Abstract
Objectives:
To assess the predictive value of preoperative gadoxetic acid (GA)-enhanced magnetic resonance imaging (MRI) features and postoperative histopathological grading for early recurrence of hepatocellular carcinoma (HCC) without microvascular invasion (MVI) after curative hepatectomy.
Methods:
A total of 85 MVI-negative HCC cases were retrospectively analyzed. Cox analyses were used to identify the independent predictors of early recurrence (within a 24 months span). The clinical prediction Model-1 or Model-2 was established without or with postoperative pathological factor, respectively. Nomogram models were constructed and receiver operating characteristic (ROC) curve analysis was used to assess the models’ predictive ability. Internal validation of the prediction models for early HCC recurrence was performed using a bootstrap re-sampling approach.
Results:
In the multivariate cox regression analysis, Edmondson-Steiner grade, peritumoral hypointensity on hepatobiliary phase (HBP), and relative intensity ratio (RIR) in HBP were identified as independent variables associated with early recurrence. The C-index of the nomogram models and internal validation were both between 0.7 and 0.8, showing good model fitting and calibration effects. The area under the ROC curve (AUC) was 0.781 for Model-1 based on the two preoperative MRI factors. When a third factor, the Edmondson-Steiner grade, was included (Model-2), the AUC increased to 0.834, and the sensitivity increased from 71.4 to 96.4%.
Conclusions:
Edmondson-Steiner grade, peritumoral hypointensity on HBP, and RIR on HBP can help predict early recurrence of MVI-negative HCC. In comparison with Model-1 (only imaging features), Model-2 (imaging features + histopathological grades) increases the sensitivity in predicting early recurrence of HCC without MVI.
Advances in knowledge:
Preoperative GA-enhanced MRI signs are of great value in predicting early postoperative recurrence of HCC without MVI, and a combined pathological model was established to evaluate the feasibility and effectiveness of this technique.
Introduction
Hepatocellular carcinoma (HCC), the sixth most common cancer worldwide and the second most lethal cancer, represents a big challenge in the clinic in terms of diagnosis and treatment. 1 For patients whose liver is free from chronic disorders and functions well, curative hepatectomy remains the best therapeutic option for long-term survival. 2,3 However, early recurrence that leads to failure of the surgical treatment is common after hepatectomy. 4 Despite that there is no consensus on the cut-off latency that categorizes early and late recurrences, it is generally recognized in the field that early recurrence is related to a substantially worse prognosis than late recurrence. 5–7
Previous studies have suggested that early recurrence, i.e., within 2 years after surgery, of HCC may be related to some tumor factors, such as serum tumor markers, tumor diameter, microvascular invasion (MVI) grouping, and the expression of related genes. 4,8,9 Because MVI is considered an extremely strong predictor of cancer recurrence and linked to a bad prognosis of HCC after surgical resection, 10 many investigators have focused on the development and use of broad imaging tools and advanced hepatobiliary-specific contrast agents to more efficiently predict the different states of MVI in HCC and its prognosis, such as recurrence-free survival. 11,12 Gadoxetic acid (GA), which exhibits advantages of combining dynamic contrast-enhanced scanning with the liver parenchyma scanning and increasing the signal contrast between HCC lesions and the liver parenchyma to provide multiaspect lesion information, not only facilitates the analysis of the functional status of liver cells but also occasionally increases the efficiency of lesion diagnosis. 12,13
MVI-negative HCC cases, an important subset of patients who may undergo curative liver resection, are also susceptible to early recurrence. However, studies that specifically evaluate predictors of poor postoperative prognoses (e.g., early recurrence) in MVI-negative HCC remains limited. 14,15 In addition to the postoperative histopathological grading, 14 preoperative MRI features 15 have been reported to be correlated with early relapse and poor survival in HCC patients without MVI; these imaging features include as mosaic architecture, larger tumor size, and non-smooth tumor margins. Of note, imaging features of the hepatobiliary phase (HBP) were not incorporated in that study due to limitations of the contrast agent, gadolinium-diethylenetriamine pentaacetic acid (Gd-DTPA). 15 Therefore, the parameters and predictive models involving imaging and histopathological data remain to be perfected and validated so that reliable predictive models can be made available to clinicians for them to better implement adjuvant therapy and to design clinical trials. To that end, the present retrospective study was aimed to establish a refined predictive model to predict early recurrence in MVI-negative HCC patients, by comparing a preoperative prediction model solely based on GA-enhanced MRI parameters (Model-1) and an integrated model involving both preoperative MRI features and postoperative histopathological measures (Model-2).
Methods and materials
Study population
This study was approved by our institutional review board, and informed consents were waived because only histopathology and imaging data were retrospectively reviewed. We retrospectively analyzed the data of patients who underwent radical hepatic resection at our hospital between February 2015 and April 2021. Patient inclusion criteria were as follows: i) patients who underwent GA-enhanced MRI examination within two weeks prior to curative resection; ii) no history of other malignancies; iii) liver mass confirmed by pathology as hepatocellular carcinoma; iv) complete laboratory and pathological data were available; and v) MRI images were complete and of good quality for easy observation. The exclusion criteria are as follows: i) relevant treatment before surgery, including radiofrequency ablation, radiation therapy, or TACE; ii) pathological confirmation of MVI-positive HCC; iii) patients who died of postoperative complications; iv) incomplete follow-up data and undetermined recurrence status; v) patients who were followed less than 2 years; and vi) recurrence time more than 2 years (Figure 1). A total of 85 cases, among which 28 showed early recurrence of HCC within 24 months after surgery and 57 did not exhibit recurrent HCC during the cut-off latency (24 months).
Figure 1.
Flow chart of the study population HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; MVI, microvascular invasion; TACE, transcatheter arterial chemoembolization; ER, early recurrence
MRI protocol
Abdominal MRI was performed using a Philips 3.0 T Achieva MR scanner with a 16-channel abdominal coil. The patient was placed in the supine position with the head elevated, and the scan was performed from the top of the diaphragm to the lower poles of both the kidneys. The specific parameters of the MRI scan sequence were as follows: (1) Fast Spoiled Gradient Recalled Echo in-Phase/opposed-phase imaging, repetition time (TR) 200 ms, echo time (TE) 2.5 ms, slice thickness 6.0 mm, slice gap 1.0 mm, matrix 230 × 250; (2) Fast Spin Echo Sequence Cross-Sectional Fat Suppression T2WI, TR 1900 ms, TE 80 ms, slice thickness 5.0 mm, slice gap 1.0 mm, matrix 250 × 250; (3) Diffusion-weighted imaging (DWI) with a spin-echo planar imaging sequence, TR 3500～6000 ms, TE 60 ms, slice thickness 5.0 mm, slice gap 1.0 mm, b values = 0, 800 seconds/mm2, matrix 128 × 160. Afterwards, a contrast agent (GA, Bayer Healthcare, Germany) was rapid bolus injected through the cubital vein at 0.1–0.15 mL/kg body weight, and the injection flow rate was 1.0–1.5 ml s−1. After injection, the solution was flushed with 20 ml normal saline. T1 high-resolution isotropic volume excitation (THRIVE) sequence was used for enhanced scanning, TR 2.8 ms, TE 1.8 ms, slice thickness 2.5 mm, scan without gaps, matrix 230 × 250. Axial images of the arterial, portal, and transitional phases were collected 22 s, 1 min, and 3 min after contrast injection, and hepatobiliary phase images were collected 20 min later.
