Changes in liver and spleen stiffness after transarterial chemoembolization for hepatocellular carcinoma and role in predicting hepatic decompensation

Abstract

BACKGROUND

Transarterial chemoembolization (TACE) alters hepatic perfusion and can potentially impact portal pressure in hepatocellular carcinoma (HCC) patients. Liver stiffness measurements (LSM) and spleen stiffness measurements (SSM) have emerged as promising tools for assessing portal hypertension. However, limited data exist regarding the effects of TACE on LSM and SSM. We hypothesized that TACE may influence short-term LSM and SSM and that baseline stiffness values can effectively identify patients at high risk for postprocedural hepatic decompensation.

AIM

To evaluate the short-term effects of TACE on stiffness measurements, and assess their baseline values for predicting 6-month hepatic decompensation.

METHODS

This prospective cohort study enrolled 102 HCC patients undergoing TACE. LSM and SSM measured by elastography at baseline, day 10, and day 30 post-procedure. A linear mixed-effects model was used to account for missing data with time as a fixed effect and participant identification as a random effect. Hepatic decompensation was defined as a composite endpoint of ascites, variceal bleeding, overt hepatic encephalopathy, and severe liver injury. Multiple logistic regression identified independent predictors of 6-month hepatic decompensation.

RESULTS

The linear mixed-effects model revealed no significant change in LSM and SSM values on days 10 and 30 after TACE compared to baseline. Within 6 months, 19 patients (18.6%) developed hepatic decompensation. Baseline SSM > 54 kilopascals (kPa) predicted hepatic decompensation with an area under the receiver operating characteristic curve of 0.83 [95% confidence interval (CI): 0.72-0.93], sensitivity of 63.2%, specificity of 90.4%. Multivariate analysis identified modified albumin-bilirubin (mALBI) grade ≥ 2B [adjusted odds ratio (aOR) = 7.05, 95%CI: 1.92-25.94, P = 0.003] and baseline SSM > 54 kPa (aOR = 16.52, 95%CI: 4.39-62.19, P < 0.001) as independent predictors of decompensation.

CONCLUSION

TACE does not significantly affect short-term stiffness measurements. However, baseline SSM can serve as an effective surrogate marker that complements the mALBI grade for stratifying hepatic decompensation risk.

Key Words: Cirrhosis; Hepatic decompensation; Hepatocellular carcinoma; Liver stiffness; Noninvasive tests; Portal hypertension; Spleen stiffness; Transarterial chemoembolization; Two-dimensional shear wave elastography

Core Tip: This prospective study evaluated the changes in liver stiffness measurements (LSM) and spleen stiffness measurements (SSM) assessed by two-dimensional shear wave elastography and transient elastography in patients with hepatocellular carcinoma undergoing transarterial chemoembolization. Both LSM and SSM showed no significant changes on days 10 and days 30 after the procedure. Importantly, baseline SSM was identified as a significant non-invasive predictor of 6-month hepatic decompensation. Incorporating baseline SSM (> 54 kilopascals) alongside the modified albumin-bilirubin grade can serve as a comprehensive preprocedural risk stratification tool, effectively bridging the prognostic gap, particularly in clinically homogeneous patients with preserved liver function.

INTRODUCTION

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, accounting for > 80% of all cases[1]. Overall, liver cancer is the sixth most prevalent cancer and the third leading cause of cancer-related deaths globally, with > 900000 new cases and 830000 deaths documented in 2020[2,3]. Transarterial chemoembolization (TACE) is a commonly used locoregional therapy for HCC that combines targeted intra-arterial chemotherapy with ischemic necrosis induced by arterial embolization. Despite its effectiveness, TACE can lead to hepatic decompensation, manifesting as complications such as ascites, variceal bleeding, hepatic encephalopathy, spontaneous bacterial peritonitis, and severe hepatotoxicity. The incidence of hepatic decompensation following TACE was 23%[4,5]. Several factors are associated with hepatic decompensation following TACE, including tumor burden [e.g., large tumor size, multiple nodules, vascular invasion, and high alpha-fetoprotein (AFP)] and underlying liver dysfunction[6-8]. Additionally, TACE can alter hepatic perfusion and potentially impact portal pressure, which may directly induce clinical liver decompensation, such as ascites or variceal bleeding[9-11]. However, the mechanisms by which TACE acutely affects liver stiffness measurements (LSM) and spleen stiffness measurements in the short term, as well as the precise relationship between these post-TACE parameters and the subsequent risk of decompensation, remain poorly understood, warranting further investigation to improve patient selection and postprocedural monitoring.

