC in similar settings compared to ultrasound [
17,
18]. Moreover, Decharatanachart et al. [
19] demonstrated that NC-MRI is a cost-effective strategy for primary HCC surveillance compared to ultrasound combined with alpha-fetoprotein (AFP), particularly in high-risk individuals. In addition to its theoretical advantages over CECT or MRI such as the absence of radiation exposure and contrast-related risks, these findings support NC-MRI as a promising alternative modality for primary HCC surveillance. Furthermore, a recent retrospective study has suggested that NC-MRI offers detection capabilities comparable to full-sequence CE liver MRI for monitoring late recurrent HCC [
9]. However, prospective evaluation of NC-MRI’s effectiveness in detecting late recurrent HCC after curative treatments, compared to the standard CECT, remains limited. This multicenter prospective study aims to determine whether NC-MRI can achieve comparable sensitivity and accuracy to CE multiphasic CT for detecting late recurrent HCC following curative treatments, such as surgical resection or local ablation.
MATERIALS AND METHODS
This prospective multicenter intra-individual head-to-head comparison non-inferiority trial was conducted across three university-affiliated tertiary referral hospitals in the Republic of Korea: Seoul National University Hospital, Asan Medical Center, and Seoul National University Bundang Hospital. Each center independently evaluated and treated their respective participants. Institutional Review Boards approved the study (Approval number: H-2106-183-1230), and written informed consent was obtained from all participants. The study was registered on both clinicaltrials. gov (study identifier: NCT05690451) and cris.nih.go.kr (study identifier: KCT0006395).
Participant enrollment
The study protocol is detailed in the
Supplementary Material. Initially, the study was designed as a non-inferiority trial based on the observed similarities in the reported accuracy and sensitivity of NC-MRI and CECT for detecting HCC [
20,
21]. To estimate sample size, following assumptions are made using results from previous studies: 1) the accuracy of CECT to detect HCC was 94.4% [
20]; 2) the accuracy of NC-MRI to detect HCC was 96.0% [
16]; 3) non-inferiority margin was set at 0.05 (e.g., 5%); 4) the power was set at 0.8; and 5) the incidence of recurrent HCC for study population was 15%. Using aforementioned assumptions, the estimated sample size was 163. Accounting for a dropout rate of 20%, as observed in previous prospective studies at these centers, the final estimated sample size was adjusted to 204. Consequently, we aimed to enroll 210 participants who had undergone curative treatment for HCC via surgical resection or local ablation and were scheduled for CECT follow-up to detect recurrence. Inclusion criteria for the study were: 1) age between 20 and 85 years; 2) prior curative treatment including margin negative surgical resection and complete local ablation for HCC; 3) absence of recurrence for more than two years post-treatment; and 4) scheduled CECT for detection of late recurrence. Exclusion criteria included: 1) claustrophobia; 2) presence of a cochlear implant and/or cardiac pacemaker; 3) history of severe allergic reaction to iodine contrast; and 4) estimated glomerular filtration rate < 60 mL/min/1.73m
2.
All enrolled participants underwent three follow-up imaging sessions, which included paired CECT and NC-MRI, scheduled at intervals of 2–6 months. The timing of these intervals was determined by the attending physicians based on each patient’s assessed risk of HCC recurrence and prior imaging findings. The first session of examinations took place 2–6 months after the most recent liver imaging examination conducted before study enrollment. Whenever feasible, both CECT and NC-MRI were conducted on the same day or within a 14-day window of each other. For the acquisition of CECT, a pre-contrast scan was initially performed, covering the area from the lower thorax to the right kidney. Following the pre-contrast scan, iodinated contrast media at a dose of 600 mg/kg was administered through an 18G intravenous catheter placed in the antecubital vein using a power injector. After the contrast injection, scans were conducted in the arterial, portal venous, and delayed phases. Detailed protocols for CECT are provided in the
Supplementary Material.
The NC-MRI protocol included the following sequences: T2-weighted fast spin-echo (FSE) sequence, heavily T2-weighted single-shot FSE or half-Fourier acquisition single-shot turbo spin-echo sequence, dual-echo T1-weighted sequence, fat-suppressed T1-weighted sequence using the Dixon technique, and diffusion-weighted imaging (DWI) with a b-value of 500–1,000 sec/mm² along with an apparent diffusion coefficient map.
Imaging evaluation and recall policy
All CECT and NC-MRI images were independently evaluated at each center based on on-site analysis by board-certified radiologists with over 10 years of experience in liver imaging (S.K.J and D.H.L at Seoul National University Hospital; D.W.K and S.Y.K at Asan Medical Center; W.C and J.C at Seoul National University Bundang Hospital). To minimize recall and learning bias, each imaging modality’s results were kept blinded from the other throughout the study period. Radiologists were blinded to the results of the other modality of the same and previous session of examination during the study period.
