RPS 601 - Chronic liver disease and portal hypertension

Author Block: D. Chaurasia, R. BATRA, R. Dixit; New Delhi/IN
Purpose: The AASLD adopts Baveno VI/VII consensus and recommends specific cut-offs for non-invasive assessment of portal hypertension (PH) and varices using elastography in adults; however, there are no established guidelines for the pediatric population due to limited high-quality data. This study aimed to evaluate the utility of hepatic stiffness (HS) and splenic stiffness (SS) values in predicting PH and assessing the risk of varices in children with chronic liver disease (CLD) using 2D shear wave elastography (2D-SWE).
Methods or Background: This cross-sectional analytical study included 64 participants under 18 years of age, comprising 34 children with CLD (26 children with PH and 11 children with varices) and 30 age-matched healthy controls. Ultrasound of the liver and spleen was performed using a curvilinear (C1-6) probe on a GE LOGIQ S8 system in a fasting state, and ten 2D-SWE measurements (in m/s) were obtained. Mean values were calculated and considered reliable if interquartile range to median ratio was ≤15%. HS and SS values were analyzed, and ROC curves were drawn to predict PH. The splenic-to-hepatic stiffness (S/L) ratio was calculated to differentiate patients with varices.
Results or Findings: HS and SS values were significantly (p < 0.05) higher in children with PH, and the S/L ratio was significantly (p < 0.05) elevated in those with varices as compared to those without.
Conclusion: SS can serve as a non-invasive surrogate marker for PH in children. The S/L ratio demonstrated superior performance over HS or SS values alone and may aid both in screening and in identifying children at increased risk of varices, allowing for tailored monitoring and timely intervention.
Limitations: Small sample size and no validated pediatric cut-offs. Larger, multicenter studies are needed to confirm the clinical utility in pediatric risk stratification.
Funding for this study: The study was institutionally supported by the Department of Radiology, Maulana Azad Medical College.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institutional Ethics Committee of Maulana Azad Medical College, New Delhi.

Author Block: A. M. Morariu-Barb, M. A. Socaciu, B. Boca, H. Stefanescu, B. Procopet, M. Lupsor-Platon; Cluj-Napoca/RO
Purpose: Clinically significant portal hypertension(CSPH) is the main determinant of complications in chronic liver disease(CLD). The gold standard is hepatic venous pressure gradient (HVPG)(>10 mmHg), but this method is invasive and not widely available. Standard CT offers little value beyond morphological assessment, but CT radiomics may represent a more accurate alternative.This study investigated whether CT-based radiomics can predict CSPH, comparing segmentation strategies (2Dvs.3D) and regions of interest (liver,spleen or both).
Methods or Background: We retrospectively analyzed 52 patients with biopsy-proven CLD who underwent HVPG and contrast-enhanced CT within three months. Nine healthy controls were included based on FibroScan F0 values and absence of signs of portal hypertension.
2D and 3D segmentations of liver and spleen were performed in portal-venous phase using Slicer software. Radiomic features were extracted and neural networks were trained to classify CSPH. Six models were developed: 2Dliver, 2Dspleen, 2Dliver+spleen, 3Dliver, 3Dspleen and 3Dliver+spleen. Performance was evaluated using area under the receiver operating characteristic curve(AUROC) and accuracy.
Results or Findings: CSPH was present in 26 patients(42.6%).The combined 3D liver and spleen model achieved the highest performance with an AUROC of 0.899 and accuracy of 66.7%. The 3Dspleen and 3Dliver models followed, with AUROCs of 0.811 and 0.816 and accuracies of 81.4% and 68%.Among the 2D models, the combined liver and spleen model performed best (AUROC 0.825,accuracy 53.3%), followed by the 2Dspleen model (AUROC 0.757,accuracy 66.7%) and the 2Dliver model (AUROC 0.703,accuracy 52.9%).
Conclusion: CT-based radiomic analysis of liver and spleen predicted CSPH with good performance.3D segmentations and combined liver-spleen models outperformed 2D and single-organ models. These findings support CT radiomics as a promising non-invasive tool for CSPH detection and prognostic stratification, extending the utility of routine CT.
Limitations: Single-center study
Small size cohort
Lack of external validation
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: S. Rivera Rojas, S. Morón Hodge, N. Saturio Galán, P. Castillo Grau, M. Cuadros Martínez, M. E. Velilla Aparicio, M. Cabero, E. Rodríguez, C. Martín Hervás; Madrid/ES
Purpose: The Fontan procedure is a surgical palliation for patients with a univentricular heart. Hemodynamic changes after the procedure lead to hepatic fibrosis and increase the risk of hepatocellular carcinoma (HCC), highlighting the need for multidisciplinary follow-up with a key role for radiology. This study aims to describe the radiological findings associated with Fontan-associated liver disease (FALD) and HCC, and to document the time to onset of these findings.