Imaging analysis
Images of GA-enhanced MRI were independently analyzed by two senior radiologists using a picture archiving and communication system (PACS, eWorldRIS v. 4.9, Hangzhou, China). Both radiologists who had over 10 years of diagnostic experience in abdominal MRI and were blinded to clinical, pathological, laboratory, and follow-up information. When the two radiologists had different verdicts, a joint review was conducted until a consensus was reached on the final decision.
Quantitative image parameter analysis The two radiologists independently measured the longest diameter of the tumor on the axial and coronal planes of the hepatobiliary images and defined it as the tumor size. The average of these two measurements was used for analysis. A region of interest (ROI) in the liver and lesion areas was drawn. The ROI was as large as possible to ensure that the ROIs of the same patient in different periods were selected from the same level of the scanned image with the same size and at a comparable position. The signal intensities (SI) of the HCC lesions and background liver parenchyma were measured in the non-enhanced, arterial, portal, transitional, and hepatobiliary phases. The average signal intensities were obtained for each sequence. Blood vessels, bile ducts, and necrotic areas were avoided as much as possible when measuring the SI of the background liver parenchyma, and the relative intensity ratio (RIR) of different phases was calculated. The calculation formula was as follows 16 : RIR = SI of the nodule/SI of the liver parenchyma.
Qualitative parameter analysis We listed preoperative GA-enhanced MRI sequences and MRI findings to differentiate them from LI-RADS imaging features. For some definitions, refer to the following references 17–19 and our clinical experience according to the different time periods of contrast agent imaging: (a) we evaluated the following signs on the plain MRI scan image. (1) intralesional fat was defined as a localized loss of signal within the tumor on out-of-phase images. (2) intratumoral hemorrhage, the most acute hemorrhage usually manifests as heterogeneous T1 hyperintensity and mixed T2 hypointensity. (3) intratumoral cystic degeneration or necrosis, defined as the intratumoral high-signal area on T2WI or the low-signal area that is not enhanced on dynamic contrast-enhanced scanning. (4) DWI target sign, defined as a tumor with diffusion restriction, inhomogeneous hypointense or mixed signal in the central area, surrounded by a high signal ring. (b) Arterial phase. (1) arterial peritumoral enhancement, which is defined as the “band-like” or “flame-like” irregularly enhanced region adjacent to the lesion observable in the arterial phase, and its enhancement degree in the portal venous phase is slightly higher than or equal to that of the surrounding normal liver parenchyma. (2) rim arterial phase hyperenhancement, regular or irregular annular hyperenhancement at the edge of the tumor on the AP, with a hypointense area in the centre. (3) intratumoral vessels and macroscopic vascular shadows in the tumor in the AP. (4) ≥50% arterial phase hypovascular component, defined as an area of 50% or less with significantly higher enhancement than the liver parenchyma. (c) Portal venous phase: peripheral washout manifested as obvious enhancement in the arterial phase and rapid clearance in the portal venous phase, that is, the “rapid rise and fall” of the contrast agent. (d) Hepatobiliary phase. (1) smooth tumor margin, a round or oval tumor with a smooth margin or nodular protrusions, or non-smooth tumor margin, appearing as a lobulated nodular tumor with a budding portion protruding into the liver parenchyma on both the axial and coronal HBP images. (2) Hepatobiliary phase peritumoral hypointensity, defined as an irregularly shaped area outside the tumor margins with a signal lower than the normal liver parenchyma and higher than the tumor itself. The integrity of the radiological capsule was further judged according to the presence or absence of thin annular hyperintensity at the tumor margins in the portal vein in the delayed and hepatobiliary phases, and the enhancing capsule was assessed on the axis of the transition phase.
Preoperative clinical data and postoperative pathological analysis
Clinical baseline data and laboratory parameters were obtained from electronic medical records, including sex, age, cause of liver disease, Child-Pugh grade, liver cirrhosis, alpha-fetoprotein (AFP), and protein induced by vitamin K absence or antagonist-II (PIVKA-II). The cut-off levels for these three serum markers were routinely used by the agency.
Pathological specimens in this study were obtained from surgically resected tumoral tissue in all patients, followed by serial sections (approximately 4–5 µm in thickness), which were confirmed by hematoxylin-eosin (HE) staining, and two pathologists histologically graded the HCC using the Edmondson-Steiner (E-S) grading system. The E-S grades I and II corresponded to well/moderately differentiated tumors (24 cases), and E-S grades III and IV corresponded to poorly differentiated tumors (61 cases). Monoclonal antibodies against CK7 and CK19 were used for immunohistochemical staining (LSAB method) to observe the bile duct epithelial markers CK7 and CK19.
Follow-up
Postoperative outpatient re-examination and telephone follow-up were performed until 1 March 2022. Follow-up was performed monthly for the first three months after surgery, every three months for the first two years, and every six months thereafter. Follow-up tests included liver function, AFP, and other serological parameters, whole-body CT or MRI examination, and abdominal B-ultrasound examination. The investigators suspected early recurrence based on the appearance of new intrahepatic or extrahepatic tumor lesions within two years after liver resection, and then confirmed the diagnosis by biopsy of excised tumor samples. 4 Recurrence-free survival (RFS) was defined as the interval between the date of surgery and the date when the tumor recurrence was first observed on radiology within two years or as a date with no recurrence until the most recent follow-up. 8,9
Statistical analysis
The Kolmogorov-Smirnov test was used to determine whether the continuous variable data met the normal distribution. Data conforming to a normal distribution are expressed as mean ± standard deviation ( ± s); otherwise, they are expressed as median (interquartile range), and the descriptions of categorical variables are presented as percentages (%). Differences in qualitative data were compared using the chi-squared test or Fisher’s exact test. The Student’s t- test was used to compare the differences in the normally distributed continuous data between two groups. The Mann-Whitney U test was used to compare abnormally distributed variables between two groups. The Kaplan-Meier method was used to calculate recurrence-free survival (RFS) and plot the survival curve, and the log-rank test was used to compare the differences in survival curves between the two groups. Stepwise cox regression multivariate analysis was performed on variables with p < 0.05 on univariate analysis to determine independent risk factors. Finally, two prediction models were established using the screened independent risk factors and were presented in the form of nomograms. Then, the bootstrap re-sampling (100 replications) method was used to verify the risk prediction models internally, and calibration plots were used to graphically assess the agreement between the probability of ER as predicted by the model and the observed probability. Assessing the discrimination of the predictive models was mainly based on the AUC area of the receiver operating characteristic (ROC) curve; 95% CI, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were also calculated. Concordance and methodological agreement between the two radiologists were assessed by calculating the intraclass correlation coefficients. All statistical analyses were performed using SPSS 25.0 software academic analysis. p < 0.05 was considered statistically significant.
