Prognostic value of CT-defined sarcopenic obesity in hepatocellular carcinoma patients undergoing transarterial chemoembolization

Abstract

Purpose

The prognostic value of sarcopenic obesity is controversial in hepatocellular carcinoma (HCC) patients undergoing transarterial chemoembolization (TACE). This study aimed to deconstruct the sarcopenic obesity phenotype to clarify the independent prognostic impacts of sarcopenia and visceral obesity.

Methods

We retrospectively analyzed 415 patients with unresectable HCC who underwent TACE between 2009 and 2020. Skeletal muscle index (SMI) and visceral fat area (VFA) were measured from baseline CT scans at the L3 vertebra level. Sarcopenia (SMI ≤ 36.2 cm²/m² in males, ≤ 29.6 cm²/m² in females) and visceral obesity (VFA ≥ 100 cm²) were defined using Asian-specific cutoffs. Overall survival (OS) was analyzed using Kaplan-Meier and Cox proportional hazards models, with sex-stratified analysis.

Results

Among 415 patients (72% male, median follow-up 21.4 months), sarcopenia prevalence was 34% and visceral obesity 59%. In sex-stratified analysis, sarcopenia predicted mortality in males and reduced the median OS by 11 months (14.4 vs. 25.4 months, p < 0.001), but not in females (23.4 vs. 22.8 months, p = 0.900). Visceral obesity showed no prognostic impact in either sex. In sex-stratified multivariate analysis adjusting for Child-Pugh class, tumor burden, BCLC stage, and AFP, sarcopenia remained independently associated with mortality in males (adjusted hazard ratio [HR] = 1.41, 95% CI: 1.06–1.86, p = 0.018) but not in females (adjusted HR = 0.84, 95% CI: 0.55–1.27, p = 0.404; p for interaction = 0.070), while visceral obesity and sarcopenic obesity composite showed no prognostic value in either sex.

Conclusion

The sarcopenic obesity composite phenotype lacks prognostic value in TACE-treated HCC patients. Sarcopenia independently predicts mortality in males but not females, while visceral obesity shows no independent prognostic significance. Clinical risk stratification could consider sarcopenia assessment in male patients rather than composite phenotypes.

Introduction

Hepatocellular carcinoma (HCC) was the sixth most prevalent cancer and the third leading cause of cancer-related mortality worldwide in 2020 [1]. Surgical resection and liver transplantation offer curative options for early-stage disease (Barcelona Clinic Liver Cancer [BCLC] stages 0 and A). Nevertheless, most patients are diagnosed at intermediate stages where transarterial chemoembolization (TACE) is the recognized standard of therapy [2, 3]. However, the marked heterogeneity in tumor burden and liver function within the BCLC-B stage, a fact acknowledged in the updated 2022 BCLC guidelines, results in variable survival outcomes post-TACE [2, 4]. Conventional prognostic models, such as the Child-Pugh score, Model for End-Stage Liver Disease, and albumin-bilirubin grade, do not adequately capture a patient’s crucial nutritional and functional status. Therefore, the identification of more comprehensive biomarkers is needed for improved risk stratification [2, 4].

Body composition analysis has become an important tool in this context. Sarcopenia, characterized by reduced muscle mass and function, affects an estimated 43.2% of patients with primary liver cancer [5]. Sarcopenia has been shown in numerous studies to be a strong independent predictor of lower objective response rates and shorter overall survival (OS) after TACE [6–11]. At the same time, obesity is a known risk factor for poor post-treatment outcomes in cancer patients, especially visceral adiposity measured by visceral fat area (VFA) [12–17]. The coexistence of these two conditions, termed sarcopenic obesity (SO), creates a clinically complex phenotype. Even though SO is increasingly recognized as an independent predictor of poor outcomes in several malignancies, especially following surgical resection [16–19], its precise significance in the context of TACE is still debated. This uncertainty is rooted in the “obesity paradox,” which is a phenomenon where being overweight is not consistently linked to poorer outcomes. This concept suggests that differences in body composition, rather than body mass index (BMI) alone, may be the true determinants of prognosis. In some studies [20–22], excess body weight was associated with favorable outcomes, while in others it had no or a negative effect. Consequently, a key question remains regarding whether visceral obesity has independent prognostic value in sarcopenic patients or if the observed outcomes are primarily driven by sarcopenia alone.

