Prediction of transarterial chemoembolization refractoriness in patients with hepatocellular carcinoma using imaging features of gadoxetic acid-enhanced magnetic resonance imaging

Abstract

Background

Repeated transarterial chemoembolization (TACE) can be associated with loss of its efficacy and subsequent tumor progression.

Purpose

To identify features of gadoxetic acid-enhanced magnetic resonance imaging (MRI) associated with TACE refractoriness and to develop a prediction model for estimating the risk of TACE refractoriness.

Material and Methods

Among 1025 patients with intermediate-stage hepatocellular carcinoma (HCC) who underwent TACE as a first-line treatment during 2010–2017, 427 patients who underwent preoperative gadoxetic acid-enhanced MRI were analyzed. According to the date of initial TACE, patients were divided into the development cohort (n = 211) and the test cohort (n = 216). TACE refractoriness was determined according to the Japan Society of Hepatology guidelines. Univariable and multivariable analyses were performed to investigate the association between clinical/MRI features and TACE refractoriness. The performance of the prediction model was internally and externally assessed using the C-index of discrimination and a Hosmer-Lemeshow goodness-of-fit test for calibration.

Results

By analyzing 427 patients, we constructed a prediction model with the following independent features associated with TACE refractoriness: maximum tumor size; tumor number; peritumoral hypointensity on hepatobiliary phase (HBP); and the presence of non-hypervascular hypointense nodule on HBP. This system enabled the prediction of TACE refractoriness in the development cohort (C-index, 0.796) and the test cohort (C-index, 0.738) with good discrimination and calibration abilities.

Conclusion

The prediction model based on gadoxetic acid-enhanced MRI features in addition to the known predictors including tumor size and number can be used to estimate the risk of TACE refractoriness in patients with intermediate-stage HCC.

Keywords

Magnetic resonance imaging liver hepatocellular carcinoma transarterial chemoembolization outcome

Introduction

Transarterial chemoembolization (TACE) is the standard treatment for patients with intermediate-stage hepatocellular carcinoma (HCC) according to the Barcelona Clinic Liver Cancer (BCLC) staging system (1,2). However, the efficacy of TACE is reduced upon repeated use, and the rate of local recurrence after initial TACE may be as high as 80% (3). Repeated TACE is associated with deterioration of liver function (4,5). Due to hypoxia caused by TACE (6,7), recurrent tumors after TACE have a significantly shorter doubling time compared with primary HCCs (3). In clinical practice, TACE is applied not only in patients with intermediate-stage HCCs but also in those with advanced-stage HCCs as well as those who already show TACE refractoriness, due to unsatisfactory results from systemic therapy (8,9).

The recent introduction of multikinase inhibitors such as lenvatinib, regorafenib, cabozantinib, and immune checkpoint inhibitors has changed the treatment landscapes of HCC (9 –11). A recent ongoing prospective trial showed a possibility for a combination of TACE and systemic therapy in improving the outcomes of intermediate-stage HCC (12). Thus, it can be beneficial for patients with high risk of TACE refractoriness, if they are quickly identified so as to allow timely switching to combination therapy or systemic therapy. In addition, patients with intermediate-stage HCC (BCLC B) are highly heterogeneous in terms of tumor stages and prognosis (13,14). Previous attempts for developing a tailored subgroup stratification for the BCLC B stage include the Bolondi subclassification (13) and the Kinki criteria (14), which are mainly based on tumor factors such as tumor size and number, and liver function similar to the BCLC staging system.

Due to its excellent diagnostic performance, gadoxetic acid-enhanced liver magnetic resonance imaging (MRI) has been increasingly used to evaluate HCC (15,16). Gadoxetic acid provides information on not only hemodynamic profiles but also the biological behaviors of HCC by being taken up into hepatocytes via the organic-anion-transporting polypeptide B1 receptor during the hepatobiliary phase (HBP) (17 –21). Imaging features on gadoxetic acid-enhanced MRI provide additional useful information beyond conventional staging systems for predicting post-treatment outcomes mainly after curative treatments, such as radiofrequency ablation (RFA), surgery, or liver transplantation (22 –25). Thus, imaging features on gadoxetic acid-enhanced MRI has the potential to predict responses after TACE. To our knowledge, however, little has been known for its applicability to predict TACE refractoriness.

The aim of the present study was to identify the imaging features on gadoxetic acid-enhanced MRI associated with TACE refractoriness in patients with HCC and to develop an easily applicable prediction model for estimating the risk of TACE refractoriness.

