ABSTRACT
Background and Aims
The applicability of non‐invasive tests (NITs) for clinically significant portal hypertension (CSPH) in patients with hepatocellular carcinoma (HCC) is controversial. We performed a head‐to‐head comparison of various NITs for CSPH and evaluated rule‐in/‐out criteria.
Methods
We conducted a retrospective study including 283 patients with HCC undergoing hepatic venous pressure gradient (HVPG) measurement at the Vienna Hepatic Haemodynamic Lab with available, same‐day data on NITs (i.e., liver stiffness measurement [LSM], platelet count [PLT], von Willebrand factor antigen [VWF:Ag], and the ratio of VWF:Ag and PLT [VITRO]). For the subsequent analysis, we focused on patients with compensated advanced chronic liver disease (cACLD).
Results
Among 164 patients with cACLD, most were BCLC 0/A (50.6%) and CSPH prevalence was 46.3%, the median LSM was 20.0 kPa (IQR: 13.3–37.9), and mean VWF:Ag/VITRO was 266 ± 119%/2.30 ± 1.81. The VITRO score yielded the highest discriminative ability to detect CSPH (AUROC: 0.834), followed by ANTICIPATE ± NASH (AUROC: 0.822), PLT (AUROC: 0.803), LSM (AUROC: 0.779), and VWF:Ag (AUROC: 0.707). The application of the Baveno VII criterium (LSM ≤ 15 kPa & PLT ≥ 150G/L) and the VITRO score (≤ 1.0) individually showed an excellent performance for ruling‐out CSPH (sensitivities: ≥ 95%, NPV: 100%/87.9%). For ruling‐in CSPH by individual criteria, the VITRO score provided the highest PPV (≥ 2.5, specificity/PPV: 89.8%/83.0%), followed by the Baveno VII criteria (specificity/PPV: 81.8%/74.6%). When sequentially applying Baveno VII criteria and VITRO, the grey zone could be significantly reduced to 16% (sensitivity/NPV: 94.7%/92.2%; specificity/PPV: 73.9%/84.4%).
Conclusion
The discriminative ability in patients with cACLD and HCC was comparable to patients without HCC. However, PPVs of established criteria were < 90%, which may be explained by the comparatively low prevalence of CSPH in this context.
Keywords: ACLD, cirrhosis, HCC, liver stiffness, VITRO, von Willebrand factor
Key Summary
- Established knowledge of the subject
-
○Data on non‐invasive tests to predict clinically significant portal hypertension in patients with hepatocellular carcinoma are controversial.
-
○A head‐to‐head comparison of the diagnostic performance of various non‐invasive tests is needed.
-
○
- Significant and/or new findings of this study
-
○The discriminative ability of non‐invasive tests in patients with compensated advanced chronic liver disease and hepatocellular carcinoma was comparable to that in patients without hepatocellular carcinoma in our study.
-
○To rule out clinically significant portal hypertension, the sequential application of the Baveno VII criteria and VITRO score significantly reduced the grey zone retaining adequate model characteristics.
-
○PPVs of established criteria to rule‐in clinically significant portal hypertension were < 90%, with the VITRO score showing highest discriminative abilities.
-
○
Lay Summary
The introduction of non‐invasive tests (i.e., elastography‐ and blood‐based biomarkers) revolutionised the staging of chronic liver disease and the diagnosis of compensated advanced chronic liver disease (cACLD). Patients with cACLD are at risk of hepatocellular carcinoma and may have clinically significant portal hypertension (CSPH), which predisposes them to complications. The utility of non‐invasive tests for CSPH in patients with hepatocellular carcinoma is controversial. We found that algorithms based on liver stiffness measurement and platelet count (Baveno VII criteria) as well as the ratio of von Willebrand factor antigen to platelet count (VITRO) can adequately rule‐out CSPH. However, broadly used cut‐offs cannot establish a diagnosis of CSPH in the context of hepatocellular carcinoma.
Abbreviations
- ACLD
advanced chronic liver disease
- AFP
alpha‐fetoprotein
- AUROC
area under the receiver operating characteristic curve
- BCLC
Barcelona clinic liver cancer
- cACLD
compensated advanced chronic liver disease
- CI
confidence interval
- CSPH
clinically significant portal hypertension
- CTP
Child‐Turcotte‐Pugh
- HCC
hepatocellular carcinoma
- HVPG
hepatic venous pressure gradient
- IQR
interquartile range
- LSM
liver stiffness measurement
- MELD
model for end‐stage liver disease
- NIT
non‐invasive test
- NPV
negative predictive value
- PLT
platelet count
- PPV
positive predictive value
- SD
standard deviation
- UNOS
united network for organ sharing
- VITRO score
von Willebrand factor antigen to platelet count ratio
- VWF:Ag
von Willebrand factor antigen
1. Introduction
Non‐invasive tests (NIT) have revolutionised staging and risk prediction in patients with chronic liver disease [1]. Liver stiffness measurement (LSM) is broadly applied to non‐invasively diagnose compensated advanced chronic liver disease (cACLD), a spectrum of advanced fibrosis and cirrhosis, and estimate the probability of clinically significant portal hypertension (CSPH) along with platelet count (PLT) [1, 2]. Von Willebrand factor antigen (VWF:Ag) is the most promising blood‐based NIT for CSPH. It is an inexpensive, widely available laboratory test [3] and is used to calculate the VITRO score [4] (i.e., VWF:Ag to PLT ratio) which has the potential to reduce the diagnostic grey zone of the Baveno VII criteria [2, 5, 6] or even replace elastography‐based NITs or estimating the probability of CSPH and hepatic decompensation [7, 8, 9, 10].
CSPH is the major pathophysiological driver of first hepatic decompensation which can be prevented by non‐selective betablockers and plays a major role in treatment allocation in patients with hepatocellular carcinoma (HCC), particularly when assessing eligibility for surgical resection. In the absence of clinically overt signs of CSPH (i.e., varices, portosystemic collaterals, or clinically evident ascites), the hepatic venous pressure gradient (HVPG) remains the minimally‐invasive reference standard for diagnosing CSPH [1, 2]. Importantly, tumour location and size [11], as well as macrovascular invasion [12], may influence LSM, whereas PLT might be increased by the presence of HCC, thereby potentially comprising diagnostic utility for CSPH [13]. A recent retrospective study questioned the role of NITs for CSPH risk estimation in patients with HCC [12]. Thus, an accurate NIT or a combination of NITs for identifying CSPH in patients with HCC remains an important unmet need. Notably, following Baveno VII [1], the clinical implications of a non‐invasive diagnosis of CSPH in patients with cACLD and HCC are not limited to candidates for resection, as carvedilol therapy should be considered in those with CSPH. Finally, given the controversy around LSM/PLT and recently emerging encouraging data on VWF:Ag (but not VITRO) [14], a head‐to‐head comparison of available CSPH‐NIT seems particularly desirable.