This study aimed to clarify the independent prognostic contributions of sarcopenia and visceral obesity in patients with unresectable HCC undergoing TACE. By analyzing these body composition components separately rather than as a composite phenotype, we sought to determine whether the SO construct provides meaningful prognostic information or if the outcomes are primarily attributable to sarcopenia alone.

Methods

Study design and patient population

This retrospective study included patients who underwent TACE for HCC between January 2009 and December 2020. The study was approved by the Institutional Review Board of the Faculty of Medicine and was conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study.

The inclusion criteria were as follows: (i) HCC patients initially treated with TACE, (ii) Child-Pugh class A or B liver disease, (iii) BCLC stage A or B HCC, and (iv) availability of a baseline abdominal CT scan within 4 weeks prior to the first TACE procedure. The exclusion criteria were as follows: (i) extrahepatic tumor metastasis, (ii) history of prior anticancer treatment for HCC, (iii) Child-Pugh class C liver function, (iv) any other malignancy, (v) missing weight-measurement data, or (vi) insufficient CT image quality for body composition analysis. A total of 415 patients were enrolled in this study. A schematic diagram of the study workflow is presented in Fig. 1.

Fig. 1.

Schematic diagram of the study workflow. The process involved three main stages. A CT image analysis: baseline axial CT scans at the third lumbar vertebra were used to segment and quantify skeletal muscle area and visceral fat area based on Hounsfield Unit (HU) thresholds. B Data extraction: key clinical factors were collected and combined with the CT-derived sarcopenia and visceral obesity status, which were defined using validated, Asian-specific cutoffs. C Model analysis: these variables were assessed using univariate (Kaplan-Meier curves and log-rank test) and multivariate (Cox proportional hazards regression) analyses to determine their independent impact on the primary outcome of overall survival

CT image analysis and body composition definitions

Skeletal muscle mass and visceral fat tissue were analyzed from the latest CT scan images prior to the first TACE procedure. Cross-sectional skeletal muscle area and visceral adipose tissue area were measured at the level of the third lumbar vertebra. Skeletal muscle mass and visceral fat tissue were analyzed using unenhanced (non-contrast) CT images to ensure the stability of Hounsfield Unit (HU)-based segmentation and to avoid variability associated with contrast enhancement. Tissue areas were determined using HU calculated with in-house software developed using MATLAB (The MathWorks, Natick, MA, USA) and Python 3.6.13 (Anaconda, Inc.) to generate a measurement model based on neural network architecture (U-Net) [23, 24]. The L3 vertebra slice was identified by a radiologist before automated segmentation by the U-Net–based pipeline. Although formal intra- and inter-observer reliability testing of slice selection or segmentation was not performed in this study, the segmentation pipeline has been previously validated against manual segmentation [23].

Sarcopenia was defined using the skeletal muscle index (SMI) calculated by normalizing the cross-sectional skeletal muscle area (cm²) by patient height squared (m²). A low SMI indicated sarcopenia as defined using Asian-specific cutoff values of 36.2 cm²/m² for males and 29.6 cm²/m² for females [24]. A uniform VFA cutoff of 100 cm² was applied for both sexes based on the Japan Society for the Study of Obesity guidelines [12]. This unified approach was selected because the JASSO threshold is the most widely validated VFA cutoff in Asian populations for increased metabolic and cardiovascular risk [25] and has been most commonly applied in prior HCC body composition studies, while consensus sex-specific VFA cutoffs for HCC prognosis have not yet been established.

Based on these definitions, patients were classified into four body composition categories: non-sarcopenic/non-obese (NN), non-sarcopenic/obese (NO), sarcopenic/non-obese (SN), and sarcopenic/obese (SO).

TACE procedure

All patients with HCC were treated using conventional TACE. After advancing the catheter tip into the tumor-feeding arteries, chemotherapeutic agents mixed with lipiodol were slowly injected. The procedure was completed when the tumor-feeding branch was embolized with gelatin sponge particles. The chemotherapeutic agents administered included doxorubicin (5–50 mg/session) or mitomycin (10–20 mg/session). The choice of anticancer agents, lipiodol emulsion dosage (4–16 mL/session), and gelatin sponge application was individualized for each patient. Repeat TACE sessions were performed on demand at 6–12 week intervals in patients with residual tumor enhancement and vascularity on follow-up imaging, provided they had favorable liver function and performance status.