For CECT, a liver nodule measuring 1 cm or larger was diagnosed as recurrent HCC based on non-invasive imaging criteria if it exhibited arterial phase non-rim hyper-enhancement followed by washout in the portal venous and/or delayed phase [
3,
5,
6,
22]. Nodules of similar size that showed arterial phase non-rim hyper-enhancement without washout in the portal venous and/or delayed phase were classified as suspicious lesions. Similarly, nodules measuring 1 cm or larger that demonstrated washout in the portal venous and/or delayed phase without arterial phase hyperenhancement were also categorized as suspicious. However, lesions with arterial phase hyper-enhancement but ill-defined margins or wedge-shaped appearances without washout were considered likely benign, potentially representing arterioportal (AP) shunts or perfusion-related changes [
23,
24]. For any recurrent HCCs or suspicious lesions detected on CECT, prior imaging studies taken before study enrollment following the initial HCC treatment were reviewed to determine whether the lesions were newly developed. Lesions that remained stable for over two years, as assessed by prior imaging studies, were classified as benign and not indicative of recurrent HCC.
For NC-MRI, which lacks the CE dynamic phases necessary for the non-invasive imaging diagnosis of HCC, only suspicious lesions could be identified. Liver nodules 1 cm or larger displaying mild to moderate hyper-intensity on T2 weighted image or diffusion restriction on DWI were considered indicative of recurrent HCC [
9,
16,
25]. If positive findings were observed on NC-MRI, previous imaging studies taken before study enrollment following the initial HCC treatment were reviewed to evaluate whether the suspicious lesions were newly developed. Lesions that remained stable for more than two years, as determined by previous imaging studies, were deemed benign and not suggestive of recurrent HCC.
To confirm recurrent HCC in suspicious lesions identified on either CECT or NC-MRI that were newly developed compared to previous imaging studies, additional assessments, such as gadoxetic acid-enhanced liver MRI or biopsy, were performed. Participants without suspected lesions for recurrent HCC during the three follow-up imaging sessions underwent an additional follow-up imaging using CECT or MRI 3–6 months after the third session to rule out potential false-negative results from the previous follow-up imaging examinations.
Reference standard for the diagnosis of recurrent HCC
Both histopathology and non-invasive imaging were employed for diagnosing recurrent HCC in this study. Histopathological confirmation was utilized when surgical resection or biopsy was performed on suspicious recurrent lesions. Non-invasive imaging diagnosis of recurrent HCC relied on the KLCA-NCC practice guidelines for the management of HCC [
5,
22].
This study was designed as a non-inferiority trial, with the non-inferiority margin for the difference in per-patient accuracy in detecting recurrent HCC between CECT and NC-MRI set at 5%. All continuous variables were reported as median with interquartile range (IQR) and compared using the Mann–Whitney U-test or Kruskal–Wallis test. Categorical variables were presented as numbers with percentages and compared using the chi-square test or Fisher’s exact test. The primary outcome of this study was the per-patient accuracy in detecting recurrent HCC for each imaging modality. Secondary outcomes included sensitivity, specificity, positive predictive value, negative predictive value, diagnostic yield and false referral rate for each modality. Per-lesion sensitivity of each modality was also assessed. The diagnostic performance was compared using the generalized estimating equation analysis considering the subject correlation, with P-values <0.05 considered statistically significant. All statistical analyses were performed using SPSS version 27.0 (IBM Co., Armonk, NY, USA).
RESULTS
Study population and development of recurrent HCC
Participant enrollment commenced on December 28, 2021, and the last follow-up examination was completed on June 20, 2024. Among the 210 participants meeting the study’s inclusion criteria and providing informed consent, 203 underwent at least one session of imaging follow-up with both CECT and NC-MRI, constituting the study cohort (
Fig. 1). All enrolled participants had remained free of HCC recurrence for more than two years following their initial treatment, and the median interval between curative treatment and study enrollment was 48.0 months (IQR 38.0–71.0 months). Baseline characteristics and details of previous HCC treatments are summarized in
Table 1. Among 203 enrolled participants, 145 (71.4%) participants had undergone surgical resection for HCC, and 104 (51.2%) had liver cirrhosis. A total of 528 paired CECT and NC-MRI imaging sessions were performed. Most follow-up imaging sessions occurred on the same day, with only six (1.1%) occurring within an interval of 2 to 10 days. A substantial proportion of participants underwent follow-up at 6-month intervals, and there was no significant difference in follow-up intervals between patients who developed recurrent HCC during the study period and those who did not.
During the study period, 23 recurrent HCCs were identified in 22 participants. Among these, one metastatic lymph node was confirmed through histopathological examination, while the remaining 22 recurrent intrahepatic HCCs were diagnosed based on non-invasive imaging criteria: 8 were diagnosed using CECT, and the remaining 14 recurrent HCCs were diagnosed using gadoxetic acid-enhanced liver MRI.
Out of the 528 follow-up imaging sessions, 20 recurrent HCCs were identified in 19 participants. Among these, 18 participants had 19 intrahepatic recurrences, with 17 participants presenting a single recurrent HCC and one participant presenting two recurrent HCCs. All of these intrahepatic recurrences were diagnosed through non-invasive imaging criteria. One participant was diagnosed with aortocaval lymph node metastasis via histopathological examination following a biopsy. Additionally, three participants were diagnosed with intrahepatic recurrent HCCs (one HCC each) on follow-up imaging studies conducted six months after negative findings on the final session of CECT and NC-MRI. These diagnoses were made using follow-up CECT (n=2) and gadoxetic acid-enhanced liver MRI (n=1).