The objective of this study was to perform a head‐to‐head comparison of the diagnostic performance of various NITs (LSM, PLT, VWF:Ag, and VITRO score) for CSPH in patients with HCC.
2. Patients and Methods
2.1. Study Design and Patients
All patients with histologically or radiologically diagnosed HCC undergoing HVPG measurement between 01/2004 and 11/2023 at the Vienna Hepatic Haemodynamic Lab were considered eligible for this retrospective study, if (i) reliable HVPG, (ii) LSM and PLT as well as (iii) VWF:Ag levels were available. The date of HVPG measurement was considered as the baseline of this study. Patients were excluded if any of the following criteria were present: Patients with other types of liver tumours, a history of liver transplantation, and/or missing information on important clinical/imaging/laboratory parameters. Study‐relevant clinical and laboratory information was retrospectively collected from patients' medical records.
For all analyses, patients were stratified into two groups, (i) cACLD (as defined by LSM ≥ 10 kPa, and/or HVPG ≥ 6 mmHg and/or histological F3/F4 fibrosis stage) and (ii) an overall cohort (including cACLD, decompensated cirrhosis, and non‐ACLD).
2.2. HVPG Measurement and CSPH Definition
HVPG measurements were performed in the absence of carvedilol/non‐selective beta‐blocker therapy and according to a local standardised operating procedure [15], which is in line with national [2, 5] and international consensuses [1, 16]. Under local anaesthesia and using ultrasound guidance, a catheter introducer sheath was inserted into the right internal jugular vein. Subsequently, a hepatic vein was cannulated by a specifically designed balloon catheter [17], and the free and wedged hepatic venous pressures were obtained/recorded at least as triplicate measurements. CSPH was defined by a threshold of HVPG ≥ 10 mmHg [16].
2.3. NITs for Portal Hypertension and Blood‐Based Biomarkers
Same‐day LSM was performed before the HVPG measurement by experienced operators in clinical routine using FibroScan (Echosens, Paris, France). All measurements were performed in fasting conditions. The M and XL‐probe were chosen based on availability and the recommendation of the device. The patients were lying in a dorsal position with the right arm in maximal abduction and measurements were performed in the right lobe of the liver, through intercostal spaces.
Blood samples were obtained via a central venous line (i.e., the side port of the catheter introducer sheath) at the time of HVPG measurement. In‐house VWF quantification was performed by an immuno‐turbidimetric assay (STA LIATEST VWF:Ag) on a STA‐R Evolution (both DIAGNOSTICA STAGO S.A.S., Asnières sur Seine, France; short Stago) analyser (i.e., an assay used in clinical routine) by the ISO‐certified Department of Laboratory Medicine of the Medical University of Vienna, immediately after the blood sample was taken. All other laboratory tests were performed by the same department, relying on methods used in clinical practice.
2.4. HCC Characteristics
The characteristics of the tumour (i.e., the number and size of nodule[s], the presence of macrovascular invasion and/or extrahepatic spread, and the exact localisation of the nodule[s]) were evaluated using the most recent imaging available at the time of the HVPG measurement. Patients were classified according to the Barcelona Clinic Liver Cancer (BCLC) classification. To compare NITs according to tumour characteristics, we divided the cohort into patients with tumours strictly located in the left lobe (including liver segment IV, no right lobe involvement) vs. right liver lobe involvement and then stratified according to presence vs. absence of CSPH.
2.5. Statistical Analysis
All statistical analyses were performed using R 4.5.2 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria). Categorical variables were reported as absolute (n) and relative frequencies (%), whereas continuous variables as mean ± standard deviation (SD) or median (interquartile range [IQR]), as appropriate. Student's t‐test was used for group comparisons of normally distributed variables and Mann–Whitney‐U‐test for non‐normally distributed variables. Group comparisons of categorical variables were performed using either Chi‐squared or Fisher's exact test, as appropriate. The correlation between continuous variables (especially LSM, PLT, VWF:Ag, VITRO score, and HVPG) was analysed using the Spearman's rank correlation coefficient and visualised via scatterplots. VWF:Ag cut‐offs according to Pereyra et al. [14] were applied, that is, > 291%. The area under the receiver operating characteristic curve (AUROC) of the different variables of interest for CSPH was calculated using the pROC‐package [1]. For sensitivity analyses, we calculated diagnostic accuracy of NITs in patients with vs. without large (i.e., ≥ 5 cm) nodules within the right liver lobe. The level of significance was set at a 2‐sided p value < 0.05.
2.6. Ethics
This study has been conducted in accordance with the principles of the Declaration of Helsinki and its amendments and has been approved by the ethics committee of the Medical University of Vienna (approval number 1869/2024), which waived the requirement of written informed consent.
3. Results
3.1. Patient and Tumour Characteristics in the Overall Study Cohort
After applying in‐ and exclusion criteria, 283 patients were finally included in the study cohort (Figure 1). Detailed information on the whole study cohort is displayed in the (Table S1).
FIGURE 1.
Flowchart. HCC, hepatocellular carcinoma; HVPG, hepatic venous pressure gradient; LSM, liver stiffness measurement; VWF:Ag, von Willebrand factor antigen.
3.2. Patient and Tumour Characteristics in Patients With cACLD
Patients with cACLD (n = 164) were predominantly male (86.0%), with leading alcohol‐/viral‐related liver disease aetiologies and a mean age of 66.0 (±9.4) years. The median HVPG was 9 (IQR: 6–14) mmHg with a CSPH prevalence of 46.3%. Included patients had well‐preserved liver function (median Child‐Turcotte‐Pugh [CTP] score 5 [IQR: 5–5] points, median united network for organ sharing model for end‐stage liver disease 2016 [UNOS MELD 2016] 9 [IQR: 8–10] points; Table 1).
TABLE 1.
Baseline characteristics of patients with cACLD.