Methods or Background: Retrospective cohort study of 87 patients at a referral hospital with abdominal imaging available in the electronic medical record (ultrasound, MRI, and/or CT). All imaging studies were reviewed. We recorded the interval from the Fontan procedure to the first imaging study, the time to diagnosis of FALD, their frequency, and the prevalence of HCC based on radiological and histological criteria.
Results or Findings: Among the 87 patients, 59.8% were male. During follow-up 92.0% of patients were diagnosed with FALD, with a mean age of 29 years (SD 8.5). Median interval from Fontan procedure to the first imaging study was 17.5 years (IQR 12). FALD-related findings were observed on the initial imaging study in 77.0%. Radiological features of FALD included heterogeneous liver parenchyma (67.4%), irregular hepatic margins (61.2%), caudate lobe hypertrophy (55.3%), splenomegaly (52.3%), suprahepatic vein dilation (33.7%), and ascites (16.3%). Nodules were detected in 43.0% of patients. Nodules suggestive of HCC developed in 6 patients (6.9%), with histological confirmation in 3 ( 3.4%).
Conclusion: Radiological findings associated with FALD are common from the first imaging study in patients post-Fontan. Although HCC is infrequent, it can develop in a subset, emphasizing the importance of standardized long-term imaging surveillance for early detection and management.
Limitations: This was a retrospective cohort study.
Funding for this study: None.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: S. Hong, J. E. Kim; Jinju/KR
Purpose: This study assessed whether dual-energy CT (DECT)–derived liver extracellular volume fraction (fECV) can reliably evaluate liver functional reserve in cirrhotic patients, as determined by Child–Pugh classification (CP class), compared with other non-invasive markers.
Methods or Background: This retrospective study included 258 patients with liver cirrhosis. fECV score was measured from iodine maps derived from equilibrium-phase images at 100/140 Sn kVp. The diagnostic performance of fECV and other non-invasive markers (aspartate transaminase-to-platelet ratio index [APRI], Fibrosis-4 [FIB-4], Model for End-Stage Liver Disease [MELD]) for differentiating CP classes was assessed by comparing the area under the receiver operating characteristic curve (AUC) using DeLong’s test with Holm correction for multiple comparisons.
Results or Findings: The fECV score was positively correlated with CP class (r = 0.53, p < 0.01). For differentiating CP class A from B, the AUCs were similar among fECV (0.84), APRI (0.83; p = 1), and FIB-4 (0.83; p = 1), whereas MELD demonstrated a significantly higher AUC (0.94; p < 0.05). For differentiating CP classes B and C, the fECV score demonstrated a significantly higher AUC (0.78), compared with FIB-4 (0.50, p = 0.04) and APRI (0.49, p = 0.04), whereas no significant difference was observed between fECV and MELD (0.92, p = 0.12).
Conclusion: Dual energy CT-derived fECV score is a useful non-invasive marker for assessing liver functional reserve based on the CP class.
Limitations: First, this study was a retrospective study conducted at a single institution. Second, iodine quantification results may vary depending on the type of DECT system. Third, there was a relatively small number of patients with Child–Pugh class C (n = 13) compared with classes A (n = 207) and B (n = 38).
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was approved by Gyeongsang National University Hospital review board (IRB No. 2025-05-005).

Author Block: J. Jin¹, C. Huang², Y. Ma¹, W. Zhang¹, Y. Tan¹, S. Chen², J. Ren¹; ¹Guangzhou/CN, ²Rochester, MN/US
Purpose: Liver microcirculatory dysfunction plays a crucial role in the development and progression of cirrhosis. There is an urgent clinical need for effective methods to assess hepatic microcirculation. This study explores the feasibility of in vivo ultrasound localization microscopy (ULM) imaging and quantitative analysis in humans. We aim to qualitatively and quantitatively investigate the similarities and differences in hepatic microcirculation between healthy volunteers and patients with cirrhosis using ULM.
Methods or Background: This study consists of two parts. Part I involves in vivo comparative imaging of normal volunteers using Color Doppler (CD), ultra-micro angiography (UMA), and ULM to validate the clinical feasibility of ULM by assessing image resolution. Part II compares microcirculatory characteristics between normal and cirrhotic livers based on ULM-derived images and quantitative parameters.