Data collection

Data were extracted from the electronic medical records and included patient demographics (age at TACE, sex, BMI), comorbidities (diabetes, hypertension, cardiovascular disease, chronic kidney disease), clinical parameters (liver disease etiology categorized as hepatitis B virus [HBV], hepatitis C virus [HCV], or others; Child-Pugh class; serum alpha-fetoprotein [AFP] level), and tumor characteristics that included BCLC stage, largest tumor diameter, and tumor number categorized as solitary or multiple.

Statistical analysis

Continuous variables are presented as median (interquartile range [IQR]) or mean ± standard deviation (SD) as appropriate, and were compared using the Mann-Whitney U test. The categorical variables were compared using the χ² test or Fisher’s exact test. OS was analyzed with the Kaplan-Meier method. Survival curves were compared using the log-rank test. Vital status was ascertained from the civil registry with follow-up through June 30, 2023. OS was calculated from the date of the first TACE procedure to death or to June 30, 2023 for patients who remained alive at the last follow-up. The correlation between continuous SMI and VFA values was quantified using Pearson’s correlation coefficient (r). The categorical association between sarcopenia status and visceral obesity status was assessed using the chi-square test. Any variable identified as significant or with a value of p < 0.2 in the univariate analysis was considered a candidate for the Cox regression multivariate analysis, and the results are presented as hazard ratio (HR) with 95% confidence interval (CI). All statistical tests were two-sided. The value of p < 0.05 was considered significant. Data analyses were performed using R software, version 4.2.0 (R foundation, Vienna, Austria).

To investigate the sex-specific effects observed in the Kaplan-Meier analyses, we performed sex-stratified multivariate Cox proportional hazards regression models. Separate models were constructed for male and female cohorts. Each model was adjusted for sarcopenia, visceral obesity (VFA ≥ 100 cm²), Child-Pugh class, tumor burden (up-to-7 criteria), BCLC stage, and AFP level. Statistical interaction between sarcopenia and sex was formally tested by including a multiplicative interaction term (sarcopenia × sex) in the full cohort model. The significance of the interaction was assessed using the likelihood ratio test.

Results

Patient characteristics

A total of 415 patients with unresectable HCC who underwent TACE were included in the analysis. Baseline demographic, clinical, tumor-related, and body composition characteristics are presented in Table 1. The study cohort had a mean age of 62.1 ± 10.9 years and was predominantly male (297 patients, 71.6%). HBV infection was the leading etiology (47.0%) followed by HCV infection (21.4%). Most patients had preserved liver function with 337 (81.2%) classified as Child-Pugh class A. The most common comorbidities were diabetes (29.4%) and hypertension (27.7%). Intermediate-stage disease (BCLC stage B) was present in 295 patients (71.1%). The median tumor size was 4.3 cm (interquartile range [IQR], 2.8–8.2 cm), and the median serum AFP level was 30 ng/mL (IQR, 8–467 ng/mL). Single tumors were observed in 179 patients (43.1%). Body composition analysis revealed a median BMI of 23.6 kg/m² (IQR, 21.1–26.1 kg/m²). CT-based measurements showed a median VFA of 119 cm² (IQR, 76–132 cm²) and a median SMI of 37.4 cm²/m² (IQR, 31.7–42.6 cm²/m²).

Table 1.

Baseline characteristics of the study population stratified by body composition phenotypes