| Patient characteristics | cACLD, n = 164 |
|---|---|
| Age, years, mean (SD) | 66.0 (9.4) |
| Sex, n (%) | |
| Male | 141 (86.0%) |
| Female | 23 (14.0%) |
| BMI, kg/m2, mean (SD) | 27.6 (4.1) |
| Aetiology of liver disease, n (%) | |
| ALD | 51 (31.1%) |
| Viral | 53 (32.3%) |
| ALD/Viral | 6 (3.7%) |
| MASLD | 27 (16.5%) |
| Other/unknown | 27 (16.5%) |
| History of alcohol use disorder, n (%) | 59 (36.0%) |
| Viral eradication/suppression, n (%) | 16 (9.8%) |
| HCC characteristics | |
| Time from HCC diagnosis to HVPG measurement, months, median (IQR) | 0.9 (0.1–2.3) |
| Prior locoregional therapy, n (%) | 6 (3.7%) |
| HCC within Milan criteria, n (%) | 83 (50.6%) |
| BCLC classification, stage, n (%) | |
| 0 | 15 (9.1%) |
| A | 68 (41.5%) |
| B | 58 (35.4%) |
| C | 23 (14.0%) |
| D | — |
| Number of nodules within the liver, n (%) | |
| 1 | 112 (68.3%) |
| 2 | 21 (12.8%) |
| 3 | 9 (5.5%) |
| 4 | — |
| ≥ 5 | 22 (13.4%) |
| Size of largest nodule, cm, median (IQR) | 4.1 (2.5–6.0) |
| Presence of macrovascular invasion, n (%) | 16 (9.8%) |
| Presence of extrahepatic spread, n (%) | 9 (5.5%) |
| Presence of non‐tumoral/apposition portal vein thrombosis, n (%) | 4 (2.4%) |
| Partial, n (%) | 4 (2.4%) |
| Localisation of tumour within the left liver lobe only, n (%) | 29 (17.7%) |
| First line treatment, n (%) | |
| Resection | 64 (39.0%) |
| Ablation | 18 (11.0%) |
| TACE/SIRT/Radiation | 66 (40.2%) |
| Systemic therapy | 8 (4.9%) |
| Best supportive care | 8 (4.9%) |
| Non‐invasive tests | |
| LSM, kPa, median (IQR) | 20.0 (13.3–37.9) |
| LSM ≥ 10 kPa, n (%) | 150 (91.5%) |
| LSM ≥ 25 kPa, n (%) | 63 (38.4%) |
| VWF:Ag, %, mean (SD) | 266 (119) |
| VITRO score, mean (SD) | 2.30 (1.81) |
| Platelet count, G/L, median (IQR) | 140 (104–187) |
| Portal hypertension characteristics | |
| HVPG, mmHg, median (IQR) | 9 (6–14) |
| CSPH, n (%) | 76 (46.3%) |
| Varices, n (%) | |
| None | 71 (43.3%) |
| Small | 21 (12.8%) |
| Large | 22 (13.4%) |
| Unknown | 50 (30.5%) |
| Red spot sign, n (%) | 5 (3.0%) |
| Portal hypertensive gastropathy, n (%) | |
| Moderate | 20 (12.2%) |
| Severe | 4 (2.4%) |
| Liver disease severity scores | |
| Child‐Turcotte‐Pugh score at baseline, points, median (IQR), stage, n (%) | 5 (5–5) |
| A | 159 (97.0%) |
| B | 5 (3.0%) |
| C | — |
| MELD score, points, median (IQR) | 9 (8–10) |
| MELD‐Na score, points, median (IQR) | 10 (9–12) |
| UNOS MELD (2016) score, points, median (IQR) | 9 (8–10) |
| Laboratory parameters | |
| Haemoglobin, g/dL, median (IQR) | 13.3 (12.2–14.4) |
| White blood cell count, G/L, median (IQR) | 5.5 (4.4–7.0) |
| Sodium, mmol/L, mean (SD) | 139.0 (2.9) |
| Creatinine, mg/dL, median (IQR) | 0.85 (0.74–1.02) |
| Bilirubin, mg/dL, median (IQR) | 0.90 (0.64–1.36) |
| Albumin, g/L, mean (SD) | 39.4 (4.2) |
| Alkaline phosphatase, IU/L, median (IQR) | 87 (66–117) |
| AST, IU/L, median (IQR) | 42 (32–66) |
| ALT, IU/L, median (IQR) | 40 (26–61) |
| GGT, IU/L, median (IQR) | 129 (69–244) |
| INR, mean (SD) | 1.18 (0.16) |
| AFP, IU/mL, median (IQR) | 7.4 (3.4–47.2) |
Note: Categorical variables were reported as absolute (n) and relative frequencies (%), whereas continuous variables as mean ± SD or median (interquartile range [IQR]), as appropriate.
Abbreviations: AFP, alpha‐fetoprotein; ALD, alcohol‐related liver disease; ALT, alanine‐aminotransferase; AST, aspartate‐aminotransferase; BCLC, Barcelona clinic liver cancer; BMI, body mass index; cACLD, compensated advanced chronic liver disease; CSPH, clinically significant portal hypertension; ECOG, Eastern Cooperative Oncology Group; GGT, gamma‐glutamyltransferase; HCC, hepatocellular carcinoma; HVPG, hepatic venous pressure gradient; INR, international normalised ratio; LSM, liver stiffness measurement; MASLD, metabolic dysfunction associated steatotic liver disease; MELD, model for end‐stage liver disease, UNOS MELD, United Network for Organ Sharing model for end‐stage liver disease; VITRO, von Willebrand factor antigen to platelet count ratio; VWF:Ag, von Willebrand factor antigen.
The median LSM was 20.0 (IQR: 13.3–37.9) kPa (≥ 10 kPa: 91.5%, ≥ 25 kPa: 38.4%), whereas the median PLT was 140 G/L (IQR: 104–187, PLT < 150 G/L: 57.3%). Mean VWF:Ag and VITRO scores were 266 (±119) % and 2.30 (±1.81, VITRO ≥ 2.5: 32.3%), respectively (Table 1).
The median time from HCC diagnosis to the HVPG measurement (or vice versa) was 0.9 (IQR: 0.1–2.3) months. Six patients (3.7%) had undergone locoregional therapy prior to study inclusion and 50.6% were classified as being within Milan criteria. Regarding the BCLC classification, 15 (9.1%) patients were classified as BCLC stage 0, 68 (41.5%) patients as BCLC stage A, 58 (35.4%) patients as BCLC B, and 23 (14.0%) patients as BCLC C. Most patients had a single HCC nodule (68.3%) compared to 13.4% of patients having ≥ 3 nodules. The tumour was located strictly in the left liver lobe in 17.7% of patients, showed macrovascular invasion in 9.8%, extrahepatic spread in 5.5%. The median alpha‐fetoprotein (AFP) levels were 7.4 (IQR: 3.4–47.2) IU/mL.
3.3. NITs for Estimating CSPH Probability in Patients With cACLD
Same‐day LSM was available for all patients (IQR/median < 30% in 79.2%; all measurements were included for subsequent analyses). LSM showed a moderate correlation with HVPG (Spearman's rho: 0.387, p < 0.001). This resulted in a high discriminative ability for CSPH with an AUROC of 0.779 (95% confidence interval [CI]: 0.705–0.852). PLT was negatively correlated with HVPG (Spearman's rho: −0.615, p < 0.001) and the AUROC was 0.803 (95% CI: 0.734–0.871). The strength of the positive correlation between VWF:Ag levels and HVPG was moderate (Spearman's rho: 0.344, p < 0.001), translating into an AUROC of 0.707 (95% CI: 0.628–0.785). Finally, the VITRO score had the numerically strongest correlation with HVPG (Spearman's rho: 0.623, p < 0.001) and showed an AUROC of 0.834 (95% CI: 0.772–0.896). Lastly, the ANTICIPATE±NASH model yielded similar discriminative abilities (AUROC 0.822 [95% CI: 0.756–0.887]; Table 2, Figure 2, Figure 3).
TABLE 2.