None of these 23 recurrent HCCs were detected in prior imaging studies taken before study enrollment. Vascular invasion was not observed in any recurrent HCC cases. The median size of the 22 intrahepatic recurrent HCCs was 1.3 cm (range 1.0–1.7 cm) and the size of the aortocaval lymph node metastasis in one participant was 3.3 cm. There were no significant differences in baseline characteristics or the stage of previously treated HCC between participants who developed recurrent HCC and those who did not, except for age; the median age of participants with recurrent HCC was significantly higher (
Table 1).
The accuracy, sensitivity, and specificity of CECT and NC-MRI in detecting recurrent HCC on a per-patient basis are summarized in
Table 2. NC-MRI demonstrated a per-patient accuracy of 96.6% (196/203), compared to 91.6% (186/203) for CECT. The lower limit of the 95% confidence interval for the difference in accuracy between NC-MRI and CECT was 1.7%, indicating that NC-MRI is not inferior to CECT, with a statistically significant advantage favoring NC-MRI (
P=0.006). When comparing the accuracy between two imaging modalities according to the presence of cirrhosis, NC-MRI provided significantly higher accuracy in detecting late recurrent HCC than CECT for 104 participants with cirrhosis (96.2% [100/104] vs. 90.4% [94/104];
P=0.031). NC-MRI also provided marginally higher accuracy than CECT for 99 participants without cirrhosis (97.0% [96/99] vs. 92.9% [92/99];
P=0.126). NC-MRI also achieved a higher per-patient sensitivity of 77.3% (17/22) compared to 36.4% (8/22) for CECT (
P=0.012). All 17 recurrent HCCs detected by NC-MRI exhibited moderately high signal intensity on T2-weighted images and diffusion restriction on DWI (
Fig. 2). Recurrent HCCs were detected exclusively by NC-MRI in 10 participants, exclusively by CECT in one participant, and by both imaging modalities in seven participants. The per-patient specificity for detecting recurrent HCC was similar between CECT (98.3%, 178/181) and NC-MRI (98.9%, 179/181) (
P=0.999). On a per-lesion basis, NC-MRI exhibited a sensitivity of 78.3% (18/23), significantly outperforming CECT, which had a sensitivity of 34.8% (8/23) (
P=0.011).
Among the 22 participants with recurrent HCC, three had elevated levels of Protein Induced by Vitamin K Absence or antagonist-II (PIVKA-II), and one had elevated levels of both AFP and PIVKA-II. The remaining 18 participants had normal levels of these tumor markers. Of the three participants with elevated PIVKA-II, one recurrent HCC was detected only on NC-MRI, one was identified by both CECT and NC-MRI, and one was not detected by either modality. The participant with elevated levels of both tumor markers had an aortocaval lymph node metastasis that was identified only on CECT.
For false-negative diagnoses, CECT failed to detect recurrent HCC in 14 participants. The reasons included misinterpretation as AP shunt (n=8) (
Fig. 2), missed detection of small HCCs (1.1 cm in size) adjacent to portal vein branches (n=2), and missed small lesion (1.2 cm) near a previous ablation zone treated with Radiofrequency ablation (n=1). Additionally, in three participants who were diagnosed with intrahepatic recurrent HCC during follow-up imaging after completing three sessions of paired imaging tests, sub-threshold arterial phase non-rim enhancing lesions without washout (measuring 5 mm, 7 mm, and 7 mm) were retrospectively identified on the final session of CECT.
Non-contrast MRI (NC-MRI) missed recurrent HCC in five participants. These included one case of aortocaval lymph node metastasis, identified retrospectively during image review (
Fig. 3), and one lesion located near the liver segment VI tip, also identified retrospectively. Three other missed lesions were later detected on follow-up imaging after completing three sessions of paired imaging tests. Retrospective review of the final NC-MRI session revealed that two of these lesions were visible but measured subthreshold sizes of 5 mm and 7 mm.
False-positive diagnoses were reported in three cases using CECT. These included two lesions identified as AP shunts and one pseudo-lesion that did not correspond to any identifiable lesion on subsequent follow-up CT or MRI. NC-MRI resulted in two false-positive diagnoses, both pseudo-lesions that were not corroborated by follow-up CT or MRI findings.
Treatment of participants with recurrent HCC
Among the 21 participants who developed intrahepatic recurrent HCCs, 12 were treated with local ablation therapy. Three participants received stereotactic body radiation therapy, while the remaining six were managed with transarterial chemoembolization. Additionally, one participant with aortocaval lymph node metastasis was treated with systemic chemotherapy using atezolizumab combined with bevacizumab.
DISCUSSION
This study evaluated and compared the performance of NC-MRI with CECT in detecting recurrent HCC among participants undergoing imaging follow-up after a recurrence-free interval of more than two years after curative treatment, including surgical resection and local ablation. Based on previously reported comparable sensitivity (79.1% for NC-MRI vs. 83.3% for CECT) and accuracy (97.9% for NC-MRI vs. 95.6% for CECT) in detecting HCC [
16,
20], the study was initially designed as a non-inferiority trial to establish the equivalence of NC-MRI to CECT. However, the findings of this study demonstrated that NC-MRI had significantly higher per-patient sensitivity for detecting recurrent HCC compared to CECT (77.3% vs. 36.4%;
P=0.012), while maintaining similar per-patient specificity (98.9% vs. 98.3%;
P=0.999). As a result, NC-MRI also achieved better per-patient accuracy in detecting recurrent HCC than CECT (96.6% vs. 91.6%;
P=0.006).