Characteristic	All patients (N = 415)	NN (n = 100)	SN (n = 69)	NO (n = 174)	SO (n = 72)	p
Age (years), mean (SD)	62.1 (10.9)	57.4 (10.0)	64 (10.5)	61.5 (10.4)	68.1 (10.7)	< 0.001
Male, n (%)	297 (71.6)	78 (78.0)	45 (65.2)	134 (77.0)	40 (55.6)	0.002
Etiology						< 0.001
HBV	195 (47.0)	59 (59.0)	31 (44.9)	80 (46.0)	25 (34.7)
HCV	89 (21.4)	25 (25.0)	22 (31.9)	32 (18.4)	10 (13.9)
Others	131 (31.6)	16 (16.0)	16 (23.2)	62 (35.6)	37 (51.4)
Diabetes	122 (29.4)	14 (14.0)	15 (21.7)	56 (32.2)	37 (51.4)	< 0.001
Hypertension	115 (27.7)	17 (17.0)	17 (24.6)	50 (28.7)	31 (43.1)	0.002
Dyslipidemia	44 (10.6)	3 (3.0)	8 (11.6)	19 (10.9)	14 (19.4)	0.007
Cardiovascular disease	26 (6.3)	2 (2.0)	7 (10.1)	9 (5.2)	8 (11.1)	0.036
Chronic kidney disease	16 (3.9)	0 (0.0)	3 (4.3)	7 (4.0)	6 (8.3)	0.023
Child-Pugh class A, n (%)	337 (81.2)	80 (80.0)	53 (76.8)	144 (82.8)	60 (83.3)	0.692
Tumor size, median (IQR)	4.3 (2.8–8.2)	5.0 (2.7–10.3)	6.2 (3.2–10.7)	3.8 (2.5–5.6)	4.3 (3.1–8.6)	0.003
Multiple tumor	236 (56.9)	60 (60)	44 (63.8)	90 (51.7)	42 (58.3)	0.298
Beyond up-to-7 criteria, n (%)	203 (48.9)	57 (57)	43 (62.3)	66 (37.9)	37 (51.4)	0.001
BCLC stage B, n (%)	295 (71.1)	74 (74.0)	52 (75.4)	119 (68.4)	50 (69.4)	0.629
AFP (ng/mL), median (IQR)	30 (8–467)	45 (9–805)	28 (6–234)	24 (7–228)	44 (10–1411)	0.130
BMI (kg/m²), median (IQR)	23.6 (21.1–26.1)	22.3 (20.3–23.8)	19.2 (17.9–21.8)	25.6 (23.9–28.4)	23.9 (21.3–26.2)	< 0.001
VFA (cm²), median (IQR)	119 (76–132)	73 (52–86)	66 (44–83)	164 (134–219)	151 (124–189)	< 0.001
SMI (cm²/m²), median (IQR)	37.4 (31.7–42.6)	40.1 (37.3–44.7)	29.4 (27.3–33.9)	41.6 (37.6–45.0)	28.3 (24.9–33.4)	< 0.001

NN non-sarcopenic/non-obese, SN sarcopenic/non-obese, NO non-sarcopenic/obese, SO sarcopenic/obese, SD standard deviation, HBV hepatitis B virus, HCV hepatitis C virus, IQR interquartile range, BCLC Barcelona Clinic Liver Cancer, AFP alpha-fetoprotein, BMI body mass index, VFA visceral fat area, SMI skeletal muscle index

Patient stratification by body composition phenotypes

Patients were categorized into four body composition phenotypes: non-sarcopenic/non-obese (NN, n = 100, 24.1%), sarcopenic/non-obese (SN, n = 69, 16.6%), non-sarcopenic/obese (NO, n = 174, 41.9%), and sarcopenic/obese (SO, n = 72, 17.3%). Overall, 141 patients (34.0%) had sarcopenia and 246 (59.3%) had visceral obesity.

Baseline characteristics differed significantly across phenotypes (Table 1). Sarcopenic patients (SN and SO) were older (mean age 64.0 vs. 60.6 years, p < 0.001) and had lower male predominance (60% vs. 77%, p = 0.002) compared to non-sarcopenic groups. Obese phenotypes (NO and SO) showed higher prevalence of metabolic comorbidities, including diabetes (42% vs. 18%, p < 0.001), hypertension (36% vs. 21%, p = 0.002), and dyslipidemia (15% vs. 7%, p = 0.007). Although Child-Pugh class A and BCLC stage B distribution were comparable across groups (p = 0.692 and p = 0.629, respectively), significant differences emerged in tumor burden. Non-obese phenotypes (NN and SN) presented with more advanced disease, with 59.2% beyond up-to-7 criteria compared to 41.9% in obese groups (p = 0.001). The sarcopenic/non-obese (SN) phenotype had the largest median tumor size (6.2 cm), which suggested this group represents the most advanced disease state.

Survival rates according to body composition

Phenotype-specific survival outcomes

Over a median follow-up period of 21.4 months (IQR 10.1–43.7), significant survival differences emerged among body composition phenotypes (log-rank test, p = 0.040) (Fig. 2). The non-sarcopenic/obese (NO) phenotype demonstrated the most favorable survival profile with a median OS of 25.0 months (95% CI: 19.3–30.7) followed by the non-sarcopenic/non-obese (NN) group at 23.4 months (95% CI: 15.9–34.0). Sarcopenic phenotypes showed progressively worse outcomes. The sarcopenic/obese (SO) group achieved 19.6 months (95% CI: 13.7–28.4) and the sarcopenic/non-obese (SN) group demonstrated the poorest survival at 15.4 months (95% CI: 11.4–23.3). Temporal survival analysis revealed sustained prognostic separation across phenotypes. Five-year OS rates were 25.5%, 19.8%, 16.7%, and 13.4% for NO, NN, SO, and SN patients, respectively. Among all pairwise comparisons, the only statistically significant difference was the inferior survival observed in the SN group compared to the NO group (p = 0.035) that represented a 38.4% reduction in median survival time.