Comparison of model characteristics for the prediction of CSPH in patients with cACLD.
| cACLD, n = 164 | |||||
|---|---|---|---|---|---|
| LSM (kPa) | VWF:Ag (%) | PLT (G/L) | VITRO score | ANTICIPATE ± NASH | |
| AUROC (95% CI) | 0.779 (0.705–0.852) | 0.707 (0.628–0.785) | 0.803 (0.734–0.871) | 0.834 (0.772–0.896) | 0.822 (0.756–0.887) |
| Ruled‐out (n) | Sens (%) | NPV (%) | Grey zone (n) | Ruled‐in (n) | Spec (%) | PPV (%) | |
|---|---|---|---|---|---|---|---|
| VITRO, ≤ 1.0/≥ 2.5 | 33 (20%) | 94.7% | 87.9% | 78 (48%) | 53 (32%) | 89.8% | 83.0% |
| Baveno VII | 33 (20%) | 100% | 100% | 68 (41%) | 63 (38%) | 81.8% | 74.6% |
| Baveno VII + VITRO, ≤ 1.0/≥ 2.5 | 51 (31%) | 94.7% | 92.2% | 26 (16%) | 87 (53%) | 73.9% | 84.4% |
| cACLD, excl. nodules ≥ 5 cm within the right liver lobe, n = 113 | |||||
|---|---|---|---|---|---|
| LSM (kPa) | VWF:Ag (%) | PLT (G/L) | VITRO score | ANTICIPATE ± NASH | |
| AUROC (95% CI) | 0.817 (0.732–0.903) | 0.701 (0.604–0.798) | 0.767 (0.679–0.854) | 0.812 (0.733–0.891) | 0.847 (0.773–0.920) |
| Ruled‐out (n) | Sens (%) | NPV (%) | Grey zone (n) | Ruled‐in (n) | Spec (%) | PPV (%) | |
|---|---|---|---|---|---|---|---|
| VITRO, ≤ 1.0/≥ 2.5 | 19 (17%) | 94.8% | 84.2% | 54 (48%) | 40 (35%) | 87.2% | 82.5% |
| Baveno VII | 23 (20%) | 100% | 100% | 48 (42%) | 42 (37%) | 85.5% | 81.0% |
| Baveno VII + VITRO, ≤ 1.0/≥ 2.5 | 30 (27%) | 94.8% | 90.0% | 21 (19%) | 62 (55%) | 74.5% | 77.4% |
| cACLD, only nodules ≥ 5 cm within the right liver lobe, n = 51 | |||||
|---|---|---|---|---|---|
| LSM (kPa) | VWF:Ag (%) | PLT (G/L) | VITRO score | ANTICIPATE ± NASH | |
| AUROC (95% CI) | 0.747 (0.607–0.886) | 0.737 (0.597–0.876) | 0.849 (0.714–0.984) | 0.859 (0.747–0.970) | 0.817 (0.699–0.935) |
| Ruled‐out (n) | Sens (%) | NPV (%) | Grey zone (n) | Ruled‐in (n) | Spec (%) | PPV (%) | |
|---|---|---|---|---|---|---|---|
| VITRO, ≤ 1.0/≥ 2.5 | 14 (28%) | 94.4% | 92.9% | 24 (47%) | 13 (25%) | 93.9% | 84.6% |
| Baveno VII | 10 (20%) | 100% | 100% | 20 (39%) | 21 (41%) | 75.8% | 83.3% |
| Baveno VII + VITRO, ≤ 1.0/≥ 2.5 | 21 (41%) | 94.4% | 95.2% | 5 (10%) | 25 (49%) | 72.7% | 64.0% |
Note: Discriminatory abilities were compared using AUROC analyses. Criteria to rule‐out CSPH with NPV/sensitivity as well as to rule‐in with PPV/specificity were calculated.
Abbreviations: AUROC, area under the receiver operating characteristic curve; CSPH, clinically significant portal hypertension; LSM, liver stiffness measurement; PLT, platelet count; VITRO, von Willebrand factor antigen to platelet count ratio; VWF:Ag, von Willebrand factor antigen.
FIGURE 2.
Area under the receiver operating characteristic curve of non‐invasive tests for CSPH and Spearman correlation plot of non‐invasive tests and HVPG in patients with cACLD. The size of the circles represents the correlation strength. BCLC, Barcelona clinic liver cancer; cACLD, compensated advanced chronic liver disease; HVPG, hepatic venous pressure gradient; LSM, liver stiffness measurement; PLT, platelet count; VITRO, von Willebrand factor antigen to platelet count ratio; VWF:Ag, von Willebrand factor antigen.
FIGURE 3.
Scatter plots of HVPG according to (A) liver stiffness, (B) platelet count, (C) VWF:Ag, and (D) VITRO score in patients with cACLD. To visualise the diagnostic accuracy of cut‐offs for CSPH, scatter plots were coloured in dark blue (rule‐in CSPH) or light blue (rule‐out CSPH). Each point of the scatterplot represents individual patient data. BCLC, Barcelona clinic liver cancer; cACLD, compensated advanced chronic liver disease; CSPH, clinically significant portal hypertension; HVPG, hepatic venous pressure gradient; LSM, liver stiffness measurement; VWF:Ag, von Willebrand factor antigen.
3.4. NITs for Estimating CSPH Probability in the Overall Cohort
To allow for comparisons with studies which also included patients with decompensated cirrhosis, analysis was repeated in all subjects. In brief, AUROCs were 0.834 for LSM, 0.784 for VWF:Ag, 0.834 for PLT, 0.886 for the VITRO score and ANICIPATE±NASH. Detailed information is provided in Table S2, Figures S1 and S2.
3.5. Baveno VII‐ and VITRO‐Based Criteria for Ruling‐In and Ruling‐Out CSPH in Patients With cACLD
The Baveno VII rule‐in criterium (i.e., LSM ≥ 25 kPa) showed a positive predictive value (PPV) of 74.6% and a specificity of 81.8% (grey zone 41%). The Baveno VII criterium to rule‐out CSPH (i.e., LSM <= 15 kPa and PLT ≥ 150 G/L) had a negative predictive value (NPV) and a sensitivity of 100%.
VITRO ≥ 2.5 showed a slightly higher PPV/specificity for ruling‐in CSPH (PPV 83.0%, specificity 89.8%), while VITRO ≤ 1.0 performed reasonably well for ruling‐out CSPH (NPV 87.9%, sensitivity 94.7%, grey zone 48%, Table 2).
When applying the Baveno VII‐VITRO sequence, the grey zone was significantly reduced (n = 26, 16%) with adequate rule‐out (NPV 92.2%, sensitivity 94.7%) but suboptimal rule‐in characteristics (PPV 84.4%, specificity 73.9%; Table 2).