Results or Findings: Compared with CD and UMA, ULM demonstrates superior capability in detecting subcapsular hepatic microcirculation. Among the quantitative parameters based on total blood flow, vessel density (VD), velocity (V), perfusion index (PI), distance metric (DM), sum of angles metric (SOAM), number of vessel branches per unit area (NOBPA) and number of branch points per unit area (NOBPPA) were higher in cirrhotic group than in control group. Overall, higher density and perfusion, higher tortuosity of microvessels were revealed in cirrhotic groups.
Conclusion: ULM enables in vivo, non-invasive, visual, and quantitative assessment of hepatic microcirculation, holding promise as a pivotal tool for early diagnosis of cirrhosis and monitoring therapeutic efficacy.
Limitations: This study did not conduct a longitudinal comparison of microvascular imaging and quantitative microcirculation parameters across different stages of liver fibrosis. We aim to include patients with varying degrees of liver fibrosis in order to better characterize the overall progression of microcirculatory changes throughout the development of liver fibrosis.
Funding for this study: National Natural Science Foundation of China (No. 82202192)
Basic and Applied Basic Research Foundation of Guangdong Province (No. 2021A1515110591)
Guangzhou Municipal Science and Technology Project (No. SL2024A04J01248)
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the Institutional Review Board (IRB) of the Third Affiliated Hospital of Sun Yat-Sen University.

Author Block: J. Li¹, J. Lei¹, B. Wang¹, X. Zhang²; ¹Lanzhou/CN, ²Xi An/CN
Purpose: To develop and validate a noninvasive dual-layer spectral computed tomography (DSCT)-based nomogram model integrating quantitative multiparameters and clinical indicators to optimize risk assessment and hemorrhage prediction of esophageal varices (EVs) in patients with cirrhosis.
Methods or Background: This retrospective study enrolled 284 cirrhotic patients undergoing DSCT , divided into training (n=194) and validation (n=90) cohorts. Patients were stratified according to the presence of high-risk esophageal varices (HREV), red color signs (RC signs), and esophageal variceal bleeding (EVB), with endoscopy as reference. Quantitative parameters were analyzed, including extracellular volume fraction (ECV) and iodine washout rate (IWR) of liver and spleen, and esophageal variceal diameter (EVD). Multivariable logistic regression identified predictors for nomogram construction. Model performance was evaluated via AUC, calibration curves, decision curve analysis (DCA), and predictive metrics [(sensitivity, specificity, positive/negative predictive value (PPV/NPV) ]in both cohorts.
Results or Findings: EVD, ECVliver, ECVspleen, and IWRliver differed significantly across subgroups stratified by HREV, RC(+), and EVB status (P < 0.05). Multivariate analysis identified EVD, ECVspleen, and IWRliver as independent predictors of HREV; EVD, ECVspleen, IWRliver, and total bilirubin for RC sign; and EVD, ECVspleen, aspartate aminotransferase, and red blood counts for EVB (all P < 0.05). The nomogram demonstrated high predictive accuracy for HREV (AUC: 0.963/0.968), RC signs (0.898/0.854), and EVB (0.847/0.785) in training/validation cohorts. DCA confirmed clinical utility, with net benefit across low-to-medium risk thresholds (0.0-0.6). At EVD thresholds of 5 mm and 3 mm, the model achieved specificity/PPV of 97.1-100.0%/97.7-100.0% and sensitivity/NPV of 94.4-94.2%/88.7-91.2% for HREV in the training and validation cohorts, respectively.
Conclusion: This DSCT-based nomogram synergizes structural (ECVspleen), functional (IWRliver), and morphological (EVD) biomarkers to noninvasively stratify risks of HREV, RC signs, and EVB in cirrhosis.
Limitations: Lack external validation.
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: F. Delavari, C. Dromain, L. Haefliger, L. Fernandes, M. Fraga Christinet, I. Jelescu, N. Vietti Violi; Lausanne/CH
Purpose: Non-invasive quantification of liver fibrosis in cirrhosis guides prognosis, surveillance, and therapy. Biopsy is invasive and sampling-limited; vibration-controlled transient elastography (VCTE) samples a small right-lobe volume and may be affected by obesity/inflammation; MR elastography requires dedicated hardware. The purpose of this work is to implement and evaluate a 3T Diffusion-based virtual elastography (DvE) workflow that converts two-point shifted ADC into voxel-wise stiffness (μ_diff, kPa) for quantifying liver fibrosis, and to test its association withVCTE in adults with cirrhosis undergoing HCC screening; secondary aims were to quantify intrahepatic heterogeneity and assess repeatability in 6 months.