Fig. 2.

Kaplan-Meier curves for overall survival based on body composition phenotypes

Sarcopenia impact

When analyzed as a dichotomous variable, sarcopenia emerged as a highly significant predictor of mortality in the overall cohort (log-rank test, p < 0.001) (Fig. 3). Sarcopenic patients (n = 141) demonstrated markedly inferior survival with a median OS of 17.1 months (95% CI: 13.6–23.7) compared to 25.0 months (95% CI: 19.5–29.5) in non-sarcopenic patients (n = 274). However, a sex-stratified analysis revealed that this prognostic impact was driven entirely by the male cohort. In males, sarcopenia remained a powerful predictor of poor survival (p < 0.001) that reduced the median OS by 11 months from 25.4 months in non-sarcopenic men down to 14.4 months. In contrast, sarcopenia had no significant impact on survival in females (p = 0.900). The median OS was nearly identical between the sarcopenic and non-sarcopenic groups (23.4 months vs. 22.8 months, respectively).

Fig. 3.

Kaplan-Meier curves for overall survival in patients with low skeletal muscle index (SMI) (sarcopenia) versus high SMI (non-sarcopenia). Survival is shown for (A) the entire cohort, (B) male patients only, and (C) female patients only. P values were calculated using the log-rank test

Visceral obesity impact

In the combined cohort, visceral obesity alone did not significantly influence survival outcomes (log-rank test, p = 0.100) (Fig. 4). Despite the lack of overall significance, there was a numerical trend toward improved survival in the visceral obesity group, which had a median OS of 23.6 months (n = 246) compared to 18.1 months in the non-obese group (n = 169). A sex-stratified analysis revealed that this protective trend was driven entirely by the male cohort. In males, visceral obesity showed a trend toward improved survival that approached but did not reach statistical significance (p = 0.06), with a median OS of 23.8 months in the obese group compared to 16.1 months in the non-obese group. Conversely, this trend was absent in females, where no significant difference was observed (p = 0.800), and the median survival times were similar between the obese (22.8 months) and non-obese (23.6 months) groups.

Fig. 4.

Kaplan-Meier curves for overall survival in patients with high visceral fat area (VFA) (visceral obesity) versus low VFA (non-visceral obesity). Survival is shown for (A) the entire cohort, (B) male patients only, and (C) female patients only. P values were calculated using the log-rank test

Correlation between SMI and VFA

Sarcopenia and visceral obesity showed a significant negative association when analyzed categorically (χ² = 5.46, p = 0.019), which indicated that these phenotypes rarely coexist. However, continuous SMI and VFA values demonstrated a weak positive correlation (r = 0.120, p = 0.014).

Prognostic factors for overall survival

To determine the independent prognostic contributions of sarcopenia and visceral obesity, both components were assessed as separate variables in the multivariate Cox proportional hazards model. Additionally, sarcopenic obesity as a combined binary phenotype was also evaluated.

Results of the multivariate Cox proportional hazards analysis are presented in Table 2. After adjusting for all significant clinical and demographic variables, five factors remained independent predictors of mortality. Sarcopenia was confirmed as a significant and independent predictor of poor survival (adjusted [adj.] HR = 1.27; 95% CI 1.01–1.60; p = 0.043). Other powerful independent predictors included BCLC stage B (adj. HR = 1.72; p = 0.001), Child-Pugh class B (adj. HR = 1.57; p = 0.002), being beyond the up-to-7 criteria (adj. HR = 1.40; p = 0.016), and an AFP level ≥ 200 ng/mL (adj. HR = 1.43; p = 0.003). In contrast, visceral obesity (VFA ≥ 100 cm²) did not show independent prognostic value in the adjusted model (adj. HR = 0.96; 95% CI 0.77–1.20; p = 0.745). When evaluated as a single composite variable, the SO phenotype also showed no significant association with survival in univariate analysis (HR = 1.14; 95% CI 0.86–1.51; p = 0.373) and was therefore not included in the final multivariate model.