3.6. Baveno VII‐ and VITRO‐Based Criteria for Ruling‐In and Ruling‐Out CSPH in the Overall Cohort
Compared to the cACLD cohort, the model characteristics to rule‐in and ‐out CSPH were numerically increased. While the Baveno VII rule‐in criterium showed a PPV of 77.8% and a specificity of 80.5%, the rule‐out criteria had an NPV/sensitivity of 100% (grey zone 35%). The VITRO ≥ 2.5 performed well for ruling‐in CSPH (PPV 85.1%, specificity 90.3%) and VITRO ≤ 1.0 for ruling‐out CSPH (NPV 93.2%, sensitivity 96.1%, grey zone 38%). Its sequence reduced the grey zone (11%) with adequate rule‐out (NPV 95.1%, sensitivity 96.1%) but also suboptimal rule‐in characteristics (PPV 77.3%, specificity 77.9%) (Table S2).
3.7. Influence of Tumour Localisation on CSPH‐NITs in Patients With cACLD
When stratifying patients according to the presence vs. absence of CSPH and comparing tumour localisation(s) (strictly left liver lobe vs. right liver lobe involvement), there were no clinically significant differences in NITs or other patient characteristics (Table 3). In patients with nodules ≥ 5 cm within the right liver lobe, the discriminative ability of LSM for CSPH was numerically decreased (AUROC 0.747 [95% CI: 0.607–0.886] vs. 0.817 [95% CI: 0.732–0.903]), which was also the case for its derivative, ANTICIPATE±NASH (AUROC 0.817 [95% CI: 0.699–0.935] vs. 0.847 [95% CI: 0.773–0.920]). Conversely, discriminative abilities of VWF:Ag, PLT, and VITRO were numerically higher.
TABLE 3.
Comparison of patient characteristics according to localisation of tumour, that is, right liver lobe involvement vs. left liver lobe only, and presence vs. absence of CSPH in patients with cACLD.
| Patient characteristics | CSPH | No CSPH | ||||
|---|---|---|---|---|---|---|
| Right liver lobe involvement, n = 63 | Left liver lobe only, n = 13 | p | Right liver lobe involvement, n = 72 | Left liver lobe only, n = 16 | p | |
| HCC within Milan criteria, n (%) | 35 (55.6%) | 7 (53.8%) | 1.000 | 32 (44.4%) | 9 (56.3%) | 0.562 |
| BCLC classification, stage, n (%) | ||||||
| 0 | 5 (7.9%) | 3 (23.1%) | — | 4 (5.6%) | 3 (18.8%) | — |
| A | 30 (47.6%) | 4 (30.8%) | 28 (38.9%) | 6 (37.5%) | ||
| B | 22 (34.9%) | 3 (23.1%) | 28 (38.9%) | 5 (31.3%) | ||
| C | 6 (9.5%) | 3 (23.1%) | 12 (16.7%) | 2 (12.5%) | ||
| Number of nodules within the liver, n (%) | ||||||
| 1 | 37 (58.7%) | 10 (76.9%) | — | 53 (73.6%) | 12 (75.0%) | — |
| 2 | 11 (17.5%) | 1 (7.7%) | 7 (9.7%) | 2 (12.5%) | ||
| 3 | 5 (7.9%) | — | 4 (5.6%) | — | ||
| 4 | — | — | — | — | ||
| ≥ 5 | 10 (15.9%) | 2 (15.4%) | 8 (11.1%) | 2 (12.5%) | ||
| Size of largest nodule, cm, median (IQR) | 3.8 (2.3–5.2) | 3.5 (1.7–4.2) | 0.258 | 4.7 (3.0–7.1) | 4.0 (2.2–5.1) | 0.200 |
| Presence of macrovascular invasion, n (%) | 5 (7.9%) | 1 (7.7%) | 0.978 | 9 (12.5%) | 1 (6.3%) | 0.782 |
| Presence of extrahepatic spread, n (%) | 1 (1.6%) | 3 (23.1%) | 0.013 | 4 (5.6%) | 1 (6.3%) | 1.000 |
| Liver stiffness measurement, kPa, median (IQR) | 30.3 (20.1–50.5) | 26.6 (20.3–45.0) | 0.614 | 14.3 (11.2–22.7) | 13.1 (10.7–15.3) | 0.274 |
| Hepatic venous pressure gradient, mmHg, median (IQR) | 15 (12–19) | 13 (12–17) | 0.480 | 6 (4–8) | 7 (5–8) | 0.793 |
| Aspartate aminotransferase, IU/L, median (IQR) | 50 (38–71) | 42 (34–44) | 0.133 | 37 (28–59) | 34 (30–67) | 0.846 |
| Alanine aminotransferase, IU/L, median (IQR) | 43 (29–61) | 31 (25–44) | 0.227 | 36 (25–63) | 32 (23–61) | 0.752 |
| Platelet count, G/L, median (IQR) | 109 (79–133) | 124 (99–141) | 0.408 | 167 (139–222) | 185 (85–233) | 0.689 |
| VWF:Ag, %, mean (SD) | 313 (132) | 278 (74) | 0.558 | 229 (100) | 244 (112) | 0.581 |
| VITRO, mean (SD) | 3.38 (2.15) | 2.74 (1.51) | 0.389 | 1.36 (0.75) | 1.94 (1.54) | 0.341 |
| Alpha‐fetoprotein, IU/mL, median (IQR) | 11 (5–100) | 5 (4–8) | 0.045 | 5 (3–28) | 5 (2–21) | 0.476 |
Note: Categorical variables were reported as absolute (n) and relative frequencies (%), whereas continuous variables as mean ± SD or median (interquartile range [IQR]), as appropriate. Student's t‐test was used for group comparisons of normally distributed variables and Mann–Whitney‐U‐test for non‐normally distributed variables. Group comparisons of categorical variables were performed using either Chi‐squared or Fisher's exact test, as appropriate. p values in bold denote p < 0.05.
Abbreviations: BCLC, Barcelona clinic liver cancer; cACLD, compensated advanced chronic liver disease; HCC, hepatocellular carcinoma; IQR, interquartile range; VITRO score, von Willebrand factor antigen to platelet count ratio; VWF:Ag, von Willebrand factor antigen.
3.8. Influence of Tumour Localisation on CSPH‐NITs in the Overall Cohort
In brief, tumour localisation had no impact on CSPH‐NITs. Detailed information is provided in the Table S3.
4. Discussion
This study performed a head‐to‐head comparison of NITs and NIT‐based criteria to estimate the probability of CSPH in patients with HCC. Although there has been controversy in this field [18], our study showed that LSM, PLT, ANTICIPATE±NASH, as well as VITRO score, had excellent discriminative abilities for CSPH in patients with HCC. The Baveno VII criteria/the VITRO cut‐offs showed robust/acceptable results for ruling‐out CSPH with sensitivities ~95%. When the Baveno VII criteria were expanded to include the VITRO cut‐offs, the grey zone was significantly reduced, with similar accuracy for ruling‐out but poor model characteristics for ruling‐in CSPH. This suggests that the VITRO score could also help to reduce the grey zone for ruling‐out patients with CSPH and HCC.