The hepatic venous pressure gradient (HVPG) is the gold standard for assessing portal pressure. However, its invasiveness and limited availability restrict its routine use in clinical practice[4,12]. In the era of noninvasive assessment of portal hypertension, LSM and SSM respectively, have gained increasing validation as noninvasive tests for the evaluation of portal hypertension in patients with chronic liver disease. These techniques offer several benefits, such as portability, cost-effectiveness, non-exposure to radiation, and capacity for dynamic monitoring[13]. Both LSM and SSM correlated well with HVPG and have been effective in predicting portal hypertension[14,15]. However, the correlation between LSM and HVPG becomes less reliable when HVPG > 1.6 kilopascals (kPa) (> 12 mmHg)[16], potentially due to the increasing influence of extrahepatic factors, such as hyperdynamic circulation and splanchnic vasodilation in patients with more advanced liver disease. Moreover, LSM alone may not reliably assess the response to a nonselective beta-blocker (NSBB)[16]. Conversely, SSM has emerged as a more reliable tool for evaluating portal hypertension, as it better reflects the hyperdynamic component of portal hypertension and has demonstrated superior predictive value for portal hypertension and related complications[17,18]. Furthermore, SSM correlated well with HVPG before transjugular intrahepatic portosystemic shunt (TIPS) and showed a significant decline within 24 hours after TIPS, whereas LSM does not exhibit a similar response[15].

Despite the clinical significance of these noninvasive tests, data on the effects of TACE on LSM and SSM remain limited. Thus, this study aimed to evaluate the effect of TACE on LSM and SSM and explore the association between baseline LSM and SSM values and the development of hepatic decompensation within 6 months following TACE, thereby contributing to a more practical risk stratification for clinicians managing HCC patients prior to TACE.

MATERIALS AND METHODS

Study design and participants

This prospective study enrolled consecutive patients with HCC who underwent TACE at the Chulalongkorn University Hospital, Bangkok, Thailand, between November 2024 and March 2025. Patients (aged ≥ 18 years) diagnosed with unresectable HCC based on typical radiologic characteristics observed in contrast-enhanced computed tomography or magnetic resonance imaging were enrolled[19]. However, those with prior liver transplantation, TIPS, splenic infarction, splenic vein thrombosis, myelofibrosis, splenectomy, contraindications for TACE, Child-Pugh score ≥ 9, or inability to perform LSM or SSM were excluded.

At baseline, patient demographics, laboratory parameters, clinical characteristics, etiology of HCC, tumor burden [e.g., tumor size, presence of vascular invasion, Barcelona Clinic Liver Cancer (BCLC) stage, and AFP levels], liver function, performance status, and previous HCC treatments were collected. Hepatic decompensation was defined as a composite of new or worsening ascites, variceal bleeding, overt hepatic encephalopathy, and grade 3-4 hepatotoxicity, as classified by the Common Terminology Criteria for Adverse Events of the National Cancer Institute[20]. Both hepatic decompensation events and mortality were assessed within 6 months following TACE.

Written informed consent was obtained from each participant. The study protocol was approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University (COA No. 1331/2024; IRB No. 0322/67), and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. The study protocol was registered in the Thai Clinical Trial Registry (No. TCTR20241021001) on October 21, 2024.

LSM and SSM by two-dimensional shear wave elastography

LSM and SSM were performed at baseline (before TACE) and on days 10 and days 30 after TACE. LSM and SSM were assessed using two-dimensional shear wave elastography (2D-SWE) with a convex broadband probe on the Supersonic Mach 30 system (Hologic, Aix-en-Provence, France), operated by a single experienced sonographer (Thaimai P) with experience in > 500 examinations.

All patients fasted for at least 8 hours under the recommendations of the European Federation of Societies for Ultrasound in Medicine and Biology[21]. The operator, guided by a real-time B-mode ultrasound image, targeted a region with optimal spatial resolution in a patient during apnea. Color elastography maps were generated to ensure complete and homogeneous filling of the selected area. A region of interest with a diameter of 15-20 mm was placed centrally within the color map, at a depth of approximately 15 mm below the capsule, and in a zone free of large vascular structures. The final LSM and SSM were calculated as the mean of five separate measurements, with their corresponding standard deviations, and reported in kPa. Measurements were deemed reliable if an interquartile range-to-median ratio was < 30%. LSM and SSM were evaluated at the same time points.

LSM by transient elastography

LSM was measured by transient elastography (TE) (FibroScan^® 630, Echosens France) by an experienced operator (Thaimai P) who had Echosens certificate training. The LSM was considered reliable if a minimum of 10 valid measurements were taken, and the interquartile range/median was < 30%[18,22,23]. TE was performed with the patient in the supine position, the right arm maximally abducted, and the transducer placed in the right intercostal spaces along the posterior axillary line[24,25] on the same day as 2D-SWE.