HCC often recurs even after curative treatment, significantly impacting long-term survival outcomes. Early recurrence, prevalent within the initial two years post-treatment, is characterized by occult tumor dissemination and biological aggressiveness, often resulting in multifocal intrahepatic recurrence, vascular invasion, and distant metastasis [
26,
27]. Effective follow-up strategies utilizing CE imaging techniques such as dynamic CT or MRI are essential for early detection during this critical period. A previous study has indicated that NC-MRI may be less effective than hepatocyte-specific CE full sequence MRI, particularly in the detection of small disseminated intrahepatic and infiltrative recurrences with vascular invasion within the first year post-hepatic resection [
25]. Consequently, NC-MRI may not be the most suitable modality for detecting early recurrence during the first two years after curative treatment. In contrast, late recurrence, occurring beyond the initial two years, tends to have a lower incidence and is more often associated with
de novo carcinogenesis in the context of underlying chronic liver disease [
7]. Cumulative recurrence rates post-curative treatment have been reported as 39.7%, 60.3%, and 71.0% at 2, 5, and 10 years, respectively [
12], highlighting the necessity for long-term continuous follow-up.
Current international guidelines recommend continuous follow-up for detecting HCC recurrence, even beyond the first two years, due to the risk of late recurrence [
3-
6,
13]. However, there is no consensus on the optimal imaging modality or follow-up interval. While ultrasound serves as a standard for HCC surveillance in high-risk populations, its effectiveness diminishes for recurrent HCC detection after curative treatment due to restricted sonic windows and post-treatment anatomical changes [
9]. Additionally, ultrasound faces challenges in evaluating resection margins post-hepatic resection and treated areas following local ablation, which complicates the detection of recurrent HCCs. As a result, CECT or MRI are widely preferred in clinical practice due to their robust performance in identifying recurrent HCC, including cases of late recurrence. However, repeated follow-up examinations carry inherent risks such as cumulative radiation exposure and adverse reactions to contrast agents, including nephrotoxicity and hypersensitivity in CECT. Indeed, a recent study reported that approximately 1.7% (1,021/59,971) of patients were diagnosed with chronic kidney disease two years after cancer diagnosis, potentially due to the repeated use of iodine contrast. Moreover, the number of CT scans involving iodine contrast might contribute to long-term renal function impairment [
28]. CE-MRI, although effective, is limited by high costs and risks associated with gadolinium-based contrast agents such as gadolinium deposition in the body. These considerations highlight the need to carefully balance the benefits and risks of repeated imaging in long-term follow-up strategies for HCC recurrence detection.
In this context, NC-MRI emerges as a promising modality for detecting HCC recurrence, particularly for late recurrence, owing to its rapid scan times and absence of contrast agent use [
21]. NC-MRI protocols for surveillance typically include only the essential sequences necessary for HCC detection. As HCC lesions commonly exhibit moderately high signal intensity on T2-weighted imaging and diffusion restriction, both T2-weighted imaging and DWI are key components of NC-MRI [
29]. In this study, all 17 recurrent HCCs detected by NC-MRI demonstrated these imaging characteristics, reinforcing the diagnostic utility of T2-weighted imaging and DWI for recurrent HCC detection. Our study demonstrates that NC-MRI significantly outperformed CECT in detecting late recurrence, highlighting its potential utility. A recent meta-analysis reported pooled detection sensitivity and specificity of NC-MRI for primary HCC surveillance in high-risk populations at 86.8% and 90.3%, respectively [
30]. Our findings of 77.3% sensitivity and 98.9% specificity for NC-MRI in detecting late recurrent HCC align closely with these meta-analysis results, further validating its diagnostic reliability and supporting its role in long-term follow-up strategies for HCC recurrence detection.
The per-patient sensitivity of CECT to detect recurrent HCC in this study was notably low at 36.4%, which contrasts with higher sensitivity reported in previous study focusing on HCC surveillance in cirrhotic patients [
20]. This reduced sensitivity may be attributed to the small median size of recurrent HCC (1.3 cm) observed in our cohort, which poses challenges for accurate detection and characterization using CECT [
31,
32]. Consistent with our findings, Sangiovanni et al. [
33] reported a sensitivity of 44% for detecting 1–2 cm HCCs in cirrhotic patients using CECT. The decreased sensitivity of CECT for small HCCs can be explained by their frequent lack of washout in the portal venous or delayed phases, despite exhibiting non-rim arterial phase hyper-enhancement. This limitation is particularly relevant because the absence of washout prevents a diagnosis of HCC using the non-invasive imaging criteria. Additionally, when an arterial enhancing lesion is located in the peripheral portion of the liver and presents with a wedge-shaped appearance (
Fig. 2), it can be challenging to distinguish it from recurrent HCC. Moreover, in this study, a major cause of false-negative diagnoses in CECT in this study was the misclassification of AP shunts due to their lack of washout and shape of the lesion. In these instances, recurrent HCC could not be reliably detected using CECT alone. MRI, in contrast, may offer superior performance due to its high tissue contrast for detecting small HCC [
34,
35]. Additionally, DWI on NC-MRI can identify small lesions with high cellularity, which may indicate the presence of recurrent HCC. In addition to false-negative diagnosis of recurrent HCC, AP shunts may also lead to false-positive diagnoses, especially when they appear nodular. In this study, two of the three false-positive cases on CECT were attributed to AP shunts.