Table 2.

Univariate and multivariate Cox proportional hazards analysis for predictors of overall survival

Variable	Reference	Univariate analysis		Multivariate analysis
		HR (95% CI)	p	HR (95% CI)	p
Age > 60 years	≤ 60 years	1.05 (0.85–1.31)	0.632
Sex: male	Female	1.05 (0.83–1.33)	0.697
Hepatitis B positive	Negative	1.08 (0.87–1.34)	0.482
Anti-hepatitis C positive	Negative	1.04 (0.81–1.33)	0.773
Diabetes: yes	No	0.89 (0.7–1.12)	0.311
Hypertension: yes	No	1.08 (0.85–1.37)	0.529
Dyslipidemia: yes	No	0.63 (0.43–0.92)	0.012	0.73 (0.49–1.08)	0.098
Cardiovascular disease: yes	No	1.03 (0.68–1.58)	0.881
Chronic kidney disease: yes	No	0.65 (0.36–1.19)	0.137	0.74 (0.40–1.36)	0.306
Child–Pugh class B	Class A	1.85 (1.42–2.40)	< 0.001	1.57 (1.19–2.08)	0.002
Up-to-7 criteria: Beyond	Within	2.08 (1.68–2.59)	< 0.001	1.40 (1.06–1.84)	0.016
BCLC stage B	BCLC Stage A	2.43 (1.88–3.15)	< 0.001	1.72 (1.24–2.39)	0.001
AFP ≥ 200 ng/mL	< 200 ng/mL	1.61 (1.28–2.02)	< 0.001	1.43 (1.14–1.81)	0.003
BMI ≥ 25 kg/m²	< 25 kg/m²	0.89 (0.71–1.12)	0.306
VFA ≥ 100 cm²	< 100 cm²	0.83 (0.67–1.04)	0.102	0.96 (0.77–1.20)	0.745
Sarcopenia: present	Absent	1.33 (1.06–1.66)	0.014	1.27 (1.01–1.60)	0.043
Sarcopenic Obesity: present	Absent	1.14 (0.86–1.51)	0.373

HR hazard ratio, CI confidence interval, BCLC Barcelona Clinic Liver Cancer, AFP alpha-fetoprotein, BMI body mass index, VFA visceral fat area

Sex-stratified multivariate analysis

Sex-stratified multivariate analysis (Table 3) revealed marked differences in the prognostic impact of body composition. In males (n = 297), sarcopenia remained an independent predictor of mortality (adj. HR = 1.41, 95% CI: 1.06–1.86, p = 0.018), alongside Child-Pugh class B (adj. HR = 1.66, 95% CI: 1.20–2.31, p = 0.002), BCLC stage B (adj. HR = 1.55, 95% CI: 1.04–2.31, p = 0.032), and elevated AFP (adj. HR = 1.48, 95% CI: 1.13–1.94, p = 0.005). In contrast, among females (n = 118), sarcopenia showed no prognostic significance (adj. HR = 0.84, 95% CI: 0.55–1.27, p = 0.404), with BCLC stage B emerging as the sole independent predictor (adj. HR = 2.36, 95% CI: 1.29–4.33, p = 0.006). Formal interaction testing did not reach conventional statistical significance (p for interaction = 0.070, likelihood ratio test), although the directionally opposing point estimates were consistent with possible effect modification by sex. Visceral obesity lacked independent prognostic value in both males (adj. HR = 0.89, 95% CI: 0.68–1.15, p = 0.370) and females (adj. HR = 0.91, 95% CI: 0.58–1.42, p = 0.666).

Table 3.

Sex-stratified multivariate Cox regression analysis

Variable	Male (n = 297)		Female (n = 118)
	HR (95% CI)	p	HR (95% CI)	p
Sarcopenia	1.41 (1.06–1.86)	0.018	0.84 (0.55–1.27)	0.404
VFA ≥ 100 cm²	0.89 (0.68–1.15)	0.370	0.91 (0.58–1.42)	0.666
Child–Pugh class B	1.66 (1.20–2.31)	0.002	1.39 (0.80–2.42)	0.242
Beyond up-to-7 criteria	1.33 (0.95–1.86)	0.095	1.63 (0.98–2.72)	0.061
BCLC stage B	1.55 (1.04–2.31)	0.032	2.36 (1.29–4.33)	0.006
AFP ≥ 200 ng/mL	1.48 (1.13–1.94)	0.005	1.50 (0.96–2.33)	0.073