Material and Methods

The present study was approved by the institutional review board of our center; the need for informed consent was waived due to the retrospective nature of the study. Analysis and reporting were carried out by adhering to the TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) guideline (26).

Study population

We reviewed the records of 4703 consecutive patients newly diagnosed with HCC and initially treated with TACE at our institution between January 2010 and December 2017. HCCs were diagnosed according to the guidelines of the American Association for the Study of Liver Diseases (AASLD) and the European Association for the Study of the Liver (EASL), or biopsy in cases with atypical imaging features (1,2). After identifying 1025 patients with BCLC B stage, we excluded the following: (i) those who did not undergo gadoxetic acid-enhanced MRI within one month before TACE (n = 535); (ii) those whose tumor response to TACE could not be assessed due to the lack of follow-up images or clinical information (n = 23); and (iii) those who were switched to or additionally underwent other treatment options for HCC (i.e. RFA, radiation therapy, surgery, or liver transplantation) immediately after TACE (n = 40). According to the date of the initial TACE, the remaining 427 patients were divided into the development cohort and the test cohort; the 211 patients who underwent TACE during 2010–2014 were allocated to the development cohort to derive the prediction model, and the 216 patients who underwent TACE during 2015–2017 were allocated to the test cohort (Fig. 1).

Fig. 1.

Flow chart of the study population. BCLC, the Barcelona Clinic Liver Cancer staging system; HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; TACE, tranarterial chemoembolization.

The clinical and demographic characteristics of the study population, such as age, sex, etiology of chronic liver disease, and laboratory data including total bilirubin, albumin, Child-Pugh class, and alpha-fetoprotein (AFP), were collected from electronic medical records.

TACE procedures

All TACE procedures were performed by highly experienced interventional radiologists. Before chemoembolization, superior mesenteric and common hepatic angiographies were performed to assess the vascular anatomy, tumor vascularity, and portal vein patency. In all patients, cisplatin-based TACE was performed. Detailed description of procedure is presented in the Supplementary Material and Methods.

Acquisition of gadoxetic acid-enhanced MRI

MRI examinations were performed using 1.5-T (Magnetom Avanto, Siemens Medical Solutions, Erlangen, Germany) or 3-T (Magnetom Skyra, Siemens Medical Solutions, Erlangen, Germany) systems. Detailed MRI techniques and parameters for MRI sequences are summarized in the Supplementary Materials and Methods and Supplementary Table 1. The average time interval between the acquisition of pre-treatment MRI and TACE for each patient was eight days (range = 1–30 days).

Pre-procedural MRI data analysis

All pre-procedural MRIs were retrospectively reviewed by two abdominal radiologists (with >6 years of experience in abdominal MRI interpretation) in consensus, who were blinded to the clinical characteristics and follow-up imaging results after TACE. The reviewers evaluated the following MRI features that potentially reflected pathologic characteristics of HCC (17,20,27 –29): (i) maximum tumor size; (ii) tumor number; (iii) morphology (nodular or multinodular vs. infiltrative); (iv) intratumoral fat (tumor area with decreased signal intensity on opposed-phase images compared with in-phase images (17)); (v) arterial peritumoral enhancement (detectable portion of crescent- or polygonal-shaped enhancement outside the tumor margin with broad contact with the tumor border on arterial phase, becoming isointense with background liver parenchyma on delayed phase (27)); (vi) enhancing capsule(peripheral rim of smooth hyperenhancement in the portal venous or transitional phase images (17,28)); (vii) peritumoral hypointensity on HBP (wedge-shaped or flame-like hypointense area of hepatic parenchyma located outside of the tumor margin on HBP images (20)); and (viii) presence of non-hypervascular hypointense nodules (NHHN; nodules showing unequivocal hypointensity on HBP images compared with the surrounding hepatic parenchyma without arterial phase hyperenhancement (29)).

Assessment of TACE refractoriness

Two abdominal radiologists that were not involved in pre-procedural MRI analyses evaluated all available follow-up CT or MRI scans performed 1–3 months after TACE to determine the radiological treatment responses of the HCCs. TACE refractoriness was defined according to the Japan Society of Hepatology (JSH) Consensus Guidelines (5): (i) two or more consecutive insufficient responses of the treated tumor (viable lesion >50%); (ii) more than two consecutive new intrahepatic lesions; (iii) continuous elevation of tumor markers immediately after TACE even with slight transient decrease; (iv) appearance of vascular invasion; and (v) appearance extrahepatic spread. For determining the elevation of tumor markers, we evaluated the level of AFP at baseline and 1–3 months after TACE and defined a continuous elevation of the tumor marker as an increase of >20% from baseline.