Despite its generally good performance, NC-MRI has limitations as a detection tool for late recurrent HCC after curative treatment. NC-MRI lacks CE dynamic sequences necessary for definitive HCC imaging diagnosis, necessitating recall examinations for lesions detected. However, given the lower recurrence rate after the initial two years compared to early recurrence, using NC-MRI for initial detection followed by confirmatory CE dynamic CT or MRI could mitigate risks associated with radiation and contrast exposure [
9,
26,
36].
In addition to the need for confirmatory testing, the detection of extrahepatic metastases may be limited with NC-MRI due to its narrower scan coverage. Furthermore, because NC-MRI does not involve the use of contrast agents, its ability to detect vascular invasion may be inferior to that of CECT or MRI. In our study, NC-MRI missed aortocaval lymph node metastasis, which was identified on CECT, highlighting a potential drawback of NC-MRI. Therefore, further randomized controlled trials incorporating both NCMRI and CE imaging are needed to comprehensively assess the diagnostic performance of NC-MRI in detecting late recurrent HCC, including extrahepatic metastasis and vascular invasion. Nonetheless, as the incidence of extrahepatic spread and vascular invasion decreases beyond two years after curative treatment, NC-MRI may still serve as a practical tool for surveillance focused on intrahepatic recurrence [
9,
26,
27].
This study had several limitations. It was designed as a single-arm intra-individual head-to-head comparison, with all participants undergoing both NC-MRI and CECT, precluding a comparison of survival outcomes between the two imaging modalities. Furthermore, intra-individual comparison design may have led to a potential underestimation of the performance of CECT in detecting small late recurrent HCCs, as such lesions might have been identified on subsequent follow-up CECT, even if missed on a single examination. Additionally, assessment of stage migration was not feasible due to the intra-individual comparison design. Randomized controlled trials are warranted to determine whether NC-MRI detection for late recurrence after curative HCC treatment could improve overall survival compared to CECT. Additionally, the majority of participants in this study had hepatitis B viral infection as the underlying liver disease (82.8%). Further studies involving diverse etiologies of liver disease are necessary to generalize the study results. Given that many patients with HCC ultimately require systemic or transarterial therapy due to high recurrence risk even after initial curative treatment, and that CE imaging is typically needed for evaluating tumor burden and treatment response, NC-MRI may be applicable only to a subset of patients. Lastly, the use of CECT to confirm potential false-negative results following the three paired imaging sessions may represent another limitation of this study. This approach could have contributed to an overestimation of sensitivity, particularly given the relatively low sensitivity of CECT for detecting late recurrent HCC observed in this study.
In conclusion, NC-MRI demonstrated significantly better sensitivity and accuracy in detecting recurrent HCC in patients with a disease-free period of more than two years after curative treatment compared to CECT. Given its lack of radiation exposure and risk of contrast-agent-related adverse events, coupled with its good detection performance, NC-MRI could represent a preferred detection modality for late recurrence of HCC after curative treatment.
FOOTNOTES
-
Authors’ contributions
Final approval of manuscript: all authors. Dong Wook Kim: data curation, perform statistical analysis, draft of initial manuscript. Won Chang: data curation, perform statistical analysis, draft of initial manuscript. So Yeon Kim: data curation, critical revision of manuscript. Young-Suk Lim: data curation, critical revision of manuscript. Jonggi Choi: data curation, critical revision of manuscript. Jungheum Cho: data curation, critical revision of manuscript. Jin-Wook Kim: data curation, critical revision of manuscript. Jai Young Cho: data curation, critical revision of manuscript. Yun Bin Lee: data curation, critical revision of manuscript. Eun Ju Cho: data curation, critical revision of manuscript. Su Jong Yu: data curation, critical revision of manuscript. Kyung-Suk Suh: data curation, critical revision of manuscript. Kwang-Woong Lee: data curation, critical revision of manuscript. Dong Ho Lee: study concept, study supervision, critical revision of manuscript.
-
Acknowledgements
This study was supported by the National R&D Program for Cancer Control through the National Cancer Center (NCC), funded by the Ministry of Health & Welfare, Republic of Korea (HA21C0163).
This study is registered at clinicaltrials.gov (study identifier: NCT05690451) and cris.nih.go.kr (study identifier: KCT0006395).
-
Conflicts of Interest
The authors have no conflicts to disclose.
SUPPLEMENTAL MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Figure 1.Flow diagram of study enrollment. CECT, contrast-enhanced computed tomography; HCC, hepatocellular carcinoma; NC-MRI, non-contrast magnetic resonance imaging.