A retrospective study questioned the role of Baveno VI criteria to rule out high‐risk varices or rule in CSPH. Among 185 patients (BCLC 0‐A in 46%; high‐risk varices in 23%) Baveno VI criteria showed a sensitivity (93%) and NPV (92%) for ruling out high‐risk varices, thereby failing the arbitrarily defined threshold of missing < 5% [12]. Moreover, in a small subgroup of patients (n = 53 with available HVPG and absence of macrovascular invasion; 43% CSPH), the NPV of LSM ≤ 15 kPa and PLT ≥ 150 G/L was found to be 77% (10/13), while the PPV of LSM ≥ 25 kPa for CSPH was only 65% (13/17).
In our study, we focused on patients with cACLD rather than patients with cirrhosis and CTP A as in the study by Allaire et al., as cACLD has largely replaced compensated cirrhosis due to the broad use of NITs. Moreover, a small proportion of patients with CTP A may have a history of hepatic decompensation, while not all compensated patients are CTP A. Notably, cACLD is the target population for NITs for CSPH, as those with less advanced disease or decompensated cirrhosis are easily classifiable [19]. In general, studies linking LSM/PLT and HVPG in patients with HCC are limited, with most of them either not consistently using balloon catheters and thereby compromising the accuracy of HVPG or including a proportion of subjects without ACLD/with decompensated cirrhosis [20, 21, 22, 23, 24, 25, 26, 27, 28, 29]. In order to provide results that are comparable with the latter studies (i.e., those including decompensated cirrhosis), we have included an analysis of all patients into the Supplement. Moreover, most of the previous studies have not evaluated the Baveno VII recommendations, which are shaping today's clinical practice.
As there are many factors potentially affecting the relationship between LSM and HVPG in patients with HCC (in particular, tumour localisation), we compared patients with vs. without CSPH and according to the localisation of the tumour (strictly left liver lobe vs. right liver lobe involvement). Importantly, we did not find any differences in LSM or other NITs according to tumour localisation in the individual CSPH strata, indicating that LSM and HVPG values that are deemed reliable despite the presence of HCC by the treating physician can be used to estimate the probability of CSPH.
However, ruling‐in CSPH in patients with cACLD and HCC using NITs at a PPV threshold of 90% remains challenging. Notably, PPV is directly related to prevalence, which was lower (45.6%) in our and previous (e.g., 43% CSPH in [12]) studies focusing on HCC, as compared to other studies on the non‐invasive diagnosis of CSPH in cACLD (~60%). Thus, the slightly lower PPV, both for LSM ≥ 25 kPa and VITRO > 2.5, may primarily be explained by plain math/statistics. Nevertheless, the lower PPV may have implications for clinical practice, as it increases the likelihood of deferring resection and/or overtreating patients with carvedilol/non‐selective betablocker therapy.
In daily practice, ruling‐out CSPH is particularly relevant in the context of evaluation for resection, and our results strongly support the use of the Baveno VII criteria. Importantly, the findings of our reasonably sized study contradict previous concerns in regard to the application of the Baveno VII criteria for ruling‐out CSPH (LSM ≤ 15 kPa and PLT ≥ 150 G/L) that were raised by Allaire et al. [12], as the NPV and a sensitivity attained 100%.
Given that rule‐out criteria were only met by 20%, with 41% remaining in the indeterminate zone after the application of the Baveno VII criteria, we evaluated whether the Baveno VII‐VITRO sequence [6] decreased the indeterminate/grey zone. By applying the Baveno VII‐VITRO sequence, the proportion of patients meeting the rule‐out criteria increased to 31%, while maintaining an adequate NPV of 92.2% and a sensitivity of 94.7%. Moreover, the indeterminate/grey zone decreased to 16%, however, the PPV was only 84.4%.
Thus, at present, HVPG measurement and/or endoscopy still have a role in a considerable proportion of patients with cACLD and HCC, given the substantial indeterminate/grey zone, as well as the limited PPV of conventional CSPH‐NIT‐based criteria.
Our study has several important strengths and limitations to acknowledge. First, the study was retrospective, with all its potential biases. Patients were included at the time of HVPG measurement. Importantly, in the majority of patients, the time between HCC diagnosis and HVPG measurement was small (77.4% within 3 months). In patients with HCC diagnosis > 3 months to HVPG measurement, HVPG measurement and imaging were within 3 months. A significant number of patients was excluded due to unavailability of LSM data (n = 171/501, 34%) and it cannot be ascertained, whether this was due to unavailability of the device, technical failure, or the physicians' concerns about the reliability of LSM in this context. Moreover, with the introduction of the 100 Hz probe for spleen stiffness measurement only at the end of the study period, data on spleen stiffness in our cohort was not available. However, the application of the NICER model may have increased the diagnostic performance, as it has been shown to be robust to a decrease in CSPH prevalence by resampling. Moreover, our study lacks information on cross‐sectional imaging‐based NITs (e.g., liver surface nodularity) for comparisons. We cannot rule out selection bias; however, haemodynamic evaluation of our patients is performed routinely for risk stratification and treatment monitoring purposes, and thus, we are convinced that our study population is not highly selected. Nevertheless, BCLC C/D patients were clearly underrepresented, although clinically CSPH‐NITs may have a role for indicating carvedilol/non‐selective betablocker therapy in those with cACLD. Unfortunately, our study does not include a validation cohort, as we are not aware of any equally sized, HVPG‐characterised cohort with data on CSPH‐NITs (especially VWF:Ag/VITRO) and HCC. Multicentre, and ideally, prospective studies are needed with a larger number of patients to validate our results and also investigate the prognostic utility of NITs in the context of HCC. Since most of included patients were classified as potentially resectable, there might be a selection bias within our study, mitigating generalisability.
In conclusion, the discriminative ability of CSPH‐NITs in patients with cACLD and HCC was comparable to patients without HCC, and the Baveno VII criteria as well as the Baveno VII‐VITRO sequence were capable of ruling out CSPH. However, PPVs of established rule‐in criteria were < 90%, which may be explained by the comparatively low prevalence of CSPH in this population.
Author Contributions
L.B.: conceptualisation, data curation, formal analysis, investigation, methodology, visualisation, writing – original draft preparation, writing – review and editing. T.J.H.: data curation, investigation, writing – review and editing. M.J.: data curation, writing – review and editing. G.S.: data curation, writing – review and editing. B.Si.: data curation, writing – review and editing. A.F.S: data curation, writing – review and editing. D.P.: data curation, writing – review and editing. C.S.: data curation, writing – review and editing. S.S.: data curation, writing – review and editing. K.K.: data curation, writing – review and editing. P.S.: data curation, writing – review and editing. P.Q.: data curation, investigation, methodology, resources, writing – review and editing. M.T.: data curation, resources, software, supervision, validation, writing – review and editing. B.Sc.: data curation, writing – review and editing. T.R.: data curation, writing – review and editing. M.P.: data curation, writing – review and editing. M.M.: conceptualisation, data curation, formal analysis, investigation, methodology, project administration, resources, supervision, validation, writing – original draft preparation, writing – review and editing.