HR hazard ratio, CI confidence interval, VFA visceral fat area, BCLC Barcelona Clinic Liver Cancer, AFP alpha-fetoprotein

Discussion

Sarcopenia, but not visceral obesity or their combination, independently predicts mortality in HCC patients undergoing TACE. While the four body composition phenotypes showed OS differences (log-rank p = 0.040), only sarcopenia retained independent prognostic significance after adjusting for tumor burden, liver function, and AFP (adj. HR = 1.27, p = 0.043). The effect was confined to male patients, where sarcopenia reduced median OS by 11 months from 25.4 to 14.4 months. Visceral obesity lacked independent prognostic value in both sexes (p = 0.745), and the SO composite showed no association with mortality (p = 0.373). Clinical risk stratification in TACE-treated HCC may benefit from prioritizing sarcopenia assessment in males rather than composite phenotypes.

The relationship between sarcopenia and visceral obesity in this cohort appears more nuanced than suggested by the composite SO construct. Rather than representing independent opposing forces, these body composition abnormalities likely reflect different stages of cancer-related metabolic derangement. The SN phenotype showed the worst survival outcomes (median OS 15.4 months), probably representing advanced cancer cachexia with severe depletion of both muscle and fat reserves. The sarcopenic/obese (SO) group showed slightly better survival (median OS 19.6 months), which suggested less advanced metabolic deterioration where some adipose tissue remains as a residual energy reservoir. The non-sarcopenic/obese (NO) phenotype achieved the most favorable survival (median OS 25.0 months) that represented preserved metabolic status with intact muscle mass and adequate energy reserves.

The consistency of the sex-specific pattern across both univariate and multivariate analyses is suggestive of true effect modification rather than a chance finding. Formal interaction testing did not reach conventional statistical significance (p for interaction = 0.070, likelihood ratio test), although the directionally opposing point estimates were consistent with possible effect modification by sex. The limited size of the female subgroup (n = 118, vs. 297 males) likely reduced statistical power to detect this interaction, and this finding should therefore be interpreted as exploratory and hypothesis-generating. Our male-specific hazard ratio aligns closely with a recent meta-analysis reporting a pooled HR of 1.38 for sarcopenia after TACE [26], which further supports the validity of this association.

These findings raise the question of why sarcopenia predicts mortality primarily in male patients. Males possess a higher proportion of type II muscle fibers, which demonstrate greater susceptibility to cancer-induced wasting [27]. Sex hormones also play divergent roles in muscle homeostasis: testosterone deficiency in cirrhotic males accelerates muscle catabolism and predicts mortality, whereas estrogen confers protection in females by enhancing mTORC1 signaling and preserving mitochondrial function [28, 29]. Treatment-related outcomes in females may be driven predominantly by tumor characteristics rather than baseline body composition, as evidenced by the stronger prognostic impact of BCLC stage in females (HR = 2.36) compared to males (HR = 1.55). Women experience 1.5- to 2-fold higher rates of chemotherapy-related adverse events compared to men [30], potentially obscuring the prognostic value of sarcopenia in this population. The mechanistic hypotheses proposed above should be regarded as hypothesis-generating rather than confirmed pathways, as the supporting evidence is largely derived from preclinical studies and cirrhotic populations outside the HCC-TACE setting.

The “obesity paradox” was also confined to males. We observed a protective trend with visceral obesity that approached statistical significance in males (p = 0.06), with a median OS of 23.8 months in obese males compared to 16.1 months in non-obese males, while this effect was entirely absent in females (p = 0.800), where median OS remained similar (22.8 vs. 23.6 months). This aligns with a large nationwide study [31] that demonstrated improved survival in overweight males post-TACE. However, this protective trend with visceral obesity in males did not translate to independent prognostic value in multivariable analysis (adj. HR = 0.89, p = 0.370), which suggested the univariate association was confounded by other clinical factors such as tumor burden and liver function.