Results
Clinicopathologic characteristics and recurrence
A total of 85 patients (60 men and 25 women; 58.13 ± 9.53 years of age) were included in the MVI-negative patient analysis cohort. The patients were divided into two groups (28 patients with early recurrence of HCC and 57 patients with non-early recurrence). The mean (± SD) time to early recurrence after hepatectomy was 10.5 ± 5.9 months (range, 1–22 months). The median follow-up period was 37 months (range, 24–64 months). A total of 24 cases (28.2%) were moderately to well differentiated, and 61 cases (71.8%) were poorly differentiated. Edmondson-Steiner grade was a high-risk factor for early recurrence of HCC without MVI according to the chi-square test (p < 0.001, Table 1). There were no significant differences between the two groups in clinical baseline and biopathological characteristics, including age, sex, serum AFP level (<400 or ≥400 ng ml−1), and CK 7, CK 19, and PIVKA-II levels (Table 1).
Table 1.
Clinicopathologic characteristics associated with early recurrence in MVI-negative HCCs
| Characteristic | Total (n = 85) | Non-ER (n = 57) | ER (n = 28) | P value |
|---|---|---|---|---|
| Age(years) a | 58.13 ± 9.53 | 59.21 ± 9.52 | 55.93 ± 9.34 | 0.137 |
| Gender | 0.101 | |||
| Male | 60 (70.6) | 37 (64.9) | 23 (82.1) | |
| Female | 25 (29.4) | 20 (35.1) | 5 (17.9) | |
| Cause of liver disease | 0.373 | |||
| Hepatitis B virus | 81 (95.3) | 53 (93.0) | 28 (100) | |
| Other | 4 (4.7) | 4 (7.0) | 0 | |
| Liver cirrhosis | 0.885 | |||
| No | 5 (5.9) | 4 (7.0) | 1 (3.6) | |
| Yes | 80 (94.1) | 53 (93.0) | 27 (96.4) | |
| Child-Pugh grade | 1.000 | |||
| Grade A | 79 (92.9) | 53 (93.0) | 26 (92.9) | |
| Grade B | 6 (7.1) | 4 (7.0) | 2 (7.1) | |
| AFP level | 0.889 | |||
| <400 ng ml−1 | 72 (84.7) | 49 (86.0) | 23 (82.1) | |
| ≥400 ng ml−1 | 13 (15.3) | 8 (14.0) | 5 (17.9) | |
| PIVKA-Ⅱ level | 0.174 | |||
| ＜40 mAU/ml | 33 (38.8) | 25 (43.9) | 8 (28.6) | |
| ≥40 mAU/ml | 52 (61.2) | 32 (56.1) | 20 (71.4) | |
| CK 7 | 0.935 | |||
| Positive | 45 (52.9) | 30 (52.6) | 15 (53.6) | |
| Negative | 40 (47.1) | 27 (47.4) | 13 (46.4） | |
| CK 19 | 0.545 | |||
| Positive | 18 (21.2) | 11 (19.3) | 7 (25.0) | |
| Negative | 67 (78.8) | 46 (80.7) | 21 (75.0) | |
| Edmondson-Steiner grade | <0.001 | |||
| I–II grade | 24 (28.2) | 23 (40.4) | 1 (3.6) | |
| III–IV grade | 61 (71.8) | 34 (59.6) | 27 (96.4) | |
| Tumor size(cm) b | 2.4 (1.50–3.95) | 2.4 (1.50–3.85) | 2.65 (1.58–4.38) | 0.584 |
AFP, alpha-fetoprotein; CK7, cytokeratin 7; CK7, cytokeratin 7; CK19, cytokeratin 19; CK19, cytokeratin 19; ER, early recurrence; PIVKA-II, protein induced by vitamin K absence or antagonist-II.
Unless otherwise indicated, data are expressed as the number of patients, with percentages in parentheses. Categorical variables were compared using the chi-square test or Fisher’s exact test
Data are continuous variables and expressed as mean ± standard deviation
Data are expressed as median with interquartile ranges in parentheses.
Preoperative MR imaging features analysis
Regarding radiological features (Table 2), MRI showed intratumor fat in 70 cases (82.4%) and more HCCs with no fat in the ER group than in the non-ER group (27/28 [96.4%] vs 43/57 [75.4%]; p = 0.037), suggesting that the prognosis of non–fat-containing HCCs was poor, and early recurrence was more common. Peritumoral hypointensity on HBP was observed more frequently in the ER group than in the non-ER group (18/28 [64.3%] vs 12/57 [21.1%]; p < 0.001). The RIR in HBP were 0.46 ± 0.08 (mean ± SD) and 0.55 ± 0.16 in the ER and non-ER groups, respectively. In contrast, the RIR in the HBP group was significantly lower than that in the ER group (p = 0.007). Regarding other MR features, there were no statistically significant differences between the two groups. Overall, the results of the inter observer agreement analysis yielded good to excellent agreement (ICC = 0.733–0.972). The interobserver agreement of the two MRI findings in predicting early recurrence of MVI-negative HCC was excellent (ICC = 0.812 for RIR in HBP and 0.877 for peritumoral hypointensity in HBP).
Table 2.
