Effect of hepatic venous pressure gradient in different hepatic venous branches on assessing portal hypertension in cirrhotic patients

Abstract

BACKGROUND

Data regarding the hepatic venous pressure gradient (HVPG) in different hepatic venous branches remain insufficient.

AIM

To assess portal hypertension using HVPG measurements from the right HVPG (R-HVPG), middle HVPG (M-HVPG), and left HVPG (L-HVPG) hepatic veins, we conducted two sub-studies: Sub-study 1 evaluated the diagnostic value of R-HVPG, M-HVPG and L-HVPG, while sub-study 2 explored the correlation between M-HVPG and portal pressure gradient (PPG) to validate the results of sub-study 1.

METHODS

We prospectively enrolled 88 patients who underwent HVPG measurements in different hepatic venous branches in sub-study 1. Sub-group 2 covered 16 cirrhotic patients for M-HVPG and PPG comparisons (NCT06358092). Agreement was analyzed using intraclass correlation coefficients (ICC) and Bland-Altman plots.

RESULTS

In sub-study 1, M-HVPG demonstrated a significantly higher technical success rate (96.6%) than standard R-HVPG (83.0%) and L-HVPG (93.2%) (P < 0.05). After excluding 15 patients with collateral shunts in the R-HVPG, 73 were included in the final analysis. The mean R-HVPG, M-HVPG and L-HVPG were 16.43 (5.34), 16.98 (5.69) and 15.29 (5.22) mmHg, respectively, with no statistically significant difference (P > 0.05). Agreement between the R-HVPG and M-HVPG was substantial [ICC: 0.70 (0.56-0.80)], superior to the moderate R-HVPG/L-HVPG agreement [ICC: 0.56 (0.37-0.70)]. In sub-study 2, the M-HVPG and PPG showed substantial agreement [ICC: 0.80 (0.52-0.92)].

CONCLUSION

Our study demonstrates that M-HVPG achieves the highest technical success rate and superior agreement with R-HVPG, which supports the priority of technical feasibility of HVPG proposed in Baveno VII consensus. Furthermore, the strong correlation with PPG suggests that the M-HVPG potentially reflects accurate portal pressure. Consequently, M-HVPG may be considered a preferred alternative when R-HVPG is unfeasible or unreliable.

Key Words: Cirrhosis; Portal hypertension; Agreement; Hepatic venous pressure gradient; Portal pressure gradient

Core Tip: Middle hepatic venous pressure gradient (M-HVPG) achieves the highest technical success rate and superior agreement with right hepatic venous pressure gradient (HVPG) than that of left HVPG. M-HVPG potentially reflects accurate portal pressure. M-HVPG may be considered a preferred alternative when right HVPG is unfeasible or unreliable.

INTRODUCTION

Portal hypertension is the main driver of decompensation and liver-related mortality in cirrhosis, making its assessment crucial for pathophysiological and therapeutic research[1]. While direct measurement of portal vein pressure is the physiological ideal, it is technically challenging because of the vessel’s anatomical position between the hepatic sinusoids and splanchnic capillaries. Consequently, the hepatic venous pressure gradient (HVPG), calculated as the difference between wedged hepatic venous pressure (WHVP) and free hepatic venous pressure (FHVP), has emerged as the standard minimally invasive surrogate[2]. It correlates strongly with direct portal pressure, particularly in patients with viral hepatitis- and alcohol-related cirrhosis[3].

The Baveno VII consensus emphasizes the necessity of stratifying patients based on liver disease severity and portal pressure dynamics to guide personalized management[1]. As the gold standard for this assessment, HVPG provides critical prognostic utility through well-established risk thresholds. Specifically, an HVPG ≥ 10 mmHg defines clinically significant portal hypertension, serving as an independent predictor of hepatic decompensation, with each one mmHg increase escalating the relative risk of decompensation by approximately 11%[1,4]. Furthermore, exceeding the 12 mmHg threshold significantly elevates the risk of variceal bleeding, while reducing pressure below this level offers protective benefits[5-7]. Higher thresholds of 16 mmHg and 20 mmHg are indicative of increased mortality risk and treatment failure (such as early rebleeding) during acute variceal bleeding episodes, respectively[8-10]. Given these precise clinical implications, ensuring the technical feasibility and accuracy of HVPG measurement is paramount for both clinical prognostication and drug development.