Figure 2.A 72-year-old woman underwent hepatic resection for HCC 28 months ago. (A) Contrast-enhanced arterial phase axial CT image shows wedge shape arterial hyper-enhancement in the subcapsular portion of segment III (arrows). (B) On the portal venous axial image, there is no washout in the corresponding area, indicating an arterioportal shunt. (C) On the T2-weighted axial image from noncontrast abbreviated MRI obtained the same day, a 1.3 cm lesion with high signal intensity is visible in the subcapsular portion of segment III (arrow). (D) This nodule also demonstrates high signal intensity on the diffusion-weighted image with a b-value of 800 (arrow). Based on the NC-MRI findings, intrahepatic recurrent HCC was suspected. (E) Subsequently, hepatocyte-specific contrast-enhanced MRI was performed. The arterial phase axial MRI shows arterial hyper-enhancement of the nodule (arrow). (F) In the hepatobiliary phase, the nodule exhibits low signal intensity (arrow). A non-invasive imaging diagnosis of recurrent HCC was confirmed, and the patient underwent treatment with transarterial chemoembolization. CT, computed tomography; HCC, hepatocellular carcinoma; NC-MRI, non-contrast magnetic resonance imaging.
Figure 3.A 74-year-old woman underwent hepatic resection for HCC 58 months ago. (A) Contrast-enhanced portal venous phase axial CT image reveals a 3.3 cm enlarged lymph node in the aortocaval space (arrow). (B) T2-weighted axial image from the non-contrast abbreviated MRI obtained the same day also shows the lymph node enlargement in the aortocaval space (arrow). However, it was missed during interpretation by human error and was not reported in the NC-MRI findings. To confirm the diagnosis, a biopsy of the enlarged aortocaval lymph node was performed, and metastatic HCC was diagnosed through histopathologic examination. CT, computed tomography; HCC, hepatocellular carcinoma; NC-MRI, non-contrast magnetic resonance imaging.
Table 1.Characteristics of study participants
Table 1.
|
Characteristic |
Total (n=203) |
HCC recurred (n=22) |
No HCC recurred (n=181) |
P-value |
|
Sex (M:F) |
154:49 |
15:7 |
139:42 |
0.374 |
|
Age (yr) |
62.0 (57.0–69.0) |
64.0 (60.0–74.0) |
61.0 (57.0–67.3) |
0.050 |
|
Etiology of liver disease |
|
|
|
0.211 |
|
Hepatitis B virus |
168 (82.8) |
17 (77.3) |
151 (83.4) |
|
|
Hepatitis C virus |
16 (7.9) |
4 (18.2) |
12 (6.6) |
|
|
Alcoholic |
10 (4.9) |
1 (4.5) |
9 (5.0) |
|
|
Others |
9 (4.4) |
0 (0.0) |
9 (5.0) |
|
|
Presence of cirrhosis |
104 (51.2) |
14 (63.6) |
90 (49.7) |
0.218 |
|
Previous treatment modality |
|
|
|
0.522 |
|
Hepatic resection |
145 (71.4) |
17 (77.3) |
128 (70.7) |
|
|
Presence of microvascular invasion |
38 (18.7) |
4 (18.2) |
34 (18.8) |
|
|
Radiofrequency ablation |
58 (28.6) |
5 (22.7) |
53 (29.3) |
|
|
Stage of previous HCC |
|
|
|
|
|
Tumor number (1:2) |
192:11 |
20:2 |
172:9 |
0.422 |
|
Largest tumor size (cm) |
2.3 (1.5–3.7) |
2.3 (1.2–2.8) |
2.3 (1.6–3.9) |
0.209 |
|
BCLC stage (0:A:B) |
80:118:5 |
9:12:1 |
71:106:4 |
0.778 |
|
AFP at the time of previous HCC treatment (ng/mL) |
6.75 (2.80–76.33) |
5.90 (2.60–9.90) |
6.93 (2.90–84.48) |
0.202 |
|
Interval between prior treatment and enrollment (mo) |
48.0 (38.0–71.0) |
47.5 (35.0–62.0) |
48.5 (39.0–71.5) |
0.498 |
|
Follow-up interval of examination |
|
|
|
0.501 |
|
3-month interval |
9 (4.4) |
1 (4.6) |
8 (4.4) |
|
|
4-month interval |
48 (23.7) |
3 (13.6) |
45 (24.9) |
|
|
6-month interval |
146 (71.9) |
18 (81.8) |
128 (70.7) |
|
|
Laboratory findings at enrollment |
|
|
|
|
|
Albumin (g/dL) |
4.25 (4.00–4.50) |
4.25 (4.10–4.50) |
4.25 (4.00–4.50) |
0.780 |
|
Total bilirubin (mg/dL) |
0.80 (0.60–1.06) |
0.95 (0.70–1.20) |
0.80 (0.60–1.00) |
0.136 |
|
Prothrombin time, INR |
1.00 (0.96–1.04) |
0.98 (0.96–1.05) |
1.00 (0.97–1.04) |
0.721 |
|
Creatinine (mg/dL) |
0.86 (0.78–0.96) |
0.78 (0.69–0.94) |
0.87 (0.79–0.96) |
0.051 |
|
Platelet counts (103/mm3) |
164 (136–202) |
162 (129–196) |
165 (136–202) |
0.520 |
|
AFP (ng/mL) |
2.36 (1.80–3.03) |
2.53 (2.10–3.20) |
2.30 (1.80–3.00) |
0.250 |
Table 2.Comparison of diagnostic performance for recurrent HCC detection between CECT and non-contrast abbreviated MRI on a perpatient basis
Table 2.