Statistical analysis

Continuous variables were expressed as mean ± SD or median (interquartile range); categorical variables were expressed as number or frequency. Data were analyzed with the Student’s t-test or the Mann–Whitney U test for continuous variables, and the chi-square test or Fisher’s exact test for categorical variables.

A logistic regression analysis was performed to assess the powers of clinical characteristics and MRI features for the prediction of TACE refractoriness. In the development cohort, univariable and multivariable analyses were performed to investigate the association between TACE refractoriness and various clinical and MRI factors. To develop a statistically robust scoring model predicting the refractoriness, we used a bootstrap resampling method. The variable selection process in the system was conducted as follows. First, we generated 1000 bootstrap resamples and a backward step-down variable selection procedure, based on a Wald chi-square statistic, was performed in the logistic regression model for each bootstrap sample. We then counted how many of each candidate variables remained in the 1000 bootstrap samples: if a variable appeared ≥500 times in the final model (i.e. bootstrap reliability criterion ≥50%), the variable was included in the scoring system. To allocate risk points in the scoring system, the bootstrapping method was used again to obtain bias-corrected regression coefficients. Risk points were obtained by bias-corrected regression coefficients multiplied by a reference value in the corresponding category. A reference risk factor profile was chosen by selecting a base category for each risk factor, which was assigned with 0 points in the scoring system. The total point scores based on patient profiles were converted into a risk estimate by using logit-transformation of the total risk score: 1/[1+exp(- $β_{0}$ -total score)], where $β_{0}$ is the regression coefficient estimate of the intercept. A shrinkage estimate (Nagelkerke Index, pseudo R2) was used to quantify over-fitting. The prediction model was internally and externally validated by measuring its discrimination and calibration abilities by the c-indices and the Hosmer-Lemeshow goodness-of-fit tests. Further details on the development of the scoring systems have been described by Sullivan et al. (30).

P values < 0.05 were considered statistically significant. Statistical analyses were performed using SAS software version 9.1.3 (SAS Inc., Cary, NC, USA) and R 2.8.1 (The R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics of the development and test cohorts

During the study period, 4703 patients with HCC were treated with TACE at our institution; after applying the exclusion criteria, 427 patients (mean age = 59.3 ± 9.7 years) were finally selected for analysis. The baseline clinical and MRI features of the development (n = 211) and the test cohort (n = 216) are summarized in Table 1. The development cohort consisted of 133 (63.0%) TACE-responsive patients and 78 (37.0%) TACE-refractory patients, and the test cohort consisted of 136 (63.0%) TACE-responsive patients and 80 (37.0%) TACE-refractory patients. The prevalence of TACE refractoriness was not significantly different between the two cohorts (P = 0.988). The median levels of bilirubin and albumin were significantly higher in the development cohort (P < 0.001 for both). The prevalence of enhancing capsule on MRI was higher in the test cohort (P = 0.001). Other characteristics showed similar results between the two cohorts.

Table 1.

Baseline characteristics of the study population.