Funding
No financial support specific to this study was received. L.B., M.J., G.S., B.Si., M.T., B.Sc., T.R., and M.M. are supported by the Clinical Research Group MOTION, Medical University of Vienna, Vienna, Austria—A project funded by the Clinical Research Groups Programme of the Ludwig Boltzmann Gesellschaft (Grant Nr: LBG_KFG_22_32) with funds from the Fonds Zukunft Österreich.
Conflicts of Interest
L.B. served as an advisory board member for Boehringer Ingelheim and received travel support from Roche. G.S. received speaker fees from DiaSorin. B.Si. received travel support from AbbVie and Gilead. M.T. received speaker fees from Agomab, BMS, Chemomab, Falk Foundation, Gilead, Intercept, Ipsen, Janssen, Madrigal, MSD, and Roche; he advised for AbbVie, Albireo, BiomX, Boehringer Ingelheim, Cymabay, Falk Pharma GmbH, Genfit, Gilead, Hightide, Intercept, Ipsen, Janssen, MSD, Novartis, Phenex, Pliant, Rectify, Regulus, Siemens, and Shire. He further received travel support from AbbVie, Falk, Gilead, Intercept, and Jannsen and research grants from Albireo, Alnylam, Cymabay, Falk, Gilead, Intercept, MSD, Takeda and UltraGenyx. He is also a co‐inventor of patents on the medical use of norUDCA filed by the Medical Universities of Graz and Vienna. B.Sc. received grant support from AstraZeneca, Eisai, and Ipsen, speaker honoraria from AstraZeneca and Eisai, as well as travel support from AbbVie, AstraZeneca, Ipsen, Gilead, and Roche. T.R. received grant support from Abbvie, Boehringer Ingelheim, Gilead, Intercept/Advanz Pharma, MSD, Myr Pharmaceuticals, Philips Healthcare, Pliant, Siemens and W. L. Gore & Associates; speaking/writing honoraria from Abbvie, Echosens, Gilead, GSK, Intercept/Advanz Pharma, Pfizer, Roche, MSD, Siemens, W. L. Gore & Associates; consulting/advisory board fee from Abbvie, Astra Zeneca, Bayer, Boehringer Ingelheim, Gilead, Intercept/Advanz Pharma, MSD, Resolution Therapeutics, Siemens; and travel support from Abbvie, Boehringer Ingelheim, Dr. Falk Pharma, Gilead, and Roche. M.P. served as a speaker and/or consultant and/or advisory board member for AstraZeneca, Bayer, Bristol‐Myers Squibb, Eisai, Ipsen, Lilly, MSD, and Roche and received travel support from Bayer and Bristol‐Myers Squibb, Ipsen, and Roche. M.M. served as a speaker and/or consultant and/or advisory board member for Boehringer Ingelheim, Chiesi, Echosens, Eli Lilly, Ipsen, Madrigal, Medpace, Novo Nordisk, Takeda, and W. L. Gore & Associates and received travel support from Ipsen as well as grants/research support from Echosens. The other authors declare no conflicts of interest.
Supporting information
Table S1: Baseline characteristics of the overall cohort.
Table S2: Comparison of model characteristics for the prediction of CSPH in the overall cohort.
Table S3: Comparison of patient characteristics according to localisation of tumour, that is, right versus left liver lobe only, and presence versus absence of CSPH in the overall cohort.
Figure S1: Area under the receiver‐operating curve of distinct non‐invasive tests of interest for CSPH and spearman correlation plot of non‐invasive tests and HVPG in the overall cohort. The size of the circles represents the correlation strength.
Figure S2: Scatter plots of HVPG according to (A) liver stiffness, (B) platelet count, (C) VWF:Ag, and (D) VITRO score in the overall cohort. To visualise the diagnostic accuracy of cut‐offs for CSPH, scatter plots were coloured in dark blue (rule‐in CSPH) or light blue (rule‐out CSPH). Each point of the scatterplot represents individual patient data.
Acknowledgements
Open Access funding provided by Medizinische Universitat Wien.
Balcar L., Haunold T. J., Jachs M., et al., “Blood‐ and Liver Stiffness‐Based Non‐Invasive Tests for Clinically Significant Portal Hypertension in Patients With cACLD and HCC ,” Liver International 46, no. 11 (2026): e70947, 10.1111/liv.70947.
Handling Editor: Luca Valenti
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. de Franchis R., Bosch J., Garcia‐Tsao G., et al., “Baveno VII—Renewing Consensus in Portal Hypertension,” Journal of Hepatology 76, no. 4 (2022): 959–974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Mandorfer M., Aigner E., Cejna M., et al., “Austrian Consensus on the Diagnosis and Management of Portal Hypertension in Advanced Chronic Liver Disease (Billroth IV),” Wiener Klinische Wochenschrift 135, no. Suppl 3 (2023): 493–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Nichols W. L., Hultin M. B., James A. H., et al., “von Willebrand Disease (VWD): Evidence‐Based Diagnosis and Management Guidelines, the National Heart, Lung, and Blood Institute (NHLBI) Expert Panel Report (USA),” Haemophilia 14, no. 2 (2008): 171–232. [DOI] [PubMed] [Google Scholar]
- 4. Hametner S., Ferlitsch A., Ferlitsch M., et al., “The VITRO Score (Von Willebrand Factor Antigen/Thrombocyte Ratio) as a New Marker for Clinically Significant Portal Hypertension in Comparison to Other Non‐Invasive Parameters of Fibrosis Including ELF Test,” PLoS One 11, no. 2 (2016): e0149230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Reiberger T., Puspok A., Schoder M., et al., “Austrian Consensus Guidelines on the Management and Treatment of Portal Hypertension (Billroth III),” Wiener Klinische Wochenschrift 129, no. Suppl 3 (2017): 135–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Jachs M., Hartl L., Simbrunner B., et al., “The Sequential Application of Baveno VII Criteria and VITRO Score Improves Diagnosis of Clinically Significant Portal Hypertension,” Clinical Gastroenterology and Hepatology 21, no. 7 (2023): 1854–1863. [DOI] [PubMed] [Google Scholar]
- 7. Dominik N., Scheiner B., Zanetto A., et al., “Von Willebrand Factor for Outcome Prediction Within Different Clinical Stages of Advanced Chronic Liver Disease,” Alimentary Pharmacology & Therapeutics 59, no. 11 (2024): 1376–1386. [DOI] [PubMed] [Google Scholar]
- 8. Mandorfer M., Schwabl P., Paternostro R., et al., “Von Willebrand Factor Indicates Bacterial Translocation, Inflammation, and Procoagulant Imbalance and Predicts Complications Independently of Portal Hypertension Severity,” Alimentary Pharmacology & Therapeutics 47, no. 7 (2018): 980–988. [DOI] [PubMed] [Google Scholar]
