This was a retrospective study. All patients with a diagnosis of HCC based on postoperative histopathology between 1 January 2014 and 31 December 2018 might be eligible for the study. Consecutive patients with HCC were identified via the electronic medical records of Guangxi Medical University Cancer Hospital, Nanning, China, and the Fourth Affiliated Hospital of Guangxi Medical University, Liuzhou, China. Patients were selected based on the following eligibility criteria: (1) patients were without preoperative neoadjuvant therapy (including transarterial chemoembolization, radiotherapy, sorafenib, lenvatinib, immune checkpoint inhibitors, etc), (2) had primary HCC and underwent potential curative liver resection (macroscopically tumor-free), (3) diagnoses of HCC were confirmed by postoperative histopathology, and (4) suffered from chronic hepatitis B (CHB) or MAFLD. Patients were excluded if they were with (1) intrahepatic cholangiocarcinoma, (2) positive anti-HCV or human immunodeficiency virus, (3) alcohol consumption, (4) autoimmune hepatitis, (5) drug-induced liver injury, (6) combined with other malignancies, or (7) incomplete medical information. All the procedures were carried out in accordance with the Helsinki Declaration of 1975. Written informed consent was not provided by the patients because this was a retrospective study. After obtaining institutional board review approval from Guangxi Medical Cancer Hospital (LW2021042) and the Fourth Affiliated Hospital of Guangxi Medical University (LW2021005), demographics, comorbid conditions, clinicopathologic data, radiology reports, and long-term outcomes of the included patients were reviewed.

Based on the aim of the study, included patients would be divided into three groups, namely, patients with MAFLD-related HCC (MAFLD group), patients with CHB-related HCC (CHB group), and those with dual etiology related HCC (CHB/MAFLD group).

The indications of liver resection were based on the Chinese guideline for HCC (18). Patients should be with Child-Pugh score of no more than 7. The presence of appropriate residual liver volume was determined by volumetric dynamic enhanced computerized tomography or magnetic resonance imaging (MRI). The selection criteria and procedures of liver resection have been detailed elsewhere (19–21). The gallbladder was routinely excised before liver resection. Adequate drainage was routinely monitored. Intraoperative ultrasound was performed when necessary.

The diagnosis of MAFLD was based on hepatic steatosis [detected either by imaging techniques (ultrasound, computerized tomography or MRI) or by postoperative liver histopathology] in addition to overweight/obesity (body mass index ≥23 kg/m²), presence of type 2 diabetes mellitus, or evidence of metabolic dysregulation (10, 11). Hepatic steatosis and cirrhosis were determined directly by imaging or histopathology reports. Fibrosis-4, waist circumference, insulin resistance score, plasma high-sensitivity C-reactive protein level, or blood biomarkers/scores of MAFLD was not assessed in this study. Alcohol consumption was defined as excessive alcohol intake (>30 g/day in men, >20 g/day in women) (22). In this study, dual etiology of HCC was defined as a patient suffered from both CHB and MAFLD (23). CHB was defined as the presence of positive of hepatitis B surface antigen HBV DNA, and/or hepatitis B core antibody. CHB-related HCC was defined as a patient with HCC accompanied by CHB. Information on alcohol consumption, autoimmune hepatitis, and drug-induced liver injury was obtained from the past medical history or personal history of medical records. Major hepatectomy was defined as the resection of three or more Couinaud segments (24).

The number or level of the following data were collected and analyzed via the electronic medical records: gender, age, body mass index, hypertension, type 2 diabetes mellitus, hepatic steatosis, liver cirrhosis, Child-Pugh liver function grade, tumor size, tumor number, macrovascular invasion, Barcelona Clinical Liver Cancer stage, hepatitis B surface antigen, alpha fetoprotein, alanine aminotransferase, aspartate aminotransferase, plasma triglycerides, high density lipoprotein cholesterol, total bilirubin, albumin, prealbumin, microvascular invasion, resection type, and adjuvant transarterial chemoembolization.

The primary outcome was overall survival, calculated from the date of liver resection to the date of death from any cause or the date of the last follow-up. No patient undergone live transplantation in this cohort. Follow-up data of overall survival was obtained from the hospital database (X.-M.L). The secondary endpoints included recurrence-free survival and perioperative mortality and morbidity. Recurrence-free survival was calculated from the date of liver resection to the date of tumor recurrence or death from any cause, whichever occurred first. HCC recurrence, which was assessed by the investigators (J.-H.Z, Y.-X.T, K.C), was diagnosed using enhanced computed tomography and/or MRI [including gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI] with or without alpha fetoprotein ≥400 ng/mL) (18). Patients with HCC recurrence would be received appropriate therapeutic approaches, with radiofrequency ablation and repeat liver resection as the preferred curative treatments (25, 26). Postoperative mortality was defined as the rate of death within 30 days after liver resection. Perioperative morbidity was graded according to the Clavien-Dindo classification (27). Clavien-Dindo grade of at least 3a was defined as major morbidity. The follow-up period was up to March 2021 or death. Patients with CHB routinely received antiviral therapy with nucleos(t)ide analogue (28). Moreover, patients with high risk factors of HCC recurrence, such as tumor size at least 5 cm, multinodules, involving macrovascular invasion or microvascular invasion, would receive adjuvant transarterial chemoembolization (29, 30). However, patients with MAFLD did not receive corresponding treatment for MAFLD.