TACE

Super-selective, conventional TACE was performed using an emulsion consisting of 6-20 mL of lipiodol and 25-75 mg of doxorubicin, with the specific volume and dosage tailored according to the tumor’s size, number, and vascularity. The emulsion was injected into the tumor-feeding artery, ensuring that the injection site was located away from the origins of the gastroduodenal, right gastric, and cystic arteries. The volume of the emulsion delivered was adjusted based on the tumor uptake and angiographic appearance. Gel foam was used as the embolic agent in all patients[6].

Statistical analysis

Sample size calculation was performed based on the primary objective of evaluating changes in spleen stiffness after TACE. Due to the lack of prior data regarding post-TACE spleen stiffness dynamics, the calculation was based on a baseline mean spleen stiffness of 34.3 ± 13.4 kPa reported in a previous study[26]. A clinically relevant difference was defined as a change of at least 10% (approximately 4 kPa) from baseline. Assuming a correlation coefficient (r) of 0.5 between measurements, a power of 80% (Z_β = 0.842), and a two-sided significance level (α) of 5% (Z_α = 1.96), a minimum of 83 patients was required for a paired mean comparison. Accounting for a potential 20% loss to follow-up, the final target sample size was determined to be 100 patients. Continuous variables were expressed as mean ± SD, whereas categorical variables were presented as n (%). Continuous variables were compared using an independent samples t test, whereas categorical variables were analyzed using Pearson’s χ² test or Fisher’s exact test when appropriate. To ensure the completeness of the 6-month mortality data, the survival status of participants who were unable to be contacted was verified through the Thailand Civil Registration database.

To analyze longitudinal changes in LSM and SSM across the three time points (baseline, day 10, and day 30 following TACE) and effectively account for missing observations, a linear mixed-effects model was employed under the “missing at random” assumption. This model used time as a fixed effect, and participant ID as a random effect, allowing for greater flexibility and minimizing attrition bias.

The diagnostic performance of LSM and SSM was assessed by receiver operating characteristic (ROC) curve analysis, and the optimal cutoff values were determined by maximizing Youden’s index. Pairwise comparisons of ROC curves were conducted using the method established by DeLong et al[27]. Univariate and multivariate logistic regression analyses were conducted to identify independent predictors of 6-month hepatic decompensation. Variables with a P value of < 0.05 in the univariate analysis were selected for entry into the multivariable model using a backward elimination strategy based on the likelihood ratio test. To avoid multicollinearity and potential selection bias arising from overlapping parameters among baseline liver function scores [e.g., Child-Pugh class, Child-Pugh score, and modified albumin-bilirubin (mALBI) grade], we constructed three separate, parallel multivariate regression models. Model 1 adjusted for the modified ALBI grade, model 2 adjusted for the Child-Pugh score, and model 3 adjusted for the Child-Pugh class. In all models, highly correlated clinical scores were evaluated separately to prevent statistical distortion and ensure model validity. All statistical analyses were conducted using StataNow/SE version 19.0 (StataCorp LLC, College Station, TX, United States). All tests were two-sided and P values < 0.05 were considered significant. All statistical methods used in this study were reviewed by a biomedical statistician from the Faculty of Medicine, Chulalongkorn University.

RESULTS

Patient characteristics

In this study, 102 of the 114 patients with HCC who underwent TACE were consecutively enrolled, whereas 12 were excluded because of unsuccessful LSM/SSM. The flowchart of patient enrollment is shown in Figure 1. The mean patient age was 63.8 ± 10.1 years, and the majority were male (84.3%, n = 86). Based on the BCLC staging, 32 (31.4%), 62 (60.7%), and 8 (7.8%) patients had BCLC stage A, BCLC stage B, and BCLC stage C, respectively. Among them, 87 (85.3%) and 15 (14.7%) patients were classified as having Child-Pugh A and Child-Pugh B, respectively. Regarding the etiology of HCC, 41 (40.2%) patients had hepatitis B virus infection, 25 (24.5%) had hepatitis C virus (HCV) infection, 9 (8.8%) had metabolic dysfunction-associated steatotic liver disease, and 17 (16.7%) had an alcohol-related liver disease. Moreover, 49% of the patients had previously received locoregional treatment, whereas 46.1% (n = 47) were untreated. Additionally, 24.5% of the patients were on NSBB. The baseline characteristics of the enrolled patients are shown in Table 1.

Figure 1 Flowchart of patient enrollment. HCC: Hepatocellular carcinoma; TACE: Transarterial chemoembolization; TIPS: Transjugular intrahepatic portosystemic shunt.