The qualitative and quantitative imaging characteristics according to ER
| Total (n = 85) | Non-ER (n = 57) | ER (n = 28) | P value | |
|---|---|---|---|---|
| Pre-images | ||||
| Intratumor fat | 0.037 | |||
| Absent | 70 (82.4) | 43 (75.4) | 27 (96.4) | |
| Present | 15 (17.6) | 14 (24.6) | 1 (3.6) | |
| Intratumor hemorrhage | 0.764 | |||
| Absent | 73 (85.9) | 48 (84.2) | 25 (89.3) | |
| Present | 12 (14.1) | 9 (15.8) | 3 (10.7) | |
| Intratumor cystic degeneration or necrosis | 0.388 | |||
| Absent | 75 (88.2) | 52 (91.2) | 23 (82.1) | |
| Present | 10 (11.8) | 5 (8.8) | 5 (17.9) | |
| DWI target sign | 0.065 | |||
| Absent | 76 (89.4) | 48 (84.2) | 28 (100) | |
| Present | 9 (10.6) | 9 (15.8) | 0 | |
| Arterial phase | ||||
| Arterial peritumoral enhancement | 0.681 | |||
| Absent | 66 (77.6) | 45 (78.9) | 21 (75) | |
| Present | 19 (22.4) | 12 (21.1) | 7 (25) | |
| Rim arterial phase hyperenhancement | 0.915 | |||
| Absent | 77 (90.6) | 51 (89.5) | 26 (92.9) | |
| Present | 8 (9.4) | 6 (10.5) | 2 (7.1) | |
| Intratumoral vessels | 0.968 | |||
| Absent | 67 (78.8) | 45 (78.9) | 22 (78.6) | |
| Present | 18 (21.2) | 12 (21.1) | 6 (21.4) | |
| ≥50% arterial phase hypovascular component | 0.453 | |||
| Absent | 69 (81.2) | 45 (78.9) | 24 (85.7) | |
| Present | 16 (18.8) | 12 (21.1) | 4 (14.3) | |
| Portal venous phase | ||||
| Peripheral washout on PVP | 0.932 | |||
| Absent | 11 (12.9) | 8 (14.0) | 3 (10.7) | |
| Present | 74 (87.1) | 49 (86.0) | 25 (89.3) | |
| Hepatobiliary phase | ||||
| Tumor margin | 0.873 | |||
| smooth | 69 (81.2) | 46 (80.7) | 23 (82.1) | |
| non-smooth | 16 (18.8) | 11 (19.3) | 5 (17.9) | |
| Peritumoral hypointensity on HBP | <0.001 | |||
| Absent | 55 (64.7) | 45 (78.9) | 10 (35.7) | |
| Present | 30 (35.3) | 12 (21.1) | 18 (64.3) | |
| Radiological capsule | 0.951 | |||
| 1, Complete | 33 (38.8) | 22 (38.6) | 11 (39.3) | |
| 0, Incomplete/Without | 52 (61.2) | 35 (61.4) | 17 (60.7) | |
| Enhancing capsule | 0.765 | |||
| Absent | 62 (72.9) | 41 (71.9) | 21 (75.0) | |
| Present | 23 (27.1) | 16 (28.1) | 7 (25.0) | |
| Quantitative image features | ||||
| RIR on Pre images a | 0.79 ± 0.13 | 0.79 ± 0.12 | 0.79 ± 0.15 | 0.838 |
| RIR in AP a | 1.51 ± 0.46 | 1.47 ± 0.44 | 1.58 ± 0.50 | 0.284 |
| RIR in PVP a | 0.85 ± 0.19 | 0.85 ± 0.18 | 0.84 ± 0.20 | 0.768 |
| RIR in TP a | 0.75 ± 0.16 | 0.75 ± 0.15 | 0.74 ± 0.17 | 0.772 |
| RIR in HBP a | 0.52 ± 0.14 | 0.55 ± 0.16 | 0.46 ± 0.08 | 0.007 |
AP, arterial phase; DWI, diffusion-weighted imaging; ER, early recurrence; HBP, hepatobiliary phase; PVP, portal venous phase; Pre, pre-contrast; RIR, relative intensity ratio; TP, transitional phase.
Unless otherwise specified, data are number of patients (percentage)
Data are continuous variables and expressed as mean ± standard deviation
Factors of univariate and multivariate analysis for ER prediction
Variables that were found to be significant in univariate analysis remained significant in the multivariate cox regression model. The results of the two analyses were consistent. Post-hepatectomy early recurrence was associated with three predictors: Edmondson-Steiner grade (hazard ratio, 8.057; 95% CI, 1.056–61.480; p = 0.044), peritumoral hypointensity on HBP (hazard ratio, 2.786; 95% CI, 1.265–6.138; p = 0.011), and the RIR in HBP (hazard ratio, 0.038; 95% CI, 0.002–0.933; p = 0.045). The optimal cut-off point for sensitivity and specificity in predicting the early recurrence of MVI-negative HCC was 0.596 for RIR in HBP (Table 3, Figures 2 and 3).
Table 3.
Univariate and multivariate cox regression analysis
| Variable | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | p value | HR | 95% CI | p value | |
| AFP level | 1.343 | 0.510,3.533 | 0.550 | |||
| PIVKA-Ⅱ level | 1.768 | 0.778,4.016 | 0.173 | |||
| Edmondson-Steiner grade | 13.142 | 1.784,96.800 | 0.011 a | 8.057 | 1.056,61.480 | 0.044 a |
| CK 7 | 1.056 | 0.503,2.220 | 0.885 | |||
| CK 19 | 1.304 | 0.554,3.070 | 0.543 | |||
| Tumor size | 1.146 | 0.943,1.393 | 0.170 | |||
| Intratumor fat | 0.144 | 0.020,1.058 | 0.057 | |||
| Intratumor hemorrhage | 0.647 | 0.195,2.144 | 0.476 | |||
| Intratumor cystic degeneration or necrosis | 2.196 | 0.833,5.785 | 0.112 | |||
| DWI target sign | 0.040 | 0.000,5.226 | 0.196 | |||
| Arterial peritumoral enhancement | 1.177 | 0.500,2.770 | 0.709 | |||
| Rim arterial phase hyperenhancement | 0.711 | 0.169,2.996 | 0.642 | |||
| Intratumoral vessels | 1.013 | 0.411,2.498 | 0.978 | |||
| ≥50% arterial phase hypovascular component | 0.687 | 0.238,1.980 | 0.486 | |||
| Peripheral washout on PVP | 1.323 | 0.399,4.384 | 0.647 | |||
| Tumor margin | 0.975 | 0.371,2.565 | 0.959 | |||
| Peritumoral hypointensity on HBP | 4.272 | 1.965,9.289 | ＜0.001 a | 2.786 | 1.265,6.138 | 0.011 a |
| Radiological capsule | 1.018 | 0.477,2.173 | 0.964 | |||
| Enhancing capsule | 0.886 | 0.376,2.084 | 0.781 | |||
| RIR on Pre images | 1.405 | 0.069,28.431 | 0.825 | |||
| RIR in AP | 1.442 | 0.692,3.002 | 0.329 | |||
| RIR in PVP | 0.758 | 0.104,5.546 | 0.785 | |||
| RIR in TP | 0.762 | 0.076,7.670 | 0.818 | |||
| RIR in HBP | 0.02 | 0.001,0.492 | 0.017 a | 0.038 | 0.002,0.933 | 0.045 a |
AFP, alpha-fetoprotein; AP, arterial phase; CI, confidence interval; CK7, cytokeratin 7; CK19, cytokeratin 19; DWI, diffusion-weighted imaging; HBP, hepatobiliary phase; HR, hazard ratio;PIVKA-II, protein induced by vitamin K absence or antagonist-II; PVP, portal venous phase; Pre, pre-contrast; RIR, relative intensity ratio; TP, transitional phase.