Standard HVPG measurement typically targets the right HVPG (R-HVPG)[11,12]. However, R-HVPG cannulation can be technically unsuccessful in certain patients owing to anatomical variations, such as ostial stenosis or right posterior lobe atrophy associated with venous malformation. Furthermore, the presence of intrahepatic veno-venous shunts can introduce systematic errors, resulting in falsely low HVPG readings that do not reflect the true portal pressure gradient (PPG)[13]. In such case, the middle HVPG (M-HVPG) and left HVPG (L-HVPG) serve as potential alternative access sites. Nevertheless, the degree of hepatic fibrosis may be heterogeneously distributed, and pressure readings could theoretically fluctuate between different liver segments[14]. Currently, there is a lack of evidence demonstrating HVPG in different hepatic venous branches on assessing portal hypertension.

Therefore, we performed two prospective sub-studies. Sub-study 1 aimed to determine the effect of HVPG in different hepatic venous branches on assessing portal hypertension in patients with hepatitis B virus (HBV)- and alcohol-related cirrhosis. Sub-study 2 explored the correlation between M-HVPG and PPG to validate sub-study 1.

MATERIALS AND METHODS

Study population and design

This prospective study recruited cirrhotic patients consecutively at the Third Affiliated Hospital of Sun Yat-Sen University from January 2024 to December 2025 (approval No. ZSSYXHNK2401; ClinicalTrials.gov identifier: NCT06358092).

In sub-study 1, we enrolled patients who underwent HVPG measurements across three hepatic veins as standalone assessment on the risk of portal hypertension. The inclusion criteria were as follows: (1) Adult patients (age, 18-75 years); (2) Confirmed cirrhosis based on previous compatible clinical, biochemical, and radiological findings; (3) HBV- and alcohol-related cirrhosis; (4) Scheduled to undergo clinically indicated transjugular HVPG measurement by means of catheterization of a hepatic vein; and (5) Written informed consent. The exclusion criteria were as follows: (1) Severe coagulation dysfunction; (2) Unfeasible HVPG measurement via the R-HVPG due to stenosis or shunt; (3) Prior transjugular intrahepatic portosystemic stent shunt surgery; (4) Hepatic carcinoma; (5) Prior liver resection or splenectomy; (6) Prior devascularization surgery; (7) Prior liver transplant; (8) Ongoing treatment with vasoactive drugs; and (9) Any active, serious, or life-threatening disease.

Sub-study 2 to observe the correlation between M-HVPG and PPG enrolled cirrhotic patients scheduled to undergo transjugular intrahepatic portosystemic stent shunt at the Third Affiliated Hospital of Sun Yat-sen University at the same time. The inclusion criteria were as follows: (1) Adult patients (age 18-75 years); (2) Confirmed cirrhosis based on previous compatible clinical, biochemical, and radiological findings; (3) HBV- and alcohol-related cirrhosis; (4) Scheduled to undergo clinically indicated transjugular intrahepatic portosystemic stent; and (5) Written informed consent. The exclusion criteria were as follows: (1) Severe coagulation dysfunction; (2) Hepatocellular carcinoma; (3) Prior liver resection or splenectomy; (4) Prior devascularization surgery; (5) Prior liver transplant; (6) Ongoing treatment with vasoactive drugs; and (7) Any active, serious, or life-threatening disease.

The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of The Third Affiliated Hospital of Sun Yat-sen University (approval No. RG2024-195-01). Written informed consent was obtained from all enrolled participants.

Measurement of HVPG and PPG

All HVPG and PPG measurements were performed by trained interventional radiologists with over 10 years of experience following the standard operating procedure (Wu C, Chang BY and Zhou CR)[1,15]. Strict standards were used to obtain an accurate HVPG, including the thickness of the catheter, the position of the catheter tip, the method and times of measurement, and so on, factors that could affect venous pressure were avoided before the operation, including discontinuation of drugs that affect portal pressure, such as nonselective beta-blockers and propofol, which can affect the HVPG and PPG during deep sedation[16].

The puncture site was disinfected and local routine sterile drapes were then placed. Under local anesthesia, HVPG measurements were conducted using a RUPS-100 set (COOK Medical, IN, United States) and a balloon catheter with a pressure transducer at the tip (Edwards Lifesciences, CA, United States) in all three hepatic venous branches. Zero measurement was needed before transjugular catheterization. The FHVP was measured in the R-HVPG (1-3 cm from the inferior vena cava). Then, as the balloon was inflated for total occlusion of the right, middle and left hepatic vein, the WHVP was measured venography of the occluded hepatic vein was performed by using a power injector to inject five mL of lopromide contrast medium (Ultravist 300; Bayer AG, Leverkusen, Germany) at a rate of two mL/s to confirm total occlusion of the balloon and to exclude the presence of hepatic venous-to-venous communications at the end of the WHVP measurement (Figure 1). Continuous recordings were performed until the pressure plateaued. Similarly, the above process was performed in the middle and left hepatic venous branches. R-, M- and L-HVPG were measured in fixed order. The HVPG was defined as the difference between the WHVP and FHVP. In this study, each HVPG measurement was based on three repeated measurements of WHVP and FHVP, the average of at least two measurements with a difference of < one mmHg was obtained, and the HVPG was calculated.