Table 1 Baseline characteristics of hepatocellular carcinoma patients undergoing transarterial chemoembolization, mean ± SD/n (%).

Characteristics	n = 102
Age (years)	63.8 ± 10.1
Males	86 (84.3)
Body mass index (kg/m2)	24.1 ± 4.2
Etiology
Hepatitis B virus infection	41 (40.2)
Hepatitis C virus infection	25 (24.5)
MASLD	9 (8.8)
Alcohol-related liver disease	17 (16.7)
BCLC stage
A	32 (31.4)
B	62 (60.7)
C	8 (7.8)
Previous treatment
No previous treatment	47 (46.1)
Locoregional treatment	50 (49.0)
Systemic therapy	1 (1.0)
Liver resection or lobectomy	4 (3.9)
Tumor characteristics
Tumor size (largest diameter)
< 5 cm in size	71 (69.6)
5-8 cm in size	16 (15.7)
> 8 cm in size	15 (14.7)
Unilobar	69 (67.6)
Bilobar	33 (32.4)
Single nodule	33 (32.4)
Multiple nodules	69 (67.6)
Non-selective beta-blocker use	25 (24.5)
Spleen diameter (cm)	10.9 ± 2.8
Platelet count (μL)	153922 ± 91509
Alpha-fetoprotein (IU/mL)	4046 ± 19144
Portal vein thrombosis	22 (21.6)
mALBI grade
Grade 1	39 (38.2)
Grade 2a	30 (29.4)
Grade 2b	26 (25.5)
Grade 3	7 (6.9)
MELD score	9.2 ± 3.0
Child-Pugh class
A	87 (85.3)
B	15 (14.7)

BCLC: Barcelona Clinic Liver Cancer; mALBI: Modified albumin-bilirubin; MASLD: Metabolic dysfunction-associated steatotic liver disease; MELD: Model for end-stage liver disease.

LSM and SSM at baseline and on days 10 and days 30 after TACE

Using a linear mixed-effects model to evaluate longitudinal changes, the estimated marginal mean of LSM on 2D-SWE remained stable without significant changes over time, measuring 26.53 kPa [95% confidence interval (CI): 23.29-29.76] at baseline, 26.49 kPa (95%CI: 23.14-29.84) on days 10, and 25.88 kPa (95%CI: 22.61-29.14) on days 30 after TACE (P = 0.979 for days 10 vs baseline and P = 0.640 for days 30 vs baseline). Similarly, the estimated marginal mean SSM on 2D-SWE showed no significant change, ranging from 38.98 kPa (95%CI: 35.77-42.19) at baseline to 36.77 kPa (95%CI: 33.42-40.11) on days 10 and 40.99 kPa (95%CI: 37.74-44.25) on days 30 post-procedure (P = 0.159 for days 10 vs baseline and P = 0.187 for days 30 vs baseline). Figure 2 demonstrates the estimated marginal means with 95%CIs of LSM and SSM on 2D-SWE at baseline and post-TACE across these time points.

Figure 2 Stiffness measurement by two-dimensional shear wave elastography prior to transarterial chemoembolization, days 10 and days 30 after transarterial chemoembolization. A: Liver stiffness measurement by two-dimensional shear wave elastography (2D-SWE) prior to transarterial chemoembolization (TACE), days 10 and days 30 after TACE; B: Spleen stiffness measurement by 2D-SWE prior to TACE, days 10 and days 30 after TACE. P values were determined using a linear mixed-effects model. 2D-SWE: Two-dimensional shear wave elastography; LSM: Liver stiffness measurement; SSM: Spleen stiffness measurement.

Consistent with these findings, LSM assessed via TE also showed no significant differences compared to baseline, with estimated marginal values of 29.04 kPa (95%CI: 25.00-33.07) at baseline, 31.39 kPa (95%CI: 27.21-35.56) on days 10, and 29.39 kPa (95%CI: 25.33-33.46) on days 30 (P = 0.191 for days 10 vs baseline and P = 0.836 for days 30 vs baseline) (Figure 3). Subgroup analyses based on Child-Pugh classification (Supplementary Table 1) and NSBB use (Supplementary Table 2) revealed no significant differences in LSM and SSM as assessed by 2D-SWE and TE following TACE.

Figure 3 Liver stiffness measurement by transient elastography prior to transarterial chemoembolization, days 10 and days 30 after transarterial chemoembolization. P values were determined using a linear mixed-effects model. LSM: Liver stiffness measurement; TE: Transient elastography.