Statistically significant results from the cox proportional hazards regression analysis. Variables with p < 0.05 in univariable analysis were applied to multivariable analysis
Figure 2.
A 50-year-old male who had hepatic cirrhosis related to hepatitis B andunderwent radical hepatectomy for HCC (a) The tumor was approximately 4.9 cm in diameter with unevenly high signals on T2 WI. (b) The mass (arrow) shows diffusion restriction with high SI on a diffusion-weighted image (b = 800 sec/mm2). (c) Gadoxetic acid–enhanced arterial phase MR image shows a hypervascular mass with peritumoral enhancement (arrow). (d) In the hepatobiliary phase (HBP), the mass showed peritumoral hypointensity (arrowheads). On histopathology after hepatectomy, the tumor was classified as Edmondson-Steiner grade III HCC without microvascular invasion.
Figure 3.
After 7 months from curative resection for HCC, intrahepatic tumor recurrence was detected by follow-up MR imaging An approximately 2.0 cm mass was seen in S8 (arrow). (a) T2 WI showed a moderately hyperintense lesion with smooth margin. (b) Diffusion restricted on DWI. (c) Dynamic contrast-enhanced scans showed enhancement of the lesions (arrow) in the arterial phase, but no enhancement in the central necrotic area. (d) The mass showed peritumoral hypointensity (arrowheads) on HBP.
Performance and comparison of Model-1 and Model-2 in predicting early recurrence of MVI-negative HCC
All independent survival factors in the multivariate analysis were selected as nomograms (Figure 4). The survival probability calibration curve showed that the nomogram predictions were in good agreement with the actual observations. The C-index of nomograms for predicting RFS was 0.719 (Model-1, 95% CI, 0.625–0.813) and 0.758 (Model-2, 95% CI, 0.674–0.842), respectively (Figure 5).
Figure 4.
Model-1 (A) and Model-2 (B) nomograms to predict the probability of 6-months, 1- or 2-years recurrence-free survival in patients with MVI-positive HCC after hepatectomy To use the nomogram, the value of an individual patient is located on each variable axis, and a line is drawn upward to determine the number of points received for each variable value. The sum of these numbers is located on the total point axis, and a line is drawn downward to the survival axes to determine the likelihood of 6-months, 1- or 2-years RFS. HBP, hepatobiliary phase; RIR, relative intensity ratio
Figure 5.
The calibration curve for predicting RFS of patients at 6 months, 1 year, and 2 years, in the Model-1(A); predicting patient RFS at 6 months, 1 year, and 2 years, in the Model-2 (B) Nomogram-predicted probability of survival is plotted on the x-axis; actual survival is plotted on the y-axis
The results of the ROC curve showed that the AUC of Edmondson-Steiner grade, peritumoral hypointensity on HBP, and RIR in HBP for predicting MVI-negative HCC-ER were 0.684, 0.716, and 0.653, respectively. When the inclusion parameters did not include the Edmondson-Steiner grade, the Model-1 had a sensitivity and specificity of 74.1% (20/28) and 77.2% (44/57), respectively. When the independent variables included this pathological factor, the sensitivity and specificity of Model-2 for predicting early recurrence of MVI-negative HCC were 96.4% (27/28) and 56.1% (32/57), respectively. The ROC areas of Model-1 and Model-2 diagnostic models were 0.781 and 0.834, respectively. There was no statistically significant difference in predictive power between the two models (p = 0.087) (Table 4), Figure 6). Therefore, the nomograms of both models had good prediction accuracy and discriminative ability.
Table 4.
Diagnostic performance of three factors and combinations for predicting for early recurrence of MVI-negative HCC
| Factors | AUC（95% CI） | Sensitivity (%) | Specificity (%) | Accuracy (%) | PPV (%) | NPV (%) |
|---|---|---|---|---|---|---|
| Peritumoral hypointensity on HBP | 0.716 (0.608–0.809) | 64.3 (18/28) | 78.9 (45/57) | 74.1 (63/85) | 60 (18/30) | 81.8 (45/55) |
| RIR in HBP | 0.653 (0.542–0.753) | 96.4 (27/28) | 35.1 (20/57) | 55.3 (47/85) | 42.2 (27/64) | 95.2 (20/21) |
| Edmondson-Steiner grade | 0.684 (0.574–0.781) | 96.4 (27/28) | 40.4 (23/57) | 58.8 (50/85) | 44.3 (27/61) | 95.8 (23/24) |
| Model-1 | 0.781 (0.678–0.864) | 71.4 (20/28) | 77.2 (44/57) | 75.3 (64/85) | 60.6 (20/33) | 84.6 (44/52) |
| Model-2 | 0.834 (0.738–0.906) | 96.4 (27/28) | 56.1 (32/57) | 69.4 (59/85) | 51.9 (27/52) | 97 (32/33) |
AUC, area under the curve; CI, confidence interval; HBP, hepatobiliary phase; NPV, negative predictive value; PPV, positive predictive value; RIR, relative intensity ratio.
Data are expressed as percentages except for AUC. The data in parentheses are the number of subjects used to calculate the percentage
Model-1=combination of two. MRI findings; Model-2= combination of all three findings.
Figure 6.
The area under the receiver operating characteristic curve for the ER prediction model were 0.781 (95% confidence interval: 0.678–0.864) and 0.834 (95%confidence interval: 0.738–0.906) in the Model-1 and Model-2, respectively.
Recurrence-free survival
Using Kaplan-Meier plots and log-rank tests, Edmondson-Steiner grade (p = 0.001; Figure 7), peritumoral hypointensity on HBP (p = 0.000; Figure 7), and RIR in HBP (p = 0.003; Figure 7) were significantly associated with shorter RFS in the 85 patients with MVI-negative HCC. The 6-, 12-, and 24 months RFS rates of patients with MVI-negative HCC were 90.6%, 78.8%, and 67.1%, respectively (Figure 7).
Figure 7.
Effect of individual parameters on recurrence-free survival of the included cases
Discussion
Our study demonstrated that Edmondson-Steiner grade, peritumoral hypointensity on HBP, and RIR on HBP were independent significant variables related to early recurrence of MVI-negative HCC, with excellent interobserver agreement. When the two MRI features of these three predictors were combined, the sensitivity of Model-1 was 71.4%, and the specificity was 77.2%. Model-2, with the addition of postoperative histopathological features to predict ER, had the potential to increase predictive performance, with a sensitivity>95%.