Characteristics	Overall cohort (n = 427)	Development cohort (n = 211)	Test cohort (n = 216)	P value
Clinical characteristics
Age (years)	59.3 ± 9.7	58.6 ± 9.7	60.0 ± 9.6	0.127
Sex				0.274
Male	375 (87.8)	189 (89.6)	186 (86.1)
Female	52 (12.2)	22 (10.4)	30 (13.9)
Etiology				0.838
HBV	326 (76.3)	160 (75.8)	166 (76.9)
HCV	40 (9.4)	23 (57.5)	17 (7.9)
Alcoholic	33 (7.7)	15 (7.1)	18 (8.3)
Unknown	28 (6.6)	13 (6.2)	15 (3.6)
Bilirubin (ng/mL)	0.8 (0.5–1.1)	0.9 (0.6–1.1)	0.7 (0.5–0.9)	<0.001
Albumin (ng/mL)	3.7 ± 0.5	3.8 ± 0.5	3.6 ± 0.5	0.001
Child-Pugh class				0.371
A	385 (90.2)	193 (91.5)	192 (88.9)
B	42 (9.8)	18 (8.5)	24 (11.1)
AFP (ng/mL)				0.877
<200	271 (63.6)	135 (64.0)	136 (63.3)
≥200	155 (36.4)	76 (36.0)	79 (36.7)
TACE response				0.988
Responsive	269 (63.0)	133 (63.0)	136 (63.0)
Refractory	158 (37.0)	78 (37.0)	80 (37.0)
MRI findings
Maximum tumor size (cm)				0.264
<7	301 (70.5)	154 (73.0)	147 (68.1)
≥7	126 (29.5)	57 (27.0)	69 (31.9)
Number of tumors				0.097
2–10	370 (86.7)	177 (83.9)	193 (89.4)
≥11	57 (13.4)	34 (16.1)	23 (10.7)
Morphology				0.936
Nodular or multinodular	384 (89.9)	190 (90.1)	194 (89.8)
Infiltrative	43 (10.1)	21 (10.0)	22 (10.2)
Intratumoral fat				0.213
Absent	350 (82.0)	168 (79.6)	182 (84.3)
Present	77 (18.0)	43 (20.4)	34 (15.7)
Arterial peritumoral enhancement				0.537
Absent	225 (52.7)	108 (51.2)	117 (54.2)
Present	202 (47.3)	103 (48.8)	99 (45.8)
Enhancing capsule				0.001
Absent	244 (57.1)	104 (49.3)	140 (64.8)
Present	183 (42.9)	107 (50.7)	76 (35.2)
Peritumoral hypointensity on HBP images				0.911
Absent	244 (57.1)	120 (56.9)	124 (57.4)
Present	183 (42.9)	91 (43.1)	92 (42.6)
NHHN				0.103
Absent	297 (69.6)	139 (5.9)	158 (73.2)
Present	130 (30.4)	72 (34.1)	58 (26.9)

Values are given as n (%), mean ± SD, or median (IQR).

AFP, alpha-fetoprotein; HBP, hepatobiliary phase; HBV, hepatitis B virus; HCV, hepatitis C virus; MRI, magnetic resonance imaging; NHHN, non-hypervascular hypointense nodule; TACE, transarterial chemoembolization.

Development of the predictive scoring system

In the development cohort, univariable analysis showed that AFP ≥ 200 (odds ratio [OR] = 2.375; P = 0.004), maximum tumor size ≥7 cm (OR = 5.099; P < 0.001), tumor number ≥1 (OR = 3.961; P < 0.001), infiltrative morphology (OR = 3.125; P = 0.016), arterial peritumoral enhancement (OR = 3.216; P < 0.001), peritumoral hypointensity on HBP (OR = 5.3; P < 0.001), and the presence of NHHN (OR = 2.527; P = 0.002) on HBP were significantly associated with TACE refractoriness (Table 2).

Table 2.

Clinical features and pre-procedural gadoxetic acid-enhanced MRI findings in the development cohort and the results of univariate analysis for factors associated with TACE refractoriness.

	TACE-responsive (n = 133)	TACE-refractory (n = 78)	OR (95% CI)	P value
Clinical characteristics
Age (years)	59.5 ± 9.1	57.0 ± 10.4	0.973 (0.945–1.002)	0.069
Sex			1.483 (0.609–3.612)	0.386
Male	121 (64.0)	68 (36.0)
Female	12 (54.6)	10 (45.5)
Child-Pugh class			2.298 (0.866–6.095)	0.095
A	125 (64.8)	68 (35.2)
B	8 (44.4)	10 (55.6)
AFP (ng/mL)			2.375 (1.327–4.250)	0.004
<200	95 (70.4)	40 (29.6)
≥200	38 (50.0)	38 (50.0)
MRI findings
Tumor size (cm)
<7	113 (73.4)	41 (26.6)	5.099 (2.660–9.775)	<0.001
≥7	20 (35.1)	37 (64.9)
Number of tumors			3.961 (1.832–8.568)	<0.001
2–10	121 (68.4)	56 (31.6)
≥11	12 (35.3)	22 (64.7)
Morphology			3.125 (1.233–7.923)	0.016
Nodular or multinodular	125 (65.8)	65 (34.2)
Infiltrative	8 (38.1)	13 (61.9)
Intratumoral fat			1.859 (0.944–3.661)	0.073
Absent	111 (66.1)	57 (33.9)
Present	22 (51.2)	21 (48.8)
Arterial peritumoral enhancement			3.216 (1.789–5.780)	<0.001
Absent	82 (75.9)	26 (24.1)
Present	51 (49.5)	52 (50.5)
Enhancing capsule			0.690 (0.393–1.209)	0.195
Absent	61 (58.7)	43 (41.4)
Present	72 (67.3)	35 (32.7)
Peritumoral hypointensity on HBP images			5.300 (2.890–9.719)	<0.001
Absent	95 (79.2)	25 (20.8)
Present	38 (41.8)	53 (58.2)
NHHN			2.527 (1.403–4.552)	0.002
Absent	98 (70.5)	41 (29.5)
Present	35 (48.6)	37 (51.4)

Values are given as n (%) or mean ± SD unless otherwise specified.

