Jin-Shu Zeng ,Jian-Xing Zeng ,Yao Huang ,Jing-Feng Liu ,Jin-Hua Zeng ,?

a Department of Ultrasonic Medical, The First Affiliated Hospital of Fujian Medical University, Fuzhou 350 0 05, China

b The Liver Center of Fujian Province, Fujian Medical University, Fuzhou 350025, China

c Department of Hepatobiliary Surgery, The First Affiliated Hospital of Fujian Medical University, Fuzhou 350 0 05, China

d Department of Hepatobiliary Surgery, Mengchao Hepatobiliary Hospital of Fujian Medical University, Fuzhou 350 0 05, China

Keywords: Hepatocellular carcinoma Liver resection Adjuvant transarterial chemoembolization Scoring system Risk stratification

ABSTRACT Background: There is currently no standard adjuvant treatment proven to prevent hepatocellular carcinoma (HCC) recurrence. Recent studies suggest that postoperative adjuvant transarterial chemoembolization (PA-TACE) is beneficial for patients at high risk of tumor recurrence. However,it is difficult to select the patients. The present study aimed to develop an easy-to-use score to identify these patients.Methods: A total of 4530 patients undergoing liver resection were recruited. Independent risk factors were identified by Cox regression model in the training cohort and the Primary liver cancer big data transarterial chemoembolization (PDTE) scoring system was established.Results: The scoring system was composed of ten risk factors including alpha-fetoprotein (AFP),albuminbilirubin (ALBI) grade,operative bleeding loss,resection margin,tumor capsular,satellite nodules,tumor size and number,and microvascular and macrovascular invasion. Using 5 points as risk stratification,the patients with PA-TACE had higher recurrence-free survival (RFS) compared with non-TACE in > 5 points group ( P < 0.001),whereas PA-TACE patients had lower RFS compared with non-TACE in ≤5 points group ( P = 0.013). In the training and validation cohorts,the C-indexes of PDTE scoring system were 0.714 [standard errors (SE) = 0.010] and 0.716 (SE = 0.018),respectively.Conclusions: The model is a simple tool to identify PA-TACE for HCC patients after liver resection with a favorable performance. Patients with > 5 points may benefit from PA-TACE.

Introduction

Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer-related death worldwide [1] . Currently,liver resection is still the most effective curative treatment for HCC. However,the prognosis remains poor and unsatisfactory because of the high incidence of tumor recurrence,which is reported to be more than 60% within 5 years [2] . Despite this high rate of tumor recurrence,no anti-recurrence adjuvant therapy is recommended because there are no universally accepted adjuvant therapies for HCC after liver resection [ 3,4 ].

Transarterial chemoembolization (TACE) is recommended for Barcelona Clinic Liver Cancer intermediate staging (BCLC B staging) patients [3] . Currently,postoperative adjuvant transarterial chemoembolization (PA-TACE) as an adjuvant therapy to prevent tumor recurrence is getting increasing attention [ 5,6 ]. Recent studies have shown that PA-TACE is beneficial for patients at high risk of tumor recurrence [ 7,8 ],such as large tumor size [ 9,10 ],multiple tumor number [ 11,12 ],nonencapsulated tumors [ 13,14 ],microvascular invasion [15–17],or portal vein tumor thrombosis [18–20],but it is difficult to define those who may benefit from PA-TACE.

Besides,there is currently no reliable tool for risk stratification of adjuvant therapy,which may lead to failure of adjuvant therapy because of suboptimal patient selection. Therefore,the identification of patients after liver resection who are at high risk of tumor recurrence is important to standardize PA-TACE and improve the efficacy of adjuvant therapy.

In this study,we employed a large-scale multicenter cohort to develop and validate a simple score (Primary liver cancer big data transarterial chemoembolization,PDTE) to identify a subgroup of patients who may benefit from PA-TACE based on risk stratification.

Methods

Patients

HCC patients who underwent liver resection between January 2009 and December 2017 were extracted from a multicenter database (primary liver cancer big data,PLCBD) including Mengchao Hepatobiliary Hospital of Fujian Medical University,Eastern Hepatobiliary Surgery Hospital and the First Affiliated Hospital of Fujian Medical University. HCC patients recruited between 2009 and 2015 were randomly divided into the training and validation cohorts in a 3:1 ratio. HCC patients recruited between 2016 and 2017 served as another validation cohort. Liver resection was considered when all tumor nodules on preoperative imaging studies could technically be resected within liver functional reserve and negative resection margins (R0 resection). This study complied with the standards oftheHelsinkiDeclarationand was approved by the Institutional Ethics Committees of all study centers.