GA is a hepatocyte-specific contrast agent. 13 Zhou et al 20 hypothesized that the expression of organic anion transport polypeptide (OATP1B3) can predict hepatocyte uptake of GA, and it is intimately linked to signal intensity on HBP. The prevalence of peritumoral hypointensity in HBP could signify compensatory insufficiency of peritumoral arterial branch blood supply, which could be caused by tumor breakthrough of the capsule, invasion of peritumoral portal vein branches, and creation of portal vein tumor thrombi. Consequently, the expression of OATP on the cell membrane of cells lacking normal hepatocyte function is reduced, GA is scarcely absorbed, and the local liver tissue around the tumor on the hepatobiliary phase image exhibits a comparatively low signal compared to the surrounding liver parenchyma. 20,21 Many clinical studies have suggested that HBP peritumoral hypointensity is significantly correlated with malignant biological behavior, such as MVI, 19,22–26 worse tumor morphological grade, 19,22 and early recurrence after radical resection of HCC. Similarly, in previous studies based on GA-enhanced MRI, Dong et al 27 used three-dimensional quantitative analysis to conclude that tumor sphericity, non-smooth tumor margin, and peritumoral hypointensity on HBP images were significant predictors of MVI in patients with a single HCC≤3 cm, which was related to its early postoperative recurrence. Wei et al 28 also pointed out that HBP peritumoral hypointensity and satellite nodules can effectively predict early recurrence through a study on the risk stratification of early recurrence after curative resection of HCC. Interestingly, this is partly consistent with the results of this study, wherein we found that HBP peritumoral hypointensity was a useful predictor for the early recurrence of MVI-negative HCC, and this sign could be observed more frequently in the ER group. In view of the correlation between GA-enhanced MRI, an indirect molecular imaging method, and the molecular biology of tumor regulatory mechanisms, 21 we believed that this sign may reveal the malignant potential of tumors. However, in this study, the sensitivity of peritumoral hypointensity on HBP to predict early recurrence of MVI-negative HCC was relatively low. We speculated that this may be because the tumor was not accompanied by microvascular invasion and that changes in blood perfusion and hepatocyte function around the tumor were relatively small.
Studies have reported that poorly differentiated tumors have a negative impact on the recurrence risk and long-term survival of patients with HCC after radical liver resection. 29,30 Under the Edmondson-Steiner grading system, MVI-negative HCC with different degrees of differentiation have been confirmed to be significantly associated with poor postoperative outcomes. 14 This is consistent with our study where patients with poorly differentiated HCC appeared more prone to early recurrence than those with well-differentiated/moderately differentiated HCC. Studies have shown that HCC tissues with low expression of OATP1B3 potentially have an advanced TNM stage, a lower degree of tumor differentiation, and a higher risk of recurrence. 31,32 Based on this conclusion, which attracted us to further explore this biomarker, new ideas for individualized medicine in liver cancer treatment may germinate.
In the multivariate analysis, the quantitative imaging parameter RIR value in the hepatobiliary phase was negatively correlated with the postoperative ER; that is, with a decrease in the RIR value, the ER rate increased significantly. The rationale of GA was consistent with the findings of previous studies. Ye. et. Al 33 showed that HCC with a lower tumor-to-liver SI ratio on HBP exhibited significantly reduced OATP expression levels and worse prognosis, including higher aggressiveness, higher tumor grade, and shorter recurrence-free survival (RFS).
It has been reported that CK19-positive HCC, which tends to have a worse clinical prognosis, has a significantly lower tumor-to-liver SI ratio on HBP images than CK19-negative HCC in the hepatobiliary stage. 34 Another study showed that highly aggressive HCC with high Ki-67LI on HBP images tended to exhibit a lower tumor-to-liver contrast ratio. 33 We postulated that this quantitative index could establish a correlation between the imaging features and the biological behaviour of tumors. It may be possible to non-invasively and effectively predict tumor response to treatment and the likelihood of early recurrence. The latest study by Mulé et al 35 used the lesion-to-liver contrast enhancement ratio (LLCER), which could accurately predict MVI and recurrence-free survival after surgical resection. The emergence of some quantitative parameters may represent greater value in predicting early postoperative recurrence of HCC, which deserved future investigation.
A previous study showed that the tumor size (mean tumor diameter>4 cm in enrolled patients) can be used to predict early recurrence of HCC in an MVI-negative cohort. 15 However, our study did not find tumor size to be an independent risk factor for ER. This apparent difference may be due to the fact that the interquartile range of tumor size of the surgical candidates enrolled in this study was 1.5–3.95 cm. Future research should enroll more lesion samples of different sizes for a more detailed analysis to improve the accuracy of the results.
Our study showed that Model-2 can improve sensitivity, whereas it cannot simultaneously maintain high specificity. Combining the three findings increased the sensitivity from 71.4 to 96.4%, however, the accuracy decreased from 75.3 to 69.4%. This phenomenon may be mainly caused by the low specificity of differentiation. Nonetheless, Model-2 improved the sensitivity (96.4%) and negative predictive value (97%) of the diagnostic model, reducing the missed diagnosis rate.
The present study has certain limitations. First, this study has not been externally validated in different geographic regions and different aetiologies, and in this study we recruited patients with hepatitis B virus (HBV)-related liver disease, of whom approximately 94% had liver cirrhosis. Second, the sample size is relatively small such that specificity may mask the average characteristics, which can further affect the statistical results. Third, this was a single-centre retrospective study, and therefore the performance of the models needs to be prospectively evaluated for more frequent follow-up and adjuvant treatment of HCC.
In conclusion, the results of the present demonstrated that our predictive model incorporating postoperative Edmondson-Steiner grade and preoperative imaging features including peritumoral hypointensity on HBP and RIR on HBP (Model-2) represents a promising model to assess the risk of early recurrence after resection of MVI-negative HCC. This predictive model may help clinicians formulate more aggressive and personalized treatment plans way earlier to improve patient prognosis and reduce early recurrence.
Footnotes
Conflict of Interest: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.
Funding: The fifth Scientific Research Project of Nantong City (The “226 Project”, Grant No. 2020–14).
Ethical approval: The study was approved by the ethics committee of Affiliated Nantong Hospital 3 of Nantong University, and the requirement of informed consent was exempted.
The authors Qi Qu and Mengtian Lu contributed equally to the work.