AFP, alpha-fetoprotein; CI, confidence interval; HBP, hepatobiliary phase; MRI, magnetic resonance imaging; NHHN, non-hypervascular hypointense nodule; OR, odds ratio; TACE, transarterial chemoembolization.

The following four variables met the bootstrap reliability criterion of ≥50%: maximum tumor size ≥7 cm; tumor number ≥11; peritumoral hypointensity on HBP; and the presence of NHHN on HBP (Supplementary Table 2). A multivariable model for predicting TACE refractoriness was developed with these four variables, and a scoring point was derived on the basis of the adjusted coefficients of the predictors in the final model. Each risk point was rescaled to designate the presence of NHHN as 1 (Table 3). Fig. 2a shows the nomogram for prediction of TACE refractoriness; the total score was in the range of 0–5 points, with a higher score reflecting a higher probability of TACE refractoriness. The highest predicted risk of being TACE refractoriness was 90.2% with a score of 5 if the following predictors were present: maximum tumor size ≥7 cm; tumor number ≥11; peritumoral hypointensity on HBP; and the presence of the NHHN on HBP (Fig. 3).

Table 3.

Predictors selected by multivariate logistic regression model for TACE refractoriness.

	β*	OR (95% CI)	P value	Points scored^†
Intercept	–2.078
Maximum tumor size (cm)
<7		1
≥7	1.296	3.636 (1.775–7.448)	<0.001	1.5
Number of tumors
2–10		1
≥11	1.101	3.397 (1.427–8.087)	0.006	1
Peritumoral hypointensity on HBP
Absent		1
Present	1.399	4.187 (2.167–8.090)	<0.001	1.5
NHHN
Absent		1
Present	0.861	2.397 (1.215–4.728)	0.012	1

*These are shrunk coefficients (by factor 0.935) to correct for over-fitting.

^†The point for each predictor was determined on the basis of the β regression coefficient of each predictor.

CI, confidence interval; HBP, hepatobiliary phase; NHHN, non-hypervascular hypointense nodule; OR, odds ratio; TACE, transarterial chemoembolization.

Fig. 2.

Prediction model for TACE refractoriness. (a) Nomogram for predicting the probability of TACE. The normogram was developed in the development cohort refractoriness based on clinical and pre-treatment MRI findings. Top, predictor points are found on the uppermost point scale that corresponds to each variable. Bottom, points of all variables are added up, and the total point projected on the bottom scale indicates the probability of TACE-refractory HCC. (b–e) ROC curve (b) and calibration curve (c) for TACE refractoriness prediction model in the development cohort show good discrimination with a C-index of 0.796 and adequate fitting and reliability. In the validation cohort, a C-index of ROC analysis (d) was 0.738. Calibration curve (e) yields good agreement between predicted and actual probability of TACE refractoriness. HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; ROC, receiver operating characteristic; TACE, tranarterial chemoembolization.

Fig. 3.

A 67-year-old man with three HCCs. (a) Arterial phase images of gadoxetic acid-enhanced MRI show a large HCC (arrows) with a maximal diameter of 8.5 cm (1.5 points) in the right lobe of the liver. (b) In the right hepatic dome, a 1-cm-sized non-hypervascular nodule (arrowhead) is seen adjacent to another 1.2-cm-sized HCC (arrow). (c) Hepatobiliary phase image reveals peritumoral hypointensity (arrowheads, 1.5 points) around the large HCC. (d) On hepatobiliary phase imaging, two non-hypervascular hypointense nodules (arrowheads, 1 point) are present adjacent to the HCC (arrow) in the right hepatic dome. According to the system developed in the present study, the total score was 4 points with an estimated risk for TACE refractoriness of 79.7%. Follow-up CT (e–g) after two sessions of TACE showing residual viable tumor around the lipodol-laden HCC (arrows, e) with rapidly progressing intrahepatic metastases (arrowheads, e), mediastinal metastatic lymph nodes (arrows, f), bone metastasis with extraosseous soft-tissue mass (arrowheads, f) and lung metastasis (arrow, g). CT, computed tomography; HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; TACE, tranarterial chemoembolization.