The inclusion criteria were 1) HCC diagnosed by pathology and 2) tumor negative resection margins,defined as complete resection of macroscopic tumor nodules with tumor-free margins confirmed by histological examination. Patients who had 1) extrahepatic metastasis,2) preoperative anticancer treatments,3) perioperative death,4) history of other malignancies,5) lost to follow up within 60 days after surgery,6) recurrence within 60 days after surgery and 7) incomplete clinical data were excluded from this study.

Clinicopathologic variables

Blood samples were obtained within 14 days before surgery for routine laboratory tests,including liver function,blood cells,total bilirubin,albumin,platelet count,alpha-fetoprotein (AFP),HBV and hepatitis C virus immunology,and HBV-DNA load. The albuminbilirubin (ALBI) grade was calculated by the formula,0.66 × log10(bilirubin,μmol/L) -0.085 × (albumin,g/L). According to a previously described cut-off,patients were stratified into three grades:ALBI grade 1 ( ≤-2.63),grade 2 (>-2.63 to -1.39) and grade 3 (>-1.39). ALBI grade 2 and ALBI grade 3 were grouped together due to the low sample size in the latter. Patient baseline characteristics including demographic information,surgical factors,laboratory parameters and tumor characteristics were obtained from postoperative pathological reports. The pathological reviews of all specimens were examined independently by two pathologists. Satellite nodules were defined as tumor cell nests on microscopy or their sizes were less than 2 cm on macroscopy presenting within 2 cm of the main tumor.

Follow-up

Follow-up was performed every 3 months in the first 2 years,and every 6 months during subsequent years. The follow-up examination included liver function,serum AFP,and at least one abdominal imaging scan,including abdominal ultrasonography,abdominal computed tomography (CT),or magnetic resonance imaging (MRI).

The diagnosis and management of tumor recurrence were relied on the evidence of imaging findings according to the current guidelines [4] . The recurrence-free survival (RFS) was defined as the interval between the date of surgery and the date of recurrence or death or last follow-up. The study was censored on 31st December 2020.

PA-TACE

PA-TACE was performed once within 1–2 months after resection without recurrence. Concisely,catheterization was placed into the proper hepatic artery through the femoral artery using the Seldinger technique,and chemotherapeutic agents including cisplatin (10–30 mg),doxorubicin hydrochloride (10 mg) or pharmorubicin (20–40 mg) were slowly injected through the catheter followed by an emulsion of lipiodol (2–10 mL). The dosage of chemotherapeutic agents and lipiodol was determined by body surface area and liver function [21] .

Statistical analysis

Continuous variables were reported as mean ± standard deviation (SD) and compared by Student’st-test or median (interquartile range,IQR) and compared by Mann-WhitneyUtest. Categorical variables were presented as number (percentage) and compared using Chi-square test or Fisher’s exact test.

Univariate and multivariate Cox proportional hazard regression analyses were performed to select the independent factors of RFS in this model. All factors withP<0.05 in univariate Cox regression were selected into multivariate Cox regression. The multivariate Cox regression was performed by stepwise backward selection to detect independent risk factors. The regression coefficients of the multivariate Cox regression model were multiplied by 3 and rounded to the nearest unit to obtain simple point numbers,facilitating the bedside calculation of the PDTE score. Kaplan-Meier curves were used to estimate RFS rates,and the difference between the two groups was analyzed by log-rank test.

Model performance was assessed by C-index,G?nen & Heller’s K,and time-dependent areas under the receiver operating characteristic curve (tdAUC) [22] . Model calibration was measured by the calibration curve. The PDTE model was also compared to early recurrence after surgery for liver tumor (ERASL) model [23],Korean model [24],8th TNM staging system [25],BCLC staging system [3],and China liver cancer (CNLC) staging system [26] .

All statistical tests were two-tailed and aPvalue<0.05 was considered statistically significant. Statistical analysis was performed with R version 3.5.2 ( http://www.r-project.org/ ). The R packages of “Table 1 ”,“rms”,“CPE”,“timeROC”,“stdca”,“survminer” and “survival” were used.

Results

Baseline characteristics of patients

During the study period,a total of 4530 HCC patients after surgical resection who met the inclusion criteria were enrolled into the study. The flowchart of these patients was shown in Fig.1.There were no significant differences in clinicopathologic features between the two cohorts ( Table 1 ). PA-TACE had a slight negative effect on tumor recurrence (not statistically significant) in the training and two validation cohorts (Fig. S1B-D).