|
CECT
|
NC-MRI
|
P-value |
|
Estimates (%) (n/N) |
95% CI (%) |
Estimates (%) (n/N) |
95% CI (%) |
|
Accuracy |
91.6 (186/203) |
87.8–95.5 |
96.6 (196/203) |
94.0–99.1 |
0.006 |
|
Sensitivity |
36.4 (8/22) |
14.5–58.2 |
77.3 (17/22) |
58.3–96.3 |
0.012 |
|
Specificity |
98.3 (178/181) |
96.5–100 |
98.9 (179/181) |
97.4–100 |
0.999 |
|
Positive predictive value |
72.7 (8/11) |
41.4–100 |
89.5 (17/19) |
74.3–100 |
0.236 |
|
Negative predictive value |
92.7 (178/192) |
89.0–96.4 |
97.3 (179/184) |
94.9–99.7 |
0.043 |
|
Diagnostic yield |
1.5 (8/528) |
0.5–2.6 |
3.2 (17/528) |
1.7–4.7 |
0.004 |
|
False referral rate |
0.9 (5/528) |
0.1–1.8 |
0.8 (4/528) |
0.1–1.5 |
0.999 |
Abbreviations
contrast-enhanced computed tomography
diffusion-weighted imaging
non-contrast magnetic resonance imaging
Protein Induced by Vitamin K Absence or antagonist-II
REFERENCES
- 1. Center MM, Jemal A. International trends in liver cancer incidence rates. Cancer Epidemiol Biomarkers Prev 2011;20:2362-2368.
- 2. Villanueva A. Hepatocellular carcinoma. N Engl J Med 2019;380:1450-1462.
- 3. European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J Hepatol 2018;69:182-236.
- 4. Marrero JA, Kulik LM, Sirlin CB, Zhu AX, Finn RS, Abecassis MM, et al. Diagnosis, staging, and management of hepatocellular carcinoma: 2018 practice guidance by the American association for the study of liver diseases. Hepatology 2018;68:723-750.
- 5. Korean Liver Cancer Association (KLCA) and National Cancer Center (NCC) Korea. 2022 KLCA-NCC Korea practice guidelines for the management of hepatocellular carcinoma. Korean J Radiol 2022;23:1126-1240.
- 6. Korean Liver Cancer Association (KLCA) and National Cancer Center (NCC) Korea. 2022 KLCA-NCC Korea practice guidelines for the management of hepatocellular carcinoma. J Liver Cancer 2023;23:1-120.
- 7. Imamura H, Matsuyama Y, Tanaka E, Ohkubo T, Hasegawa K, Miyagawa S, et al. Risk factors contributing to early and late phase intrahepatic recurrence of hepatocellular carcinoma after hepatectomy. J Hepatol 2003;38:200-207.
- 8. Montorsi M, Santambrogio R, Bianchi P, Donadon M, Moroni E, Spinelli A, et al. Survival and recurrences after hepatic resection or radiofrequency for hepatocellular carcinoma in cirrhotic patients: a multivariate analysis. J Gastrointest Surg 2005;9:62-67 discussion 67-68.
- 9. Jeon SK, Lee DH, Hur BY, Park SJ, Kim SW, Park J, et al. Abbreviated MRI for secondary surveillance of recurrent hepatocellular carcinoma after presumed curative treatment. J Magn Reson Imaging 2023;58:1375-1383.
- 10. Poon RT. Differentiating early and late recurrences after resection of HCC in cirrhotic patients: implications on surveillance, prevention, and treatment strategies. Ann Surg Oncol 2009;16:792-794.
- 11. Zheng J, Chou JF, Gönen M, Vachharajani N, Chapman WC, Majella Doyle MB, et al. Prediction of hepatocellular carcinoma recurrence beyond milan criteria after resection: validation of a clinical risk score in an international cohort. Ann Surg 2017;266:693-701.
- 12. Kim J, Kang W, Sinn DH, Gwak GY, Paik YH, Choi MS, et al. Substantial risk of recurrence even after 5 recurrence-free years in early-stage hepatocellular carcinoma patients. Clin Mol Hepatol 2020;26:516-528.
- 13. Omata M, Cheng AL, Kokudo N, Kudo M, Lee JM, Jia J, et al. Asia-Pacific clinical practice guidelines on the management of hepatocellular carcinoma: a 2017 update. Hepatol Int 2017;11:317-370.
- 14. Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology 2023;78:1922-1965.
- 15. Korean Liver Cancer Association (KLCA) and National Cancer Center (NCC) Korea. 2022 KLCA-NCC Korea practice guidelines for the management of hepatocellular carcinoma. Clin Mol Hepatol 2022;28:583-705.
- 16. Park HJ, Jang HY, Kim SY, Lee SJ, Won HJ, Byun JH, et al. Non-enhanced magnetic resonance imaging as a surveillance tool for hepatocellular carcinoma: Comparison with ultrasound. J Hepatol 2020;72:718-724.