Tumor burden at presentation also contributes to the observed survival differences. The lean, non-obese phenotypes (NN and SN) presented with more aggressive tumor features that demonstrated a significantly higher proportion of patients beyond the up-to-7 criteria compared to the obese phenotypes (59.2% vs. 41.9%, p = 0.001). The SN group had the largest median tumor size at 6.2 cm, compared to 5.0 cm in NN, 3.8 cm in NO, and 4.3 cm in SO groups. This pattern supports the interpretation that the SN phenotype represents the most severe cachexia state, which itself is a manifestation of advanced, aggressive tumor biology rather than a simple nutritional deficiency [32]. Consistent with this concept, advanced tumor burden and aggressive disease features remain strongly associated with poor survival outcomes across HCC subgroups, including those with macrovascular invasion [33].

Routine body composition assessment from standard-of-care CT scans could be considered part of pre-treatment evaluation for patients undergoing TACE, particularly for male patients in whom sarcopenia demonstrates strong prognostic value [34, 35], alongside other elements of preoperative and pre-procedural assessment in HCC, such as evaluation of hepatic vascular anatomy and surgical planning [36]. Beyond its prognostic role, sarcopenia may also be amenable to intervention [37]. While we cannot establish whether interventions targeting sarcopenia would improve outcomes, emerging evidence suggests that multimodal prehabilitation programs, including dietary supplementation and resistance exercise, may enhance physiological resilience in cancer patients [38]. As the HCC treatment landscape continues to evolve toward sequential and combination strategies, including emerging approaches such as immune checkpoint inhibitor rechallenge [39], integrating baseline body composition assessment may help guide individualized management across treatment lines.

This study has several strengths, including a large, well-defined cohort of patients who received uniform initial treatment with conventional TACE that was followed by a comprehensive sex-stratified analysis. CT-quantified VFA allowed direct assessment of metabolically active adiposity rather than relying on less precise BMI metrics. Using validated Asian-specific cutoffs for sarcopenia and visceral obesity enhanced the applicability of our findings in Asian populations.

Some limitations should be acknowledged. The retrospective, single-center design limits generalizability beyond Asian populations. We lacked functional assessments (grip strength) and longitudinal body composition data. In addition, our analyses were based on all-cause mortality, as cause-of-death classification was not available; therefore, a competing risk analysis could not be performed and should be considered in future studies. Our uniform VFA cutoff (100 cm² for both sexes) follows Japanese guidelines but may not be optimal for sex-specific prognostic stratification. Future prospective studies should incorporate functional measurements, serial body composition assessments, and evaluate targeted interventions in sarcopenic male patients.

Conclusion

The SO composite phenotype has no independent prognostic value in TACE-treated HCC patients. Sarcopenia independently predicts mortality in male patients but shows no prognostic impact in females, which highlights sex-specific differences in body composition-survival associations. Visceral obesity lacks independent predictive value in both sexes after adjusting for clinical confounders. The pattern where severe sarcopenia without visceral fat represents the worst outcomes suggests these body composition changes reflect different stages of cancer-related metabolic derangement rather than independent biological processes. Clinical risk stratification may benefit from sex-specific sarcopenia assessment in male patients rather than the sarcopenic obesity combination. Future research should focus on validating these sex-specific findings and evaluating targeted prehabilitation interventions to improve outcomes in patients undergoing TACE.

Abbreviations

AFP

Alpha-fetoprotein

BCLC

Barcelona Clinic Liver Cancer

BMI

Body mass index

CI

Confidence interval

CT

Computed tomography

HBV

Hepatitis B virus

HCC

Hepatocellular carcinoma

HCV

Hepatitis C virus

HR

Hazard ratio

HU

Hounsfield Unit

IQR

Interquartile range

NN

Non-sarcopenic/non-obese

NO

Non-sarcopenic/obese

OS

Overall survival

SMI

Skeletal muscle index

SN

Sarcopenic/non-obese

SO

Sarcopenic/obese

TACE

Transarterial chemoembolization

VFA

Visceral fat area

Authors’ contributions

Settawit Chatsuntiprapa: Conceptualization, Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Kittipitch Bannangkoon: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Writing – original draft, Writing – review & editing. Teeravut Tubtawee: Data curation, Formal analysis, Investigation, Writing – original draft. Natee Ina: Data curation, Formal analysis, Investigation, Writing – original draft.

Funding

This research received no external funding.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of the Faculty of Medicine, Prince of Songkla University and Songklanagarind Hospital (REC. 67-382-7-4, approved on 18 September 2024).

The requirement for written informed consent was waived by the Institutional Review Board of the Faculty of Medicine, Prince of Songkla University, due to the retrospective nature of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.