Development of the PDTE scoring system in the training cohort

Univariate Cox regression analysis for determining the risk factors associated with RFS was performed in the training cohort. Results were shown in Table S1. Multivariate analysis revealed that AFP>400 ng/mL,ALBI grade>2.63,operative bleeding loss>800 mL,resection margin<1 cm,tumor size ≥5 cm,multiple tumor number,presence of microvascular invasion,presence of macrovascular invasion,absence of tumor capsular and presence of satellite nodules were the independent factors associated with poor RFS ( Table 2 ).

Table 2Multivariate Cox regression analysis of factors associated with RFS in the training cohort.

Fig. 1. The flowchart of the study design. HCC: hepatocellular carcinoma; PDTE: primary liver cancer big data transarterial chemoembolization; PA-TACE: postoperative adjuvant transarterial chemoembolization.

The regression coefficients of the multivariate Cox regression model were multiplied by 3 and rounded to the nearest unit to obtain simple point numbers,facilitating the bedside calculation of the PDTE scoring system. The PDTE scoring system for a patient could be calculated using the following formula,by adding the sum of multiplying these factors by their respective weights:

PDTE score = AFP ( ≤400 ng/mL = 0;>400 ng/mL = 1) + ALBI( ≤2.63 = 0;>2.63 = 1) + operative bleeding loss ( ≤800 mL = 0;>800 mL = 1) + resection margin ( ≥1 cm = 0;<1 cm = 1) + tumor size (<5 cm = 0; ≥5 cm = 2) + tumor number (solitary = 0; multiple = 2) + microvascular invasion (absence = 0;presence = 2) + macrovascular invasion (absence = 0; presence = 2) + tumor capsular (presence = 0; absence = 1) + satellite nodules (absence = 0; presence = 1).

The effect of PA-TACE based on risk stratification according to the PDTE scoring system

Using 5 points as risk stratification according to the PDTE scoring system (which corresponds to the 66th centile),patients were stratified into 2 subgroups: ≤5 points and>5 points groups.

Next,we investigated the effect of PA-TACE after the risk stratification. In the training cohort,the 1-,3-,and 5-year RFS rates of the TACE group (35.9%,18.5%,and 13.6%) were significantly higher than those of the non-TACE group (29.8%,11.9%,and 4.3%;P<0.001) in patients with>5 points; however,the 1-,3-,and 5-year RFS rates of the TACE group (74.7%,52.1%,and 36.5%) were significantly lower than those of the non-TACE group (78.3%,58.2%,and 42.1%;P= 0.016) in patients with ≤5 points ( Fig. 2 A-B; Table 3 ).In the two validation cohorts,the RFS rates of the TACE group were better than those of the non-TACE group in patients with>5 points; while there were no significant differences in the RFS rates between the two groups in patients with ≤5 points ( Fig. 2 C-F;Table 3 ). Similarly,we found that in>5 points group of the entire cohort,patients undergone PA-TACE had significantly increased RFS compared with non-TACE patients (P<0.001),whereas PATACE was significantly associated with decreased RFS in the ≤5 points group of the entire cohort (P= 0.013) ( Fig. 2 G-H; Table 3 ).These results demonstrated that the PDTE scoring system was able to identify a subgroup of patients who would benefit from PA-TACE based on risk stratification.

Fig. 2. Comparison of RFS between non-TACE and TACE groups based on risk stratification. A: ≤5 points in the training cohort; B: > 5 points in the training cohort; C:≤5 points in the validation cohort; D: > 5 points in the validation cohort; E: ≤5 points in the other validation cohort; F: > 5 points in the other validation cohort; G:≤5 points in the entire cohort; H: > 5 points in the entire cohort. RFS: recurrence-free survival; HR: hazard ratio; 95% CI: 95% confidence interval; TACE: transarterial chemoembolization.

Table 3The effect of PA-TACE based on risk stratification.

Assessing and comparing model performance

In the training and validation cohorts,the C-indexes of PDTE scoring system were 0.714 [standard errors (SE) = 0.010] and 0.716 (SE = 0.018),respectively; those of the G?nen& Heller’s K were 0.681 (SE = 0.004) and 0.683 (SE = 0.008),respectively( Table 4 ). The time-dependent AUC (1-,2- and 3-year) of PDTE scoring system were 0.800 (SE = 0.008),0.804 (SE = 0.008),0.787(SE = 0.008) in the training cohort and 0.818 (SE = 0.010),0.797(SE = 0.010),0.763 (SE = 0.015) in the validation cohort,respectively ( Table 4 ; Fig. 3 ). The above indicators of the PDTE scoring system were greater than ERASL model,Korean model,8th TNM staging system,BCLC staging system and CNLC staging system,suggesting a satisfactory discriminative performance ( Table 4 ). In addition,the calibration curves were performed a good correlation between the predicted and actual outcome in the probability of 1-,2- and 3-year RFS in the training and validation cohorts ( Fig. 3 ).