Validation of the predictive scoring system

In the internal validation analysis, the final prediction model had a bias-corrected C-index of 0.796 (95% confidence interval [CI] = 0.733–0.860; Fig 2b). The calibration curve for the predicted and actual probabilities of TACE refractoriness yielded good calibration abilities based on the results of the Homer-Lemeshow test (χ² = 1.031; df = 2; P = 0.597; Fig 2c). The external validation analysis confirmed that the prediction model well-performed in terms of discrimination (C-index = 0.738; 95% CI = 0.670–0.807; Fig. 2d) and calibration (Hosmer-Lemeshow goodness-of-fit test, χ² = 2.817; df = 2; P = 0.245; Fig 2e).

Discussion

Our results show that gadoxetic acid-enhanced MRI features provide additional information to the conventional staging system in predicting TACE refractoriness in patients with HCC. In addition to the known predictors, including maximum tumor size and tumor number, multivariable analysis showed that peritumoral hypointensity and the presence of NHHN on HBP on gadoxetic acid-enhanced MRI were associated with TACE refractoriness. We developed a prediction model derived from these independent predictors, which showed good performances in the development cohort (C-index = 0.796) and was well-validated in the temporally distinct test cohort (C-index = 0.738).

The maximum tumor size and tumor number are well-validated predictors of early HCC recurrence after TACE (31 –33). In line with these studies, we also found that increased tumor burden was significantly related to a higher risk of TACE refractoriness. Due to the heterogeneity of intermediate-stage HCC with respect to tumor burden, previous attempts for further stratification of such patients including the Bolondi subclassification (13) and the Kinki criteria (14) also included tumor size and tumor number as important factors. Our results support these modified systems for selecting a subset of patients with higher risk of TACE refractoriness in whom alternative remedies may be needed.

The present study unveiled the potential of tailored staging using imaging features on gadoxetic acid-enhanced MRI including peritumoral hypointensity and the presence of NHHN on HBP. Among the clinical and imaging features investigated in the study, peritumoral hypointensity on HBP showed the highest OR of 4.187 and was allocated 1.5 points in the scoring system for predicting TACE refractoriness. The clinical implications of peritumoral hypointensity on HBP has been actively studied: recent studies showed that peritumoral hypointensity on HBP is strongly associated with the presence of microvascular invasion in pathologic analysis (20,24), which is a strong predictor of early recurrence after curative hepatectomy or RFA (34,35) as well as TACE-refractory HCC (36). Although several studies showed that the presence of peritumoral hypointensity on HBP can predict early recurrence after curative treatment (23,24,37), only few studies focused on its clinical implications following TACE. Our results suggest that the peritumoral hypointensity on HBP can also be considered as a predictive feature of poor treatment outcomes after TACE. As the use of gadoxetic acid-enhanced MRI in evaluating HCC expands, NHHN on HBP continues to gain attention. NHHN may be considered as a borderline lesion that may transition into HCC (17) and ∼30% of NHHNs eventually transform into hypervascular HCC (38). Thus, the presence of NHHN is considered to indicate enhanced hepatocarcinogenic potential in background liver (29). The concurrence of NHHN on HBP is a significant risk factor for the recurrence of HCC after hepatectomy and RFA, especially for multicentric recurrences (23,29,39). Our results show that NHHN is also a prognostic factor for poor outcomes after TACE. We assume that NHHN has low responsiveness to TACE due to its non-hypervascular nature and the high risk of transformation into overt HCC after TACE.

By using the aforementioned tumor-related factors and imaging features from gadoxetic acid-enhanced MRI, we developed a scoring system that estimates the risk for TACE refractoriness with good discrimination and calibration, and validated its performance on a separate test cohort. Each total score presents the probability of TACE refractoriness and supposedly provides reliable guidance for predicting TACE refractoriness. Gadoxetic acid-enhanced MRI is expected to improve the accuracy of conventional staging systems based on tumor size and number by adding new information on tumor biology. In line with our results, a recent study demonstrated that additional gadoxetic acid-enhanced MRI before HCC treatment was associated with better survival by assisting patient selection (40). Our new staging system for intermediate-stage HCC may be useful for improving patient selection as well by identifying those at risk for TACE refractoriness.