Figure 1 Hepatic venous pressure gradient measurement. A: Right free hepatic venous pressure (FHVP); B: Middle-FHVP; C: Left-FHVP; D: Right wedged hepatic venous pressure (WHVP); E: Middle-WHVP; F: Left-WHVP.

In the pilot cohort, the inferior vena cava pressure was measured in the inferior vena cava during HVPG measurement. The liver parenchyma and portal vein were then punctured through the inferior vena cava or the hepatic vein. After successful puncture of the portal vein, the pigtail or end-to-side hole catheter was inserted into the splenic vein or superior mesenteric vein for intravenous angiography. Before shunting, the pressure in the portal vein was measured (three measurements, averaged) and the PPG value was calculated. PPG was defined as the difference between portal venous pressure and inferior vena cava.

Statistical analysis

The primary outcome was agreement among the readings of the R-HVPG, M-HVPG, and L-HVPG. Secondary outcomes included the measurement success rate of M-HVPG and L-HVPG, and the correlation and agreement between the M-HVPG and PPG. The technical success of HVPG measurements was defined as: Successful transjugular catheterization, venography, and recording of plateaued pressure. The exploratory purpose was to determine the diagnostic value of M-HVPG and L-HVPG for determination of R-HVPG ≥ 10, 12, 16, and 20 mmHg.

Categorical data were expressed as n (%), and continuous variables were expressed as mean ± SD or median (interquartile range). Intraclass correlation coefficients (ICC) and Bland-Altman plots were used to analyze the agreement. The magnitude of the ICC was interpreted according to the Landis and Koch convention: A poor agreement of 0, a slight agreement between 0 and 0.20, a fair agreement between 0.21 and 0.40, a moderate agreement between 0.41 and 0.60, a substantial agreement between 0.61 and 0.80, and an almost-perfect agreement between 0.81 and 1.00[17]. Major discrepancies between the two techniques were considered when the difference was > 5 mmHg. Diagnostic performance for detecting different cutoffs of R-HVPG was assessed by the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity, and 95% confidence interval (CI) was computed using the method of Delong (1988). The cut-off values of the M-HVPG and L-HVPG were defined as the maximal sum of sensitivity and specificity. The AUC estimates the probability of the criteria to correctly distinguish between two randomly selected patients, one of whom was randomly selected from all those with the condition of interest and the other was randomly selected from all those without the condition of interest. Statistical significance was set at P < 0.05. All statistical calculations were performed using R language (version 4.4.3, R Core Team, 2025).

RESULTS

Study population

The flow diagram of the study population is shown in Figure 2. In sub-study 1, 88 patients with HBV- and alcohol-related cirrhosis underwent HVPG measurements in different hepatic venous branches. All patients finished successful venipuncture and catheterization. Of them, 15 patients with collateral shunts in the R-HVPG were excluded (successful cannulation but unreliable measurement due to collateral shunting). Two of them with cavernous transformation of portal veins failed to perform HVPG measurements in all three hepatic venous branches, and two patients failed in left hepatic veins (besides the right) due to incomplete occlusion. In addition, there were one patient failing only in the middle and two only in the left hepatic venous all due to collateral shunting, respectively. The M-HVPG had a better technical success rate than the R-HVPG and L-HVPG (96.6% vs 83.0% vs 93.2%, P < 0.05). Finally, 73 patients (mean age 53.58 years; male, 83.6%) were included in analysis. These included 43 (58.9%) patients with HBV, 20 (27.4%) with alcohol-related liver disease (ALD), and 10 (13.7%) with HBV + ALD. The number of Child-Pugh classes A, B, and C was 52 (71.2%), 18 (24.6%), and 3 (4.1%), respectively. There were 59 decompensated patients. Of them, 54 patients occurred variceal bleeding, 11 ascites needing treatment and one hepatic encephalopathy. The baseline characteristics of the enrolled patients were summarized in Table 1.