Fig. 3. Discrimination and calibration plots for the PDTE scoring system. A: Time-dependent AUC in the training cohort; B: time-dependent AUC in the validation cohort; C:calibration curve in the training cohort; D: calibration curve in the validation cohort. PDTE: primary liver cancer big data transarterial chemoembolization; AUC: area under receiver operating characteristic curve.

Table 4Comparison of performance between PDTE scoring system and 5 other models.

Discussion

Tumor recurrence is a major cause of death in HCC,which accounts for more than 60% within 5 years after liver resection [2] .Therefore,prevention of HCC recurrence via adjuvant therapy is an important unmet medical need. However,there is no standard adjuvant treatment proven to prevent tumor recurrence [ 3,4 ]. TACE is the recommended treatment option for BCLC intermediate stage patients [3] . In recent years,PA-TACE,as an adjuvant therapy to prevent tumor recurrence,has attracted more and more attention [5] . However,the adjuvant role of PA-TACE remains controversial [6] . A phase III randomized controlled trial (RCT) found that HCC patients with intermediate or high risk of recurrence could benefit from PA-TACE [8] . Nevertheless,a second RCT recruiting low-risk patients did not confirm these results [27] . A recent metaanalysis indicated that PA-TACE was only beneficial for patients at high risk of postoperative recurrence [7] . In this study,we found that PA-TACE had no impact on tumor recurrence in unstratified cohort (Fig. S1A,P= 0.096).

Toward this goal,a simple score,PDTE scoring system,was developed to identify the patients who might benefit from PA-TACE.In this study,patients with PDTE score>5 points had more aggressive tumor characteristics and higher risk of recurrence than those with PDTE score ≤5 points,and therefore,may benefit from PA-TACE. However,PA-TACE was harmful to the patients with PDTE score ≤5 points. Hence,the model aids decision-making for the application of PA-TACE.

The PDTE scoring system is composed of ten risk factors. Consistent with previous studies,the aggressive tumor-related factors including preoperative AFP level,tumor size and number,microvascular and macrovascular invasion,tumor capsular,and satellite nodules represent the main components of the predictive risk factors [2] . Surgical factors (operative bleeding loss and resection margin) are known to be associated with high tumor recurrence rates in HCC patients [ 28,29 ]. ALBI grade is another independent factor to predict tumor recurrence,which has been used in another model [23] .

There are some limitations to our study. Firstly,the selection bias is inevitable in retrospective studies. However,this bias is minimized by a multicenter cohort. Secondly,this study was performed in China and most HCC patients had HBV infection,and further validation is necessary in different geographic regions.Thirdly,the frequency,drugs and dosages of PA-TACE could vary across medical centers. The standard of PA-TACE considering both the efficacy and safety should be guided.

In conclusion,based on a large multicenter cohort of 4530 patients,the PDTE scoring system is the first easy-to-use tool to identify PA-TACE for patients with HCC after liver resection. Patients with>5 points may benefit from PA-TACE.

Acknowledgments

None.

CRediT authorship contribution statement

Jin-ShuZeng:Conceptualization,Data curation,Formal analysis,Writing – original draft,Writing – review & editing.Jian-Xing Zeng:Conceptualization,Data curation,Formal analysis,Writing original draft,Writing – review & editing.YaoHuang:Data curation,Project administration,Resources,Supervision,Writing – review & editing.Jing-FengLiu:Data curation,Funding acquisition,Project administration,Resources,Supervision,Writing – review &editing.Jin-HuaZeng:Conceptualization,Data curation,Funding acquisition,Project administration,Resources,Supervision,Writing review & editing.

Funding

This study was supported by grants from the Special Fund of Fujian Development and Reform Commission (31010308),the Natural Science Foundation of Fujian Province (2018J01140) and the Key Clinical Specialty Discipline Construction Program of Fuzhou(201912002).

Ethical approval

This study complied with the standards oftheHelsinkiDeclara-tionand was approved by the Institutional Ethics Committees of all study centers.

Competing interest

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

Supplementary materials

Supplementary material associated with this article can be found,in the online version,at doi: 10.1016/j.hbpd.2022.07.007 .