The present study has some limitations. First, there may have been a potential selection bias due to the retrospective nature of this study. Approximately half of the initial eligible patients were excluded from the study as they did not undergo pre-procedural gadoxetic acid-enhanced MRI. This reflects the current practice pattern of pre-procedural imaging in which gadoxetic acid-enhanced MRI is more frequently performed before curative treatments than before TACE. Our results suggest that gadoxetic acid-enhanced MRI may be encouraged before determining treatment options, not only due to its excellent diagnostic performance but also its ability for better patient selection and prognostic implications. Second, although we used robust statistical methods to enhance the reliability of our results, the present study was performed using data from a single center. Therefore, our results should be further validated through prospective studies in multiple institutions.

In conclusion, we developed a simple and easily applicable scoring system based on gadoxetic acid-enhanced pre-procedural MRI for estimating the risk of TACE refractoriness in patients with HCC. By combining conventional staging information with detailed information from gadoxetic acid-enhanced MRI, HCC patients at risk for TACE refractoriness may be identified before TACE and allocated to optimal treatment options.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

So Yeon Kim

Seung Soo Lee

References

Bruix

Sherman

Management of hepatocellular carcinoma: an update. Hepatology 2011; 53:1020–1022.

Bruix

Sherman

Llovet

, et al. Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. J Hepatol 2001; 35:421–430.

Tezuka

Hayashi

Kubota

, et al. Growth rate of locally recurrent hepatocellular carcinoma after transcatheter arterial chemoembolization: comparing the growth rate of locally recurrent tumor with that of primary hepatocellular carcinoma. Dig Dis Sci 2007; 52:783–788.

Raoul

J-L

Gilabert

Piana

How to define transarterial chemoembolization failure or refractoriness: a European perspective. Liver cancer 2014; 3:119–124.

Kudo

Matsui

Izumi

, et al. Transarterial chemoembolization failure/refractoriness: JSH-LCSGJ criteria 2014 update. Oncology 2014; 87 (Suppl. 1):22–31.

Xie

Chen

Hypoxia and hepatocellular carcinoma: The therapeutic target for hepatocellular carcinoma. J Gastroenterol Hepatol 2007; 22:1178–1182.

Lai

Conley

Knudsen

, et al. Hypoxia after transarterial chemoembolization may trigger a progenitor cell phenotype in hepatocellular carcinoma. Histopathology 2015; 67:442–450.

Park

Amarapurkar

Chao

, et al. Consensus recommendations and review by an International Expert Panel on Interventions in Hepatocellular Carcinoma (EPOIHCC). Liver Int 2013; 33:327–337.

Kudo

Lenvatinib may drastically change the treatment landscape of hepatocellular carcinoma. Liver Cancer 2018; 7:1–19.

10.

Llovet

Ricci

Mazzaferro

, et al. Sorafenib in advanced hepatocellular carcinoma. N Engl J Med 2008; 359:378–390.

11.

Cheng

Kang

Chen

, et al. Efficacy and safety of sorafenib in patients in the Asia-Pacific region with advanced hepatocellular carcinoma: a phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol 2009; 10:25–34.

12.

Kudo

Ueshima

Ikeda

, et al. Randomized, open label, multicenter, phase II trial comparing transarterial chemoembolization (TACE) plus sorafenib with TACE alone in patients with hepatocellular carcinoma (HCC): TACTICS trial. J Clin Oncol 2018; 36:206–206.

13.

Bolondi

Burroughs

Dufour

, et al. Heterogeneity of patients with intermediate (BCLC B) Hepatocellular Carcinoma: proposal for a subclassification to facilitate treatment decisions. Semin Liver Dis 2012; 32:348–359.

14.

Kudo

Arizumi

Ueshima

, et al. Subclassification of BCLC B stage hepatocellular carcinoma and treatment strategies: Proposal of modified Bolondi’s subclassification (Kinki Criteria). Dig Dis 2015; 33:751–758.

15.

Lee

, et al. Hepatocellular carcinoma: diagnostic performance of multidetector CT and MR imaging-a systematic review and meta-analysis. Radiology 2015; 275:97–109.

16.

Hanna

Miloushev

Tang

, et al. Comparative 13-year meta-analysis of the sensitivity and positive predictive value of ultrasound, CT, and MRI for detecting hepatocellular carcinoma. Abdom Radiol (NY) 2016; 41:71–90.

17.

Choi

Lee

Sirlin

CB.

CT and MR imaging diagnosis and staging of hepatocellular carcinoma: part II. Extracellular agents, hepatobiliary agents, and ancillary imaging features. Radiology 2014; 273:30–50.