- 17. Kim DH, Yoon JH, Choi MH, Lee CH, Kang TW, Kim HA, et al. Comparison of non-contrast abbreviated MRI and ultrasound as surveillance modalities for HCC. J Hepatol 2024;81:461-470.
- 18. Rhee H, Kim MJ, Kim DY, An C, Kang W, Han K, et al. Non-contrast Magnetic resonance imaging vs ultrasonography for hepatocellular carcinoma surveillance: a randomized, single-center trial. Gastroenterology 2025;168:1170-1177.e1112.
- 19. Decharatanachart P, Pan-Ngum W, Peeraphatdit T, Tanpowpong N, Tangkijvanich P, Treeprasertsuk S, et al. Cost-utility analysis of non-contrast abbreviated magnetic resonance imaging for hepatocellular carcinoma surveillance in cirrhosis. Gut Liver 2024;18:135-146.
- 20. Yoon JH, Lee JM, Lee DH, Joo I, Jeon JH, Ahn SJ, et al. A comparison of biannual two-phase low-dose liver CT and US for HCC Surveillance in a group at high risk of HCC development. Liver Cancer 2020;9:503-517.
- 21. Park HJ, Kim SY. Imaging modalities for hepatocellular carcinoma surveillance: expanding horizons beyond ultrasound. J Liver Cancer 2020;20:99-105.
- 22. Joo I, Lee JM, Koh YH, Choi SH, Lee S, Chung JW. 2022 Korean Liver Cancer Association-National Cancer Center Korea Practice Guidelines for imaging diagnosis of hepatocellular carcinoma: what’s new? Korean J Radiol 2023;24:1-5.
- 23. Kim JH, Joo I, Lee JM. Atypical appearance of hepatocellular carcinoma and its mimickers: how to solve challenging cases using gadoxetic acid-enhanced liver magnetic resonance imaging. Korean J Radiol 2019;20:1019-1041.
- 24. Kim TK, Choi BI, Han JK, Chung JW, Park JH, Han MC. Nontumorous arterioportal shunt mimicking hypervascular tumor in cirrhotic liver: two-phase spiral CT findings. Radiology 1998;208:597-603.
- 25. Min JH, Kim YK, Choi SY, Kang TW, Jeong WK, Kim K, et al. Detection of recurrent hepatocellular carcinoma after surgical resection: non-contrast liver MR imaging with diffusion-weighted imaging versus gadoxetic acid-enhanced MR imaging. Br J Radiol 2018;91:20180177.
- 26. Portolani N, Coniglio A, Ghidoni S, Giovanelli M, Benetti A, Tiberio GA, et al. Early and late recurrence after liver resection for hepatocellular carcinoma: prognostic and therapeutic implications. Ann Surg 2006;243:229-235.
- 27. Kang HJ, Kim H, Lee DH, Hur BY, Hwang YJ, Suh KS, et al. Gadoxetate-enhanced MRI features of proliferative hepatocellular carcinoma are prognostic after surgery. Radiology 2021;300:572-582.
- 28. Koo JH, Lee M, Kim EH, Oh HJ, Lim JS, Hyung WJ, et al. Harmful effect of repetitive intravenous iodinated contrast media administration on the long-term renal function of patients with early gastric cancer. Sci Rep 2023;13:19448.
- 29. Lee DH. Recent advances and issues in imaging modalities for hepatocellular carcinoma surveillance. J Liver Cancer 2025;25:31-40.
- 30. Chan MV, Huo YR, Trieu N, Mitchelle A, George J, He E, et al. Noncontrast MRI for hepatocellular carcinoma detection: a systematic review and meta-analysis - a potential surveillance tool? Clin Gastroenterol Hepatol 2022;20:44-56.e42.
- 31. Pocha C, Dieperink E, McMaken KA, Knott A, Thuras P, Ho SB. Surveillance for hepatocellular cancer with ultrasonography vs. computed tomography -- a randomised study. Aliment Pharmacol Ther 2013;38:303-312.
- 32. Roberts LR, Sirlin CB, Zaiem F, Almasri J, Prokop LJ, Heimbach JK, et al. Imaging for the diagnosis of hepatocellular carcinoma: A systematic review and meta-analysis. Hepatology 2018;67:401-421.
- 33. Sangiovanni A, Manini MA, Iavarone M, Romeo R, Forzenigo LV, Fraquelli M, et al. The diagnostic and economic impact of contrast imaging techniques in the diagnosis of small hepatocellular carcinoma in cirrhosis. Gut 2010;59:638-644.
- 34. Kim HD, Lim YS, Han S, An J, Kim GA, Kim SY, et al. Evaluation of early-stage hepatocellular carcinoma by magnetic resonance imaging with gadoxetic acid detects additional lesions and increases overall survival. Gastroenterology 2015;148:1371-1382.
- 35. Lee YJ, Lee JM, Lee JS, Lee HY, Park BH, Kim YH, et al. Hepatocellular carcinoma: diagnostic performance of multidetector CT and MR imaging-a systematic review and meta-analysis. Radiology 2015;275:97-109.
- 36. Xu XF, Xing H, Han J, Li ZL, Lau WY, Zhou YH, et al. Risk factors, patterns, and outcomes of late recurrence after liver resection for hepatocellular carcinoma: a multicenter study from China. JAMA Surg 2019;154:209-217.