Author contributions: Qi Qu: Conceptualization, Data curation, Methodology, Writing-original draft, Software, Investigation. Mengtian Lu: Data curation, Methodology, Software, Investigation. Lei Xu: Resources, Investigation, Validation. Jiyun Zhang: Visualization, Investigation. Maotong Liu: Formal analysis. Jifeng Jiang: Investigation. Xiance Zhao: Visualization. Xueqin Zhang: Conceptualization, Methodology, Project administration. Tao Zhang: Conceptualization, Methodology, Writing - review & editing, Funding acquisition.
Contributors: Qi Qu: Conceptualization, Data curation, Methodology, Writing-original draft, Software, Investigation. Mengtian Lu: Data curation, Methodology, Software, Investigation. Lei Xu: Resources, Investigation, Validation. Jiyun Zhang: Visualization, Investigation. Maotong Liu: Formal analysis. Jifeng Jiang: Investigation. Xiance Zhao: Visualization. Xueqin Zhang: Conceptualization, Methodology, Project administration. Tao Zhang: Conceptualization, Methodology, Writing - review & editing, Funding acquisition.
Contributor Information
Qi Qu, Email: 872788633@qq.com.
Mengtian Lu, Email: 2434133631@qq.com.
Lei Xu, Email: xuleicome@126.com.
Jiyun Zhang, Email: 734128513@qq.com.
Maotong Liu, Email: 270397421@qq.com.
Jifeng Jiang, Email: 8679636@qq.com.
Xiance Zhao, Email: shane.zhao@philips.com.
Xueqin Zhang, Email: 13962981245@163.com.
Tao Zhang, Email: 19931067@qq.com.
REFERENCES
- 1.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021; 71: 209–49. doi: 10.3322/caac.21660 [DOI] [PubMed] [Google Scholar]
- 2.Akoad ME, Pomfret EA. Surgical resection and liver transplantation for hepatocellular carcinoma. Clin Liver Dis 2015; 19: 381–99. doi: 10.1016/j.cld.2015.01.007 [DOI] [PubMed] [Google Scholar]
- 3.Rahbari NN, Mehrabi A, Mollberg NM, Müller SA, Koch M, Büchler MW, et al. Hepatocellular carcinoma: current management and perspectives for the future. Ann Surg 2011; 253: 453–69. doi: 10.1097/SLA.0b013e31820d944f [DOI] [PubMed] [Google Scholar]
- 4.Wang M-D, Li C, Liang L, Xing H, Sun L-Y, Quan B, et al. Early and late recurrence of hepatitis B virus-associated hepatocellular carcinoma. Oncologist 2020; 25: e1541–51. doi: 10.1634/theoncologist.2019-0944 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Li W-F, Yen Y-H, Liu Y-W, Wang C-C, Yong C-C, Lin C-C, et al. Preoperative predictors of early recurrence after resection for hepatocellular carcinoma. Am J Surg 2022; 223: 945–50. doi: 10.1016/j.amjsurg.2021.08.012 [DOI] [PubMed] [Google Scholar]
- 6.Zhang K, Tao C, Wu F, Siqin T, Wu J, Rong W. Establishment and evaluation of a predictive model for early postoperative recurrence of hepatocellular carcinoma in patients with microvascular invasion. Int J Gen Med 2021; 14: 2259–74. doi: 10.2147/IJGM.S303896 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yamamoto Y, Ikoma H, Morimura R, Konishi H, Murayama Y, Komatsu S, et al. Optimal duration of the early and late recurrence of hepatocellular carcinoma after hepatectomy. World J Gastroenterol 2015; 21: 1207–15. doi: 10.3748/wjg.v21.i4.1207 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zhu Y, Gu L, Chen T, Zheng G, Ye C, Jia W. Factors influencing early recurrence of hepatocellular carcinoma after curative resection. J Int Med Res 2020; 48: 0300060520945552. doi: 10.1177/0300060520945552 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chan AWH, Zhong J, Berhane S, Toyoda H, Cucchetti A, Shi K, et al. Development of pre and post-operative models to predict early recurrence of hepatocellular carcinoma after surgical resection. J Hepatol 2018; 69: 1284–93. doi: 10.1016/j.jhep.2018.08.027 [DOI] [PubMed] [Google Scholar]
- 10.Chen S-L, Xiao H, Xie Z-L, Shen J-X, Chen Z-B, Wang Y-Q, et al. The presence of microvascular invasion guides treatment strategy in recurrent HBV-related HCC. Eur Radiol 2020; 30: 3473–85. doi: 10.1007/s00330-019-06640-8 [DOI] [PubMed] [Google Scholar]
- 11.Çelebi F, Görmez A, Ilgun AS, Tokat Y, Balcı NC. The value of 18F-FDG PET/MRI in prediction of microvascular invasion in hepatocellular carcinoma. Eur J Radiol 2022; 149: S0720-048X(22)00046-8. doi: 10.1016/j.ejrad.2022.110196 [DOI] [PubMed] [Google Scholar]
- 12.Wang X, Zhang Z, Zhou X, Zhang Y, Zhou J, Tang S, et al. Computational quantitative measures of gd-EOB-DTPA enhanced MRI hepatobiliary phase images can predict microvascular invasion of small HCC. Eur J Radiol 2020; 133: S0720-048X(20)30550-7. doi: 10.1016/j.ejrad.2020.109361 [DOI] [PubMed] [Google Scholar]
- 13.Hui CL, Mautone M. Patterns of enhancement in the hepatobiliary phase of gadoxetic acid-enhanced MRI. Br J Radiol 2020; 93(1112): 20190989. doi: 10.1259/bjr.20190989 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhou L, Rui JA, Zhou WX, Wang SB, Chen SG, Qu Q. Edmondson-steiner grade: a crucial predictor of recurrence and survival in hepatocellular carcinoma without microvascular invasio. Pathol Res Pract 2017; 213: 824–30. doi: 10.1016/j.prp.2017.03.002 [DOI] [PubMed] [Google Scholar]
- 15.Wei Y, Pei W, Qin Y, Su D, Liao H. Preoperative MR imaging for predicting early recurrence of solitary hepatocellular carcinoma without microvascular invasion. Eur J Radiol 2021; 138: S0720-048X(21)00143-1. doi: 10.1016/j.ejrad.2021.109663 [DOI] [PubMed] [Google Scholar]
- 16.Kim HY, Choi JY, Kim CW, Bae SH, Yoon SK, Lee YJ, et al. Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid-enhanced magnetic resonance imaging predicts the histological grade of hepatocellular carcinoma only in patients with Child-Pugh class A cirrhosis. Liver Transpl 2012; 18: 850–57. doi: 10.1002/lt.23426 [DOI] [PubMed] [Google Scholar]