Baseline and changes in LSM and SSM, and risk of hepatic decompensation

During the 6-month follow-up period, 6 deaths and 19 (18.6%) cases of hepatic decompensation were recorded. Among these, worsening ascites occurred in 8 (42.1%) patients, variceal bleeding in 1 (5.2%), overt hepatic encephalopathy in 1 (5.2%), and grade 3 to grade 4 hepatotoxicity in 9 (47.3%). The mean time to hepatic decompensation was 31.2 days. Of the 19 patients who experienced decompensation, 3 died within 6 months. Patients who experienced hepatic decompensation had significantly higher baseline LSM (37.8 ± 19.2 kPa vs 24.0 ± 15.6 kPa, P = 0.001) and SSM (56.3 ± 16.4 kPa vs 35.0 ± 14.8 kPa, P < 0.001) by 2D-SWE than those who did not experience decompensation. Likewise, baseline LSM by TE was significantly higher in patients who experienced hepatic decompensation than in those who did not experience it (60.8 ± 21.7 kPa vs 46.6 ± 21.5 kPa, P = 0.011). Regarding changes in LSM and SSM measured by 2D-SWE from baseline, SSM progression (> 10% increase) was observed in 28 (31.5%) patients on days 10 and in 50 (51.0%) on day 30. LSM progression (> 10% increase) occurred in 32 (36.0%) patients on days 10 and in 36 (36.7%) on days 30. Progression in serial LSM or SSM was not significantly associated with hepatic decompensation (P > 0.05).

Performance of baseline LSM and SSM in predicting hepatic decompensation

Baseline LSM and SSM evaluated by 2D-SWE and TE demonstrated good predictive accuracy. The areas under the ROC curve (AUROCs) in predicting hepatic decompensation were 0.72 (95%CI: 0.59-0.85) for LSM by 2D-SWE, 0.83 (95%CI: 0.72-0.93) for SSM by 2D-SWE, and 0.85 (95%CI: 0.77-0.93) for LSM by TE. Pairwise comparisons indicated no significant difference in predictive performance among these three modalities (all P > 0.05). Supplementary Figure 1A shows the ROC curve analysis of LSM and SSM in predicting hepatic decompensation in patients with HCC who underwent TACE.

Furthermore, the predictive performance of baseline SSM for hepatic decompensation after TACE was compared with traditional clinical scoring systems. The AUROCs of baseline SSM, mALBI grade, Child-Pugh score, and model for end-stage liver disease (MELD) score were 0.83 (95%CI: 0.72-0.93), 0.73 (95%CI: 0.61-0.86), 0.68 (95%CI: 0.55-0.81), and 0.66 (95%CI: 0.51-0.80), respectively. Pairwise comparisons using DeLong’s test revealed no significant differences between the AUROCs of SSM and those of the mALBI grade (P = 0.421) or Child-Pugh score (P = 0.123). However, baseline SSM demonstrated a significantly superior predictive performance compared to the MELD score (P = 0.022). These results underscore that SSM is a highly robust non-invasive indicator, providing predictive value that is either comparable or superior to established clinical models in identifying HCC patients at high risk for decompensation following TACE (Supplementary Figure 1B).

By maximizing Youden’s index to determine the optimal cutoff values, a baseline LSM by 2D-SWE > 42 kPa was identified as the best threshold for predicting hepatic decompensation within 6 months following TACE, achieving an accuracy of 76.4%, sensitivity of 47.4%, specificity of 83.1%, positive predictive value (PPV) of 39.1%, and negative predictive value (NPV) of 87.3%. For SSM by 2D-SWE, the optimal cutoff was 54 kPa, resulting in an accuracy of 85.3%, sensitivity of 63.2%, specificity of 90.4%, PPV of 60.0%, and NPV of 91.5%. In comparison, LSM measured by TE had an optimal threshold of 24.5 kPa, with corresponding values of 70.5%, 94.7%, 65.1%, 38.3%, and 98.2%, respectively. Among all methods, SSM by 2D-SWE > 54 kPa demonstrated the highest overall accuracy for predicting hepatic decompensation.