- 9. Ferlitsch M., Reiberger T., Hoke M., et al., “von Willebrand Factor as New Noninvasive Predictor of Portal Hypertension, Decompensation and Mortality in Patients With Liver Cirrhosis,” Hepatology 56, no. 4 (2012): 1439–1447. [DOI] [PubMed] [Google Scholar]
- 10. La Mura V., Reverter J. C., Flores‐Arroyo A., et al., “Von Willebrand Factor Levels Predict Clinical Outcome in Patients With Cirrhosis and Portal Hypertension,” Gut 60, no. 8 (2011): 1133–1138. [DOI] [PubMed] [Google Scholar]
- 11. Llop E., Berzigotti A., Reig M., et al., “Assessment of Portal Hypertension by Transient Elastography in Patients With Compensated Cirrhosis and Potentially Resectable Liver Tumors,” Journal of Hepatology 56, no. 1 (2012): 103–108. [DOI] [PubMed] [Google Scholar]
- 12. Allaire M., Campion B., Demory A., et al., “Baveno VI and VII Criteria Are Not Suitable for Screening for Large Varices or Clinically Significant Portal Hypertension in Patients With Hepatocellular Carcinoma,” Alimentary Pharmacology & Therapeutics 58, no. 3 (2023): 346–356. [DOI] [PubMed] [Google Scholar]
- 13. Berzigotti A., Reig M., Abraldes J. G., Bosch J., and Bruix J., “Portal Hypertension and the Outcome of Surgery for Hepatocellular Carcinoma in Compensated Cirrhosis: A Systematic Review and Meta‐Analysis,” Hepatology 61, no. 2 (2015): 526–536. [DOI] [PubMed] [Google Scholar]
- 14. Pereyra D., Mandorfer M., Santol J., et al., “Von Willebrand Factor Antigen Improves Risk Stratification for Patients With a Diagnosis of Resectable Hepatocellular Carcinoma,” Annals of Surgical Oncology 31 (2024): 6526–6536. [DOI] [PubMed] [Google Scholar]
- 15. Reiberger T., Schwabl P., Trauner M., Peck‐Radosavljevic M., and Mandorfer M., “Measurement of the Hepatic Venous Pressure Gradient and Transjugular Liver Biopsy,” Journal of Visualized Experiments 160 (2020). [DOI] [PubMed] [Google Scholar]
- 16. Bettinger D., Berzigotti A., Mandorfer M., et al., “Transjugular Diagnostic Procedures in Hepatology: Indications, Techniques and Interpretation,” Journal of Hepatology Reports 7, no. 8 (2025): 101437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Ferlitsch A., Bota S., Paternostro R., et al., “Evaluation of a New Balloon Occlusion Catheter Specifically Designed for Measurement of Hepatic Venous Pressure Gradient,” Liver International 35, no. 9 (2015): 2115–2120. [DOI] [PubMed] [Google Scholar]
- 18. Thabut D. and Kudo M., “Treatment of Portal Hypertension in Patients With HCC in the Era of Baveno VII,” Journal of Hepatology 78, no. 3 (2023): 658–662. [DOI] [PubMed] [Google Scholar]
- 19. Mandorfer M., Abraldes J. G., and Berzigotti A., “Non‐Invasive Assessment of Portal Hypertension: Liver Stiffness and Beyond,” Journal of Hepatology Reports 7, no. 3 (2025): 101300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Azoulay D., Ramos E., Casellas‐Robert M., et al., “Liver Resection for Hepatocellular Carcinoma in Patients With Clinically Significant Portal Hypertension,” Journal of Hepatology Reports 3, no. 1 (2021): 100190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Busch F., De Paepe K. N., Gibbs P., et al., “The Clinical Value of the Hepatic Venous Pressure Gradient in Patients Undergoing Hepatic Resection for Hepatocellular Carcinoma With or Without Liver Cirrhosis,” Open Medicine (Warsaw, Poland) 19, no. 1 (2024): 20230851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Cucchetti A., Cescon M., Golfieri R., et al., “Hepatic Venous Pressure Gradient in the Preoperative Assessment of Patients With Resectable Hepatocellular Carcinoma,” Journal of Hepatology 64, no. 1 (2016): 79–86. [DOI] [PubMed] [Google Scholar]
- 23. Cucchetti A., Ercolani G., Vivarelli M., et al., “Is Portal Hypertension a Contraindication to Hepatic Resection?,” Annals of Surgery 250, no. 6 (2009): 922–928. [DOI] [PubMed] [Google Scholar]
- 24. Iranmanesh P., Vazquez O., Terraz S., et al., “Accurate Computed Tomography‐Based Portal Pressure Assessment in Patients With Hepatocellular Carcinoma,” Journal of Hepatology 60, no. 5 (2014): 969–974. [DOI] [PubMed] [Google Scholar]
- 25. Kim T., Kim B. W., Wang H. J., et al., “Quantitative Assessment of the Portal Pressure for the Liver Surgery Using Serological Tests,” Annals of Surgery 264, no. 2 (2016): 330–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Molina V., Sampson‐Dávila J., Ferrer J., et al., “Benefits of Laparoscopic Liver Resection in Patients With Hepatocellular Carcinoma and Portal Hypertension: A Case‐Matched Study,” Surgical Endoscopy 32, no. 5 (2018): 2345–2354. [DOI] [PubMed] [Google Scholar]
- 27. Ursu C. P., Ciocan A., Ursu Ș., et al., “Prognostic Indicators of Overall Survival in Hepatocellular Carcinoma Patients Undergoing Liver Resection,” Cancers (Basel) 16, no. 7 (2024): 1427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Wang T., Du Y. N., Sun J., et al., “Drug‐Eluting Bead Transarterial Chemoembolization Could Improve the Hepatic Hemodynamics of Patients With Unresectable Hepatocellular Carcinoma: A Retrospective Cohort Study,” Journal of Gastrointestinal Oncology 14, no. 1 (2023): 302–311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Hobeika C., Guyard C., Sartoris R., et al., “Performance of Non‐Invasive Biomarkers Compared With Invasive Methods for Risk Prediction of Posthepatectomy Liver Failure in Hepatocellular Carcinoma,” British Journal of Surgery 109, no. 5 (2022): 455–463. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Baseline characteristics of the overall cohort.
Table S2: Comparison of model characteristics for the prediction of CSPH in the overall cohort.
Table S3: Comparison of patient characteristics according to localisation of tumour, that is, right versus left liver lobe only, and presence versus absence of CSPH in the overall cohort.
Figure S1: Area under the receiver‐operating curve of distinct non‐invasive tests of interest for CSPH and spearman correlation plot of non‐invasive tests and HVPG in the overall cohort. The size of the circles represents the correlation strength.
Figure S2: Scatter plots of HVPG according to (A) liver stiffness, (B) platelet count, (C) VWF:Ag, and (D) VITRO score in the overall cohort. To visualise the diagnostic accuracy of cut‐offs for CSPH, scatter plots were coloured in dark blue (rule‐in CSPH) or light blue (rule‐out CSPH). Each point of the scatterplot represents individual patient data.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.