- 17.Kang H-J, Kim H, Lee DH, Hur BY, Hwang YJ, Suh K-S, et al. Gadoxetate-enhanced MRI features of proliferative hepatocellular carcinoma are prognostic after surgery. Radiology 2021; 300: 572–82. doi: 10.1148/radiol.2021204352 [DOI] [PubMed] [Google Scholar]
- 18.Lee S, Kim SH, Lee JE, Sinn DH, Park CK. Preoperative gadoxetic acid-enhanced MRI for predicting microvascular invasion in patients with single hepatocellular carcinoma. J Hepatol 2017; 67: 526–34. doi: 10.1016/j.jhep.2017.04.024 [DOI] [PubMed] [Google Scholar]
- 19.Lee S, Kim KW, Jeong WK, Kim M-J, Choi GH, Choi JS, et al. Gadoxetic acid-enhanced MRI as a predictor of recurrence of HCC after liver transplantation. Eur Radiol 2020; 30: 987–95. doi: 10.1007/s00330-019-06424-0 [DOI] [PubMed] [Google Scholar]
- 20.Zhou X, Long L, Mo Z, Li Y. Oatp1B3 expression in hepatocellular carcinoma correlates with intralesional gd-EOB-DTPA uptake and signal intensity on gd-EOB-DTPA-enhanced MRI. Cancer Manag Res 2021; 13: 1169–77. doi: 10.2147/CMAR.S292197 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kitao A, Matsui O, Yoneda N, Kozaka K, Kobayashi S, Koda W, et al. Gadoxetic acid-enhanced MR imaging for hepatocellular carcinoma: molecular and genetic background. Eur Radiol 2020; 30: 3438–47. doi: 10.1007/s00330-020-06687-y [DOI] [PubMed] [Google Scholar]
- 22.Ahn SJ, Kim JH, Park SJ, Kim ST, Han JK. Hepatocellular carcinoma: preoperative gadoxetic acid-enhanced MR imaging can predict early recurrence after curative resection using image features and texture analysis. Abdom Radiol (NY) 2019; 44: 539–48. doi: 10.1007/s00261-018-1768-9 [DOI] [PubMed] [Google Scholar]
- 23.Liang X, Shi S, Gao T. Preoperative gadoxetic acid-enhanced MRI predicts aggressive pathological features in LI-RADS category 5 hepatocellular carcinoma. Clin Radiol 2022; 77: 708–16. doi: 10.1016/j.crad.2022.05.018 [DOI] [PubMed] [Google Scholar]
- 24.Zhang L, Yu X, Wei W, Pan X, Lu L, Xia J, et al. Prediction of HCC microvascular invasion with gadobenate-enhanced MRI: correlation with pathology. Eur Radiol 2020; 30: 5327–36. doi: 10.1007/s00330-020-06895-6 [DOI] [PubMed] [Google Scholar]
- 25.Yang H, Han P, Huang M, Yue X, Wu L, Li X, et al. The role of gadoxetic acid-enhanced MRI features for predicting microvascular invasion in patients with hepatocellular carcinoma. Abdom Radiol (NY) 2022; 47: 948–56. doi: 10.1007/s00261-021-03392-2 [DOI] [PubMed] [Google Scholar]
- 26.Jiang H, Wei J, Fu F, Wei H, Qin Y, Duan T, et al. Predicting microvascular invasion in hepatocellular carcinoma: a dual-institution study on gadoxetate disodium-enhanced MRI. Liver Int 2022; 42: 1158–72. doi: 10.1111/liv.15231 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dong S-Y, Wang W-T, Chen X-S, Yang Y-T, Zhu S, Zeng M-S, et al. Microvascular invasion of small hepatocellular carcinoma can be preoperatively predicted by the 3D quantification of MRI. Eur Radiol 2022; 32: 4198–4209. doi: 10.1007/s00330-021-08495-4 [DOI] [PubMed] [Google Scholar]
- 28.Wei H, Jiang H, Zheng T, Zhang Z, Yang C, Ye Z, et al. LI-RADS category 5 hepatocellular carcinoma: preoperative gadoxetic acid-enhanced MRI for early recurrence risk stratification after curative resection. Eur Radiol 2021; 31: 2289–2302. doi: 10.1007/s00330-020-07303-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Shen J, Liu J, Li C, Wen T, Yan L, Yang J. The impact of tumor differentiation on the prognosis of HBV-associated solitary hepatocellular carcinoma following hepatectomy: a propensity score matching analysis. Dig Dis Sci 2018; 63: 1962–69. doi: 10.1007/s10620-018-5077-5 [DOI] [PubMed] [Google Scholar]
- 30.Chen W, Zhang Z, Fang X, Xiong L, Wen Y, Zhou J, et al. Prognostic value of the ALBI grade among patients with single hepatocellular carcinoma without macrovascular invasion. Medicine (Baltimore) 2021; 100(24): e26265. doi: 10.1097/MD.0000000000026265 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kitao A, Matsui O, Yoneda N, Kozaka K, Shinmura R, Koda W, et al. The uptake transporter OATP8 expression decreases during multistep hepatocarcinogenesis: correlation with gadoxetic acid enhanced MR imaging. Eur Radiol 2011; 21: 2056–66. doi: 10.1007/s00330-011-2165-8 [DOI] [PubMed] [Google Scholar]
- 32.Chen S, Li K, Jiang J, Wang X, Chai Y, Zhang C, et al. Low expression of organic anion-transporting polypeptide 1B3 predicts a poor prognosis in hepatocellular carcinoma. World J Surg Oncol 2020; 18: 127. doi: 10.1186/s12957-020-01891-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ye Z, Cao L, Wei Y, Chen J, Zhang Z, Yao S, et al. Preoperative prediction of hepatocellular carcinoma with highly aggressive characteristics using quantitative parameters derived from hepatobiliary phase MR images. Ann Transl Med 2020; 8: 85. doi: 10.21037/atm.2020.01.04 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Choi S-Y, Kim SH, Park CK, Min JH, Lee JE, Choi Y-H, et al. Imaging features of gadoxetic acid-enhanced and diffusion-weighted MR imaging for identifying cytokeratin 19-positive hepatocellular carcinoma: A retrospective observational study. Radiology 2018; 286: 897–908. doi: 10.1148/radiol.2017162846 [DOI] [PubMed] [Google Scholar]
- 35.Mulé S, Chalaye J, Legou F, Tenenhaus A, Calderaro J, Galletto Pregliasco A, et al. Hepatobiliary MR contrast agent uptake as a predictive biomarker of aggressive features on pathology and reduced recurrence-free survival in resectable hepatocellular carcinoma: comparison with dual-tracer 18F-FDG and 18F-FCH PET/CT. Eur Radiol 2020; 30: 5348–57. doi: 10.1007/s00330-020-06923-5 [DOI] [PubMed] [Google Scholar]