Figure 2 Flow diagram of study population. HVPG: Hepatic venous pressure gradient; PPG: Portal pressure gradient; HBV: Hepatitis B virus; ALD: Alcohol-related liver disease.

Table 1 Baseline characteristics of patients in sub-study 1.

	All (n = 73)	HBV (n = 43)	ALD (n = 20)	HBV + ALD (n = 10)
Age (year), mean ± SD	53.58 ± 9.10	55.50 ± 9.00	51.10 ± 9.51	50.60 ± 10.24
Male	61 (83.6)	32 (74.4)	19 (95.0)	10 (100.0)
BMI (kg/m2)	22.84 (3.28)	22.12 (3.14)	23.48 (3.36)	23.57 (3.04)
Decompensation, n (%)	59 (80.8)	37 (86.0)	13 (65.0)	9 (90.0)
Variceal bleeding	54 (74.0)	36 (83.7)	9 (45.0)	9 (90.0)
Ascites needing treatment	11 (15.1)	4 (9.3)	5 (25.0)	2 (20.0)
Hepatic encephalopathy	1 (1.4)	0 (0)	1 (5.0)	0 (0)
CSPH, n (%)	63 (86.3)	35 (81.4)	18 (90.0)	10 (100.0)
Etiology, n (%)
HBV	43 (58.9)	NA	NA	NA
ALD	20 (27.4)	NA	NA	NA
HBV and ALD	10 (13.7)	NA	NA	NA
Child-Pugh class, n (%)
Class A	52 (71.2)	35 (81.4)	11 (55.0)	6 (60.0)
Class B	18 (24.6)	7 (16.3)	7 (35.0)	4 (40.0)
Class C	3 (4.1)	1 (2.3)	2 (10.0)	0 (0)
Laboratory test, median (IQR)
PLT (109/L)	54.00 (11.00)	72.00 (52.50)	87.00 (54.75)	72.00 (33.00)
TBIL (μmol/L)	19.60 (19.25)	15.40 (11.10)	27.90 (27.35)	22.35 (22.90)
Hb (109/L)	101.00 (43.00)	109.00 (46.00)	101.00 (50.00)	76.50 (47.50)
ALT (U/L)	25.00 (20.00)	24.00 (15.00)	29.00 (22.25)	25.00 (13.00)
AST (U/L)	34.00 (24.50)	30.00 (13.50)	62.00 (73.75)	40.00 (20.00)
GGT (U/L)	41.50 (67.50)	35.00 (31.25)	132.00 (194.00)	27.00 (57.00)
Alb (g/L)	34.55 (6.50)	37.20 (9.00)	33.00 (5.70)	33.20 (3.90)
Cr (μmol/L)	68.00 (21.00)	72.00 (23.50)	61.00 (15.25)	65.50 (18.00)
PT (seconds)	15.90 (2.50)	15.85 (2.70)	15.95 (3.10)	16.50 (2.10)
INR	1.27 (0.26)	1.26 (0.27)	1.30 (0.32)	1.34 (0.23)

HBV: Hepatitis B virus; ALD: Alcoholic-related liver disease; BMI: Body mass index; CSPH: Clinically significant portal hypertension; IQR: Interquartile range; PLT: Platelet counts; TBIL: Total bilirubin; Hb: Hemoglobin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; GGT: Gamma-glutamyl transpeptidase; Alb: Albumin; Cr: Creatinine; PT: Prothrombin time; INR: International normalized ratio.

In sub-study 2, we enrolled 16 patients (mean age 50.50 years; male, 81.3%) underwent M-HVPG and PPG measurements. There were 13 (81.3%) patients with HBV and 3 (18.7%) with ALD. The number of Child-Pugh classes A, B, and C was 9 (56.3%), 6 (37.5%), and 1 (6.2%), respectively (Table 2).

Table 2 Baseline characteristics of patients in sub-study 2.

	All (n = 16)
Age (year), mean ± SD	50.50 ± 10.73
Male, n (%)	13 (81.3)
HVPG, mean ± SD	23.44 ± 4.29
PPG, mean ± SD	23.63 ± 4.22
Etiology, n (%)
HBV	13 (81.3)
ALD	3 (18.7)
Child-Pugh class, n (%)
Class A	9 (56.3)
Class B	6 (37.5)
Class C	1 (6.2)
Laboratory test, median (IQR)
PLT (109/L)	78.50 (90.5)
TBIL (μmol/L)	26.00 (37.90)
Alb (g/L)	33.90 (8.80)
PT (seconds)	16.65 (1.50)

HVPG: Hepatic venous pressure gradient; PPG: Portal pressure gradient; HBV: Hepatitis B virus; ALD: Alcoholic-related liver disease; IQR: Interquartile range; PLT: Platelet counts; TBIL: Total bilirubin; Alb: Albumin; PT: Prothrombin time.