Categorical variables were expressed as number and percentage and compared with Pearson’s chi-square tests. Bonferroni test was used for multiple comparisons. Continuous variables were expressed as mean ± standard deviation and compared by using the one-way analysis of variance for normal distributions. For skewed distributions, variables were expressed as median (interquartile range) and compared by using the Kruskal-Wallis test. Overall and recurrence-free survival were calculated with the Kaplan-Meier method, with the values compared using the log-rank test. To reveal the association between different counterparts (MAFLD, CHB/MAFLD, and CHB) and overall or recurrence-free survival after liver resection, multivariate analyses were performed using a Cox proportional hazards model. Any confounders that a change in effect estimate was more than 5% or recognized to be associated with the outcomes were entered into a Cox regression analysis. A two-tailed p value < 0.05 was considered to indicate statistically significant. GraphPad Prism 8.0 and IBM SPSS (ver. 26.0 SPSS Inc., Chicago, IL, USA) softwares were used to perform the analyses.

To reduce the potential effect of baseline variable imbalance on prognosis, propensity score matching between groups at a 1:3 ratio was used. All collected variables were involved in propensity score matching except body mass index, hypertension, type 2 diabetes mellitus, triglycerides, and high density lipoprotein cholesterol, which were used to define MAFLD. The propensity score was generated by a logistic regression. Nearest-neighbor caliper matching without replacement (random order or closest distance) was used to pair MAFLD (or MAFLD plus CHB/MAFLD) and CHB patients with similar propensity score values (31).

A total of 1887 patients underwent potential curative liver resection of HCC from January 2014 from December 2018; 368 had with anti-HCV or HIV positive, alcohol consumption, combined with other malignancies, autoimmune hepatitis or drug-induced liver injury and were excluded from this study. Other 194 patients without CHB or MAFLD were also excluded. Of the remaining 1325 patients, 67 (5.0%) had MAFLD alone, 1082 (81.7%) had CHB alone, and 176 (13.3%) had CHB concurrent with MAFLD ( Figure 1 ). The diagnosis of hepatic steatosis was detected by imaging techniques alone (13, 3.8%), liver histopathology alone (299, 88.2%) or confirmed by both (27, 8.0%).

Compared to patients with CHB, MAFLD patients were older (58.8 vs 48.9 yrs), more frequently had overweight/obesity (62.7% vs 33.3%), hypertension (35.8% vs 6.4%), type 2 diabetes mellitus (26.9% vs 5.8%), and early stage disease (77.6% vs 59.0%, BCLC stage 0/A). Hepatic synthetic function at HCC presentation (measured by prothrombin time, prealbumin, alanine aminotransferase, and aspartate aminotransferase levels) was better in MAFLD patients. There were no differences in rates of gender, smoking, Child-Pugh grade B, or macrovascular invasion. Moreover, fewer MAFLD patients had cirrhosis (50.7% vs 71.3%), preoperative alpha fetoprotein concentration ≥ 400 ng/mL (23.9% vs 42.2%), tumor size ≥ 5 cm (49.3% vs 61.6%), mulinodular (10.4% vs 22.5%), microvascular invasion (41.8% vs 56.7%), and receiving major hepatectomy (19.4% vs 34.1%) or adjuvant transarterial chemoembolization (32.8% vs 55.0%) than CHB patients. However, MAFLD patients had higher level of triglycerides ( Table 1 ).

During a median follow-up of 47 months, 450 of 1325 (34.0%) patients died mainly because of HCC progression and/or hepatic failure. Of the total population (n=1325), 1-year, 3-year, and 5-year overall survival were 87.4%, 68.6%, and 59.8%, respectively ( Supplementary Figure 1A ). One-year, 3-year, and 5-year recurrence-free survival were 57.8%, 39.0%, and 30.7%, respectively ( Supplementary Figure 1B ). Univariable and multivariable analyses for overall and recurrence-free survival are summarized in Table 2 . MAFLD was independently associated with overall and recurrence-free survival (all P<0.05) on multivariable analyses.

Postoperative mortality within 30 days after liver resection was observed in two (0.2%) patients in CHB group. However, no postoperative mortality was observed in other two groups. MAFLD group (4.5%, 3/67) had the lowest major morbidity than CHB group (7.9%, 85/1082) and CHB/MAFLD group (10.2%, 18/176) (all P>0.05).

Median follow-up for patients in the MAFLD, CHB/MAFLD, and CHB group were 46 (range, 7-87), 44 (range, 2-86), and 47 (range, 1-86) months, respectively. MAFLD patients had significantly higher overall and recurrence-free survival than CHB/MAFLD patients (P=0.002, Figure 2A ; P=0.038, Figure 2B ) or CHB patients (P<0.001, Figure 2A ; P<0.001, Figure 2B ). In addition, CHB/MAFLD patients had higher overall (P=0.059; Figure 2A ) and recurrence-free survival than CHB patients (P=0.010; Figure 2B ).

Baseline clinical characteristics between MAFLD (n=58) and CHB (n=180) groups were comparable after propensity score matching ( Table 3 ). MAFLD patients had significantly higher overall and recurrence-free survival than CHB patients (P=0.004, Figure 3A ; P=0.043, Figure 3B ).

To further explore the difference in prognosis between MAFLD and CHB, patients with MAFLD or CHB/MAFLD were combined in one group. Baseline clinical characteristics of the two groups before and after propensity matching were described in Table 4 . Patients with MAFLD or CHB/MAFLD had significantly higher overall and recurrence-free survival than CHB patients before or after propensity score matching (all P<0.05, Figure 4 ).