18.

Choi

Kim

Park

, 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.

19.

Yamashita

Kitao

Matsui

, et al. Gd-EOB-DTPA-enhanced magnetic resonance imaging and alpha-fetoprotein predict prognosis of early-stage hepatocellular carcinoma. Hepatology 2014; 60:1674–1685.

20.

Kim

Jeon

, et al. Prediction of microvascular invasion of hepatocellular carcinoma: usefulness of peritumoral hypointensity seen on gadoxetate disodium-enhanced hepatobiliary phase images. J Magn Reson Imaging 2012; 35:629–634.

21.

Kitao

Matsui

Yoneda

, et al. Hypervascular hepatocellular carcinoma: correlation between biologic features and signal intensity on gadoxetic acid-enhanced MR images. Radiology 2012; 265:780–789.

22.

Kim

Park

, et al. Single hepatocellular carcinoma: preoperative MR imaging to predict early recurrence after curative resection. Radiology 2015; 276:433–443.

23.

Lee

, et al. Non-hypervascular hepatobiliary phase hypointense nodules on gadoxetic acid-enhanced MRI: risk of HCC recurrence after radiofrequency ablation. J Hepatol 2015; 62:1122–1130.

24.

Lee

Kim

Lee

, et al. Preoperative gadoxetic acid-enhanced MRI for predicting microvascular invasion in patients with single hepatocellular carcinoma. J Hepatol 2017; 67:526–534.

25.

Kim

Sinn

Jeong

, et al. Hepatobiliary MRI as novel selection criteria in liver transplantation for hepatocellular carcinoma. J Hepatol 2018; 68:1144–1152.

26.

Moons

Altman

Reitsma

, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015; 162:W1–73.

27.

Kim

Park

Choi

, et al. Can microvessel invasion of hepatocellular carcinoma be predicted by pre-operative MRI? Eur Radiol 2009; 19:1744–1751.

28.

Marrero

Kulik

Sirlin

, 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.

29.

Toyoda

Kumada

Tada

, et al. Non-hypervascular hypointense nodules detected by Gd-EOB-DTPA-enhanced MRI are a risk factor for recurrence of HCC after hepatectomy. J Hepatol 2013; 58:1174–1180.

30.

Sullivan

Massaro

D’Agostino

Sr.

Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Stat Med 2004; 23:1631–1660.

31.

Douhara

Namisaki

Moriya

, et al. Predisposing factors for hepatocellular carcinoma recurrence following initial remission after transcatheter arterial chemoembolization. Oncol Lett 2017; 14:3028–3034.

32.

Rou

Lee

Moon

, et al. Risk factors and therapeutic results of early local recurrence after transcatheter arterial chemoembolization. World J Gastroenterol 2014; 20:6995–7004.

33.

Kim

Park

Joo

, et al. Severity and timing of progression predict refractoriness to transarterial chemoembolization in hepatocellular carcinoma. J Gastroenterol Hepatol 2012; 27:1051–1056.

34.

Lauwers

Terris

Balis

, et al. Prognostic histologic indicators of curatively resected hepatocellular carcinomas: a multi-institutional analysis of 425 patients with definition of a histologic prognostic index. Am J Surg Pathol 2002; 26:25–34.

35.

Sumie

Kuromatsu

Okuda

, et al. Microvascular invasion in patients with hepatocellular carcinoma and its predictable clinicopathological factors. Ann Surg Oncol 2008; 15:1375–1382.

36.

Kim

Nam

Cho

, et al. Plasma micoRNA-122 as a predictive marker for treatment response following transarterial chemoembolization in patients with hepatocellular carcinoma. J Gastroenterol Hepatol 2017; 32:199–207.

37.

Lee

Kang

Song

, et al. Effect of microvascular invasion risk on early recurrence of hepatocellular carcinoma after surgery and radiofrequency ablation. Ann Surg 2019. DOI: 10.1097/SLA.0000000000003268.

38.

Suh

Kim

Pyo

, et al. Hypervascular transformation of hypovascular hypointense nodules in the hepatobiliary phase of gadoxetic acid–enhanced MRI: a systematic review and meta-analysis. Am J Roentgenol 2017; 209:781–789.

39.

Toyoda

Kumada

Tada

, et al. Non-hypervascular hypointense nodules on Gd-EOB-DTPA-enhanced MRI as a predictor of outcomes for early-stage HCC. Hepatol Int 2015; 9:84–92.

40.

Kim

Lim

Han

, 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.