Factors related to hepatic decompensation following TACE

Univariate analysis identified several factors significantly associated with hepatic decompensation following TACE, including male, HCV infection, NSBB use, splenic length, Child-Pugh score, Child-Pugh grade, mALBI grade ≥ 2B, SSM by 2D-SWE > 54 kPa, and LSM by 2D-SWE > 42 kPa. Subsequently, three separate, parallel multivariate regression models were constructed to account for overlapping baseline liver function scores (Table 2). In model 1, adjusting for the modified ALBI grade, mALBI grade ≥ 2B [adjusted odds ratio (aOR) = 7.05; 95%CI: 1.92-25.94; P = 0.003] and baseline SSM by 2D-SWE > 54 kPa (aOR = 16.52; 95%CI: 4.39-62.19; P < 0.001) emerged as independent factors related to hepatic decompensation. The reliable independent predictive value of baseline SSM > 54 kPa was consistently maintained when adjusting for the continuous Child-Pugh score in model 2 (aOR = 18.59; 95%CI: 5.11-67.66; P < 0.001) and the categorical Child-Pugh class in model 3 (aOR = 17.71; 95%CI: 5.04-62.19; P < 0.001). To evaluate the robustness of these findings given the cohort size, a post-hoc power analysis was performed based on the observed decompensation rates (8.5% in the SSM ≤ 54 kPa group vs 60.0% in the SSM > 54 kPa group). With the total sample size of 102 patients, the study achieved a statistical power of 99.8% at a significance level (α) of 0.05, confirming that the study was adequately powered to detect these significant differences.

Table 2 Predictors of hepatic decompensation in hepatocellular carcinoma patients after transarterial chemoembolization.

Variables	Univariate analysis		Multivariate analysis model 1		Multivariate analysis model 2		Multivariate analysis model 3
	OR (95%CI)	P value	aOR (95%CI)	P value	aOR (95%CI)	P value	aOR (95%CI)	P value
Age, years	1.01 (0.96-1.06)	0.699
Female (%)	3.37 (1.04-10.87)	0.042
Body mass index (kg/m2)	0.97 (0.86-1.09)	0.582
Etiology (%)
Hepatitis B virus infection	Reference
Hepatitis C virus infection	4.05 (1.17-14.02)	0.027
MASLD	0.90 (0.09-8.80)	0.928
Alcohol	0.96 (0.17-5.51)	0.963
Cryptogenic	2.06 (0.33-12.81)	0.439
Others	0 (0)	1
BCLC stage (%)
A	Reference
B	2.82 (0.75-10.65)	0.126
C	3.22 (0.44-23.65)	0.250
Previous treatment (%)
No previous treatment	Reference
Locoregional treatment	0.80 (0.28-2.30)	0.684
Systemic therapy	0 (0)	1.000
Liver resection	1.41 (0.13-15.16)	0.778
Number of TACE	1.13 (0.84-1.51)	0.418
Tumor size (cm)	1.05 (0.92-1.18)	0.489
Beta-blocker use (%)	3.77 (1.31-10.8)	0.014
Splenic length (mm)	1.03 (1.01-1.05)	0.006
Platelet count (μL)	1.00 (1.00-1.00)	0.084
Alpha-fetoprotein (IU/mL)	1.00 (1.00-1.00)	0.295
Portal vein thrombosis (%)	1.93 (0.64-5.87)	0.245
MELD score	1.08 (0.94-1.25)	0.290
mALBI grade ≥ 2B	6.83 (2.29-20.31)	< 0.001	7.05 (1.92-25.94)	0.003
Child-Pugh score	2.02 (1.20-3.41)	0.008			2.24 (1.20-4.19)	0.011
Child-Pugh grade
A	Reference						Reference
B	3.79 (1.16-12.47)	0.028					4.73 (1.10-20.38)	0.037
SSM by 2D-SWE above 54 kPa	16.07 (4.92-52.48)	< 0.001	16.52 (4.39-62.19)	< 0.001	18.59 (5.11-67.66)	< 0.001	17.71 (5.04-62.19)	< 0.001
LSM by 2D-SWE above 42 kPa	4.44 (1.52-12.91)	0.006

aOR: Adjusted odds ratio; OR: Odds ratio; CI: Confidence interval; BCLC: Barcelona Clinic Liver Cancer; MASLD: Metabolic dysfunction-associated steatotic liver disease; MELD: Model for end-stage liver disease; kPa: Kilopascals; TACE: Transarterial chemoembolization; mALBI: Modified albumin-bilirubin; 2D-SWE: Two-dimensional shear wave elastography; LSM: Liver stiffness measurement; SSM: Spleen stiffness measurement.

In a subgroup analysis focusing on patients with preserved liver function (Child-Pugh class A, n = 87), baseline SSM > 54 kPa remained a highly significant predictor of hepatic decompensation (OR = 21.54; 95%CI: 5.25-88.37; P < 0.001). This confirms the robust predictive value of SSM even within a clinically homogeneous group of patients with preserved liver function.