Methods or Background: Prospective, single-centre study with MRI and VCTE at each visit. DWI: ss-EPI, b=200/1500 s/mm². μ_diff was computed from sADC using a calibrated linear model. Processing included liver masking with vessel exclusion, variability filtering, a right-lobe region of interest (ROI) mirroring VCTE practice, and a whole-liver heterogeneity index (HI=coefficient of variation of μ_diff). Associations were tested with Spearman’s ρ; repeatability with two-way random-effects, absolute-agreement inter-class correlation (ICC).
Results or Findings: Eighty-two MRI–VCTE pairs were analysed. Right-lobe μ_diff correlated with VCTE stiffness (ρ=0.34, p=0.034). In 24 rescans, right-lobe μ_diff showed moderate repeatability (ICC=0.58). Whole-liver μ_diff showed a weaker, non-significant association (ρ=0.18, p=0.09). HI>20% classified 26.8% of livers as heterogeneous; excluding these strengthened the whole-liver association to significance (ρ=0.27, p=0.043; n=60), indicating spatial variability as a confounder when comparing local vs global stiffness.
Conclusion: DvE provides simple MRI-integrated, whole-liver stiffness maps for fibrosis quantification in cirrhosis screening. Observing the association withVCTE, which samples a right-lobe ROI, we propose intrahepatic heterogeneity can drive discordance. DvE may complement VCTE as for non-invasive whole-liver fibrosis assessment within routine HCC surveillance.
Limitations: Empirical calibration out of the dataset; diffusion artefacts and SNR/motion, limited sample size.
Funding for this study: SNSF - Project funding in biology and medicine (division III) 2021 October - 207944
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Prospective uniccentric study with local ethics approval (CER-VD 2022-D0116)

Author Block: D. Catucci¹, L. Hulbert¹, L. Zbinden¹, M. P. Brönnimann¹, V. Obmann¹, A. Berzigotti¹, A. Christe¹, R. Sznitman¹, A. T. Huber²; ¹Bern/CH, ²Lucerne/CH
Purpose: To analyse whether artificial intelligence (AI)-automated volumetry of liver structures and data extraction from liver T1 mapping can differentiate stages of chronic liver disease (CLD) and detect clinically significant portal hypertension (CSPH).
Methods or Background: In this retrospective study, 159 patients underwent liver MRI including T1 maps before (T1 long) and 20 minutes after Gd-EOB-DTPA-administration (T1 short) and native T1 Dixon-sequence for volumetry between 09/2018 and 05/2019. CLD-patients were grouped based on liver biopsy into early CLD (F0-F2; n=34), advanced CLD (F3-F4; n=27) and advanced CLD with CSPH (F3-F4 with presence of portosystemic collaterals, splenomegaly >12 cm with thrombocytopenia <100 G/L or ascites; n=53). 45 patients without CLD served as a negative control group. Total liver volume (TLV), hepatic veins volume (HVV), liver segmental volumes I-VIII, T1 long and T1 short were automatically measured using an in-house developed liver-segmentation-algorithm. A T1 reduction rate (ΔT1) was calculated as (T1 long – T1 short)/T1 long. In addition, a functional-right-lobe-volume-ratio (FRLVR) was calculated as (volumes segments IV-VIII/TLV)*ΔT1 and a hepatic-veins-ratio (HVR) was calculated as (HVV/TLV)*1000. Parameters were compared between groups using the Kruskal-Wallis-test. Univariate logistic regression analysis was used to assess the discriminatory power of parameters for group-differentiation.
Results or Findings: All parameters differed significantly between the groups (p<0.05). TLV initially increased with CLD-severity and decreased at advanced CLD-stages with CSPH. In contrast, ΔT1, FRLVR and HVR continuously decreased with increasing CLD-severity. HVR and FRLVR showed strong discriminatory power for CLD (odds-ratio OR 0.71 and 0.88), advanced CLD (OR 0.72 and 0.86) and CSPH (OR 0.66 and 0.87), all with p<0.001.
Conclusion: Our findings highlight the potential of integrating AI-automated liver segmentation for MRI data extraction and volumetry into clinical workflows to improve management of CLD patients.
Limitations: Retrospective study design.
Funding for this study: This study received funding from the Swiss National Science Foundation (grant number 10003604).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Cantonal ethics comittee of Bern, Switzerland.