DISCUSSION

This study investigated LSM and SSM changes within 30 days in patients with HCC treated with TACE and evaluated their utility in predicting hepatic decompensation following TACE. Our primary longitudinal analysis revealed that neither LSM nor SSM exhibited significant short-term changes following TACE. However, patients who experienced hepatic decompensation had significantly higher baseline LSM and SSM values than those who did not experience it. Furthermore, baseline LSM and SSM by 2D-SWE and LSM measured by TE showed a good predictive accuracy for hepatic decompensation. In addition, a baseline mALBI grade ≥ 2B and baseline SSM by 2D-SWE > 54 kPa were independent factors related to hepatic decompensation following TACE. These findings support the potential clinical value of SSM particularly when measured by 2D-SWE as a noninvasive tool for risk stratification and early identification of patients at higher risk of hepatic decompensation following TACE.

The stability of both LSM and SSM observed within 30 days after the procedure provides insight into post-TACE portal hemodynamics. Previous studies investigating the effect of TACE on portal pressure have reported conflicting results. Okada et al[10] demonstrated that 50% of patients with HCC experienced an increase in esophageal variceal pressure, as assessed by an endoscopic pneumatic pressure sensor. Additionally, 88.9% of patients exhibited a significant increase in portal blood flow measured by doppler ultrasonography, 3 days after TACE, compared with baseline. Similarly, Moriyasu et al[11] found that portal blood flow increased 1 week after transcatheter arterial embolization (TAE), although no significant changes were observed 4 weeks after TAE. In contrast, Scheiner et al[5] reported that TACE had neither acute nor intermediate effects on HVPG within 2 months. Nonetheless, repeated TACE was associated with a significant long-term increase in HVPG at 6 months. Likewise, Elia et al[9] found no change in HVPG 3 days after TACE. The findings of the present study are consistent with these latter reports. In this study, no significant changes were noted in either LSM or SSM within 10 days or 30 days following TACE, suggesting that TACE might not significantly affect portal pressure during the 30-day follow-up. The discrepancies among previous studies may be attributed to differences in the methods used to assess portal hypertension and variations in the timing of post-TACE evaluations.

To gain deeper clinical insights, subgroup analyses based on baseline liver function and medical therapy were performed. After stratifying patients according to baseline hepatic reserve, those in Child-Pugh class B presented with noticeably higher baseline LSM than those in Child-Pugh class A, reflecting more advanced background cirrhosis. However, regardless of the baseline hepatic reserve, post-TACE LSM and SSM values remained remarkably stable across both Child-Pugh cohorts, showing no significant variations at days 10 and days 30 after the procedure. This high consistency demonstrates that TACE-induced localized ischemia does not trigger widespread acute parenchymal injury or aggravate background liver stiffness, confirming an excellent short-term hepatic safety profile even among patients with compromised hepatic reserve.

Regarding concomitant medications, subgroup analyses based on NSBB therapy also revealed no significant differences in post-TACE LSM and SSM trajectories. Both the NSBBs and the non-NSBB cohorts exhibited stable stiffness measurements throughout the 30-day follow-up period. This observed stability across both groups further supports the short-term clinical safety of performing TACE, suggesting that the procedure does not acutely disrupt portal hemodynamics or splanchnic circulation, irrespective of concurrent NSBB use. Nevertheless, these subgroup findings should be interpreted cautiously due to the limited sample size in the medication cohort, and warrant further validation in larger targeted studies.

The overall rate of hepatic decompensation following TACE was 18.6%, with worsening ascites being the most common complication in the present study. This finding aligns with the results of previous studies that have documented decompensation rates ranging from 2.4% to 23%[4-6]. Previous studies have identified tumor burden (e.g., large tumor size and high serum AFP) and impaired hepatic reserve, such as low serum albumin[6], albumin-bilirubin score[8] or impaired indocyanine green retention[4] as significant risk factors for hepatic decompensation following TACE. To our knowledge, no studies have specifically investigated the roles of LSM and SSM in predicting hepatic decompensation following TACE. This study provides novel insights, demonstrating that LSM and SSM are valuable in predicting post-TACE hepatic decompensation, with an accuracy rate ranging from 72.1% to 84.7%. Baseline SSM by 2D-SWE > 54 kPa predicted hepatic decompensation with an accuracy of 85.3%, PPV of 60.0%, and NPV of 91.5%. Moreover, baseline SSM > 54 kPa and mALBI grade ≥ 2B were independent factors related to post-TACE hepatic decompensation. Notably, the independent predictive value of SSM > 54 kPa remained highly consistent across separate adjusted models accounting for different clinical liver function scores (mALBI grade, Child-Pugh score, and Child-Pugh class). The discrepancies between our findings and those of previous studies may be explained by the differences in study design, definitions of hepatic decompensation, and patient characteristics. For instance, Kohla et al[6] defined decompensation based on changes in the Child-Pugh score, whereas the present study used a composite endpoint of worsening ascites, variceal bleeding, overt hepatic encephalopathy, and severe hepatotoxicity, using Common Terminology Criteria for Adverse Events criteria to define post-TACE liver failure within 6 months[4,20]. Furthermore, the analyzed cohort had a smaller tumor burden than that reported by Kohla et al[6] (69.6% vs 20% with tumors < 5 cm). Previous predictive models have primarily focused on hepatic function or tumor burden. In contrast, this study highlights the independent contribution of SSM a surrogate for portal hypertension alongside mALBI grade, indicating that the risk of hepatic decompensation is multifactorial. These findings underscore the importance of incorporating both hepatic reserve (mALBI grade) and hemodynamic status SSM into preprocedural risk stratification to more accurately identify patients at high risk.

The clinical utility of SSM in this context is further demonstrated when compared to traditional clinical scoring systems. While the Child-Pugh and MELD scores are established standards for assessing hepatic reserve and mortality risk, they primarily focus on markers of synthetic function. However, post-TACE hepatic decompensation is often driven by acute hemodynamic stress rather than a simple decline in liver function alone. Our results support this, as baseline SSM (AUROC = 0.83) significantly outperformed the MELD score (AUROC = 0.66, P = 0.022).

Rather than completely replacing traditional staging systems, baseline SSM should be viewed as a complementary non-invasive tool that adds incremental prognostic value, particularly by capturing the vascular changes of portal hypertension that metabolic scores might miss. This complementary role is especially useful for identifying “high-risk” individuals among patients with preserved liver function (e.g., Child-Pugh A) who may still have underlying significant subclinical portal hypertension. Our subgroup analysis of 87 Child-Pugh class A patients strongly supports this, revealing that those with baseline SSM > 54 kPa had a 21-fold increased risk of developing decompensation (OR = 21.54, P < 0.001). Although baseline SSM showed high diagnostic accuracy, its prognostic superiority was not uniformly demonstrated across all statistical comparisons. Therefore, integrating SSM alongside conventional clinical scores offers a more comprehensive, multifactorial preprocedural evaluation in routine clinical practice.

This study has several key strengths. To our knowledge, it is the first prospective investigation to establish the role of SSM in predicting hepatic decompensation following TACE. Despite the single-center design, a post-hoc power analysis of 99.8% confirms that the sample size was highly sufficient to detect significant clinical differences. Furthermore, baseline SSM demonstrated superior predictive performance compared to the conventional MELD score (P = 0.022), highlighting its clinical utility for personalized risk stratification.

However, some limitations must be acknowledged. First, HVPG measurements were not performed owing to the invasiveness of the procedure and the impracticality of repeating it three times. However, multiple studies have confirmed the association among LSM, SSM, and HVPG[15,28,29]. Second, although standardized conventional TACE techniques using lipiodol and gelfoam were performed in all patients, the specific volumes and dosages were tailored based on individual tumor characteristics, which might have introduced subtle variations in postprocedural ischemic effects. However, to minimize potential bias, the intervention radiologists who conducted TACE were completely blinded and were unaware of LSM and SSM data. Further studies are warranted to evaluate whether alternative modalities, such as drug-eluting bead TACE, would yield distinct stiffness trajectories. Third, our study relied on a composite endpoint for hepatic decompensation, which combines clinical events with different severities and weights, ranging from reversible grade 3-4 hepatotoxicity to life-threatening variceal bleeding. This clinical heterogeneity may influence how the observed associations are interpreted and warrants cautious application in routine clinical practice. Finally, given that this was a single-center study, the proposed optimal baseline SSM cutoff (> 54 kPa) should be interpreted with caution. External validation in larger, independent multicenter cohorts is necessary to confirm its general applicability and diagnostic reproducibility.

CONCLUSION

The present study revealed that short-term liver stiffness measured through both 2D-SWE and TE, as well as spleen stiffness measured via 2D-SWE, remained stable without significant variations within 30 days after TACE. Nonetheless, a mALBI grade ≥ 2B and a baseline SSM > 54 kPa were identified as key preprocedural predictors of hepatic decompensation following TACE. In conclusion, baseline SSM can be an effective, non-invasive surrogate marker that reflects the severity of portal hypertension and complements conventional clinical scores when stratifying the risk of short-term hepatic decompensation after TACE. Nonetheless, further large-scale comparisons are needed to determine its precise incremental value.

ACKNOWLEDGEMENTS

We thank the staff of the Division of Gastroenterology and Hepatology, Excellence Center in Liver Diseases, Center of Excellence in Hepatic Fibrosis and Cirrhosis, Faculty of Medicine, Chulalongkorn University, and King Chulalongkorn Memorial Hospital, Thai Red Cross Society, for their technical assistance and clinical support.