Measurement of R-HVPG, M-HVPG and L-HVPG

The mean R-HVPG, M-HVPG and L-HVPG were 16.43 (5.34), 16.98 (5.69) and 15.29 (5.22), respectively, with no statistically significant difference (R-HVPG and M-HVPG, P = 0.35; R-HVPG and L-HVPG, P = 0.13; M-HVPG and L-HVPG, P = 0.07) (Table 3, Figure 3A). Major discrepancies were found in 16 patients (21.9%) between the R-HVPG and M-HVPG, 14 (19.2%) between the R-HVPG and L-HVPG, and 14 (19.2%) between the M-HVPG and L-HVPG.

Figure 3 Bland-Altman plots and box plots of right hepatic venous pressure gradient, middle hepatic venous pressure gradient and left hepatic venous pressure gradient. A: Hepatic venous pressure gradient (HVPG) in all patients; B: HVPG in hepatitis B virus; C: HVPG in alcohol-related liver disease; D: HVPG in hepatitis B virus + alcohol-related liver disease. L-HVPG: Left hepatic venous pressure gradient; M-HVPG: Middle hepatic venous pressure gradient; R-HVPG: Right hepatic venous pressure gradient; HBV: Hepatitis B virus; ALD: Alcohol-related liver disease; NS: Not significant.

Table 3 Measurement of right hepatic venous pressure gradient, middle hepatic venous pressure gradient and left hepatic venous pressure gradient, mean ± SD.

	R-HVPG	M-HVPG	L-HVPG
All	16.43 ± 5.34	16.98 ± 5.69	15.29 ± 5.22
HBV	15.60 ± 5.13	16.65 ± 5.55	14.53 ± 5.21
ALD	17.65 ± 6.25	18.10 ± 5.99	15.74 ± 4.47
HBV + ALD	17.60 ± 3.89	16.30 ± 5.98	17.78 ± 6.34

HVPG: Hepatic venous pressure gradient; HBV: Hepatitis B virus; ALD: Alcoholic-related liver disease.

In patients with HBV, the mean R-HVPG was similar to that of the M-HVPG [15.60 (5.13) vs 16.65 (5.55), P = 0.61]. The results were same in R-HVPG and L-HVPG [15.60 (5.13) vs 14.53 (5.21), P = 0.91], M-HVPG and L-HVPG [16.65 (5.55) vs 14.53 (5.21), P = 0.69] (Table 3, Figure 3B). In the patients with ALD, the mean R-HVPG, M-HVPG and L-HVPG were 17.65 (6.25), 18.10 (5.99) and 15.74 (4.47), respectively, with no statistically significant difference (R-HVPG and M-HVPG, P = 0.86; R-HVPG-L-HVPG, P = 0.16; M-HVPG and L-HVPG, P = 0.22) (Table 3, Figure 3C). In the patients with HBV + ALD, R-HVPG, M-HVPG and L-HVPG were 17.60 (3.89), 16.30 (5.98) and 17.78 (6.34), respectively, with no statistically significant difference (R-HVPG and M-HVPG, P = 0.22; R-HVPG and L-HVPG, P = 0.17; M-HVPG and LHVPG, P = 0.86) (Table 3, Figure 3D).

Agreement of R-HVPG, M-HVPG and L-HVPG

When analyzing the agreement between the R-HVPG and M-HVPG, an ICC of 0.70 (95%CI: 0.56-0.80) was obtained, suggesting a substantial agreement. The ICC was 0.56 (95%CI: 0.37-0.70) with a moderate agreement between R-HVPG and L-HVPG and 0.47 (95%CI: 0.27-0.63) with a moderate agreement between M-HVPG and L-HVPG (Table 4, Figure 3A).

Table 4 Agreement of right hepatic venous pressure gradient, middle hepatic venous pressure gradient and left hepatic venous pressure gradient.

	R-HVPG-M-HVPG	R-HVPG-L-HVPG	M-HVPG-L-HVPG
All	0.70 (0.56-0.80)	0.56 (0.37-0.70)	0.47 (0.27-0.63)
HBV	0.70 (0.51-0.82)	0.56 (0.31-0.74)	0.55 (0.31-0.72)
ALD	0.63 (0.26-0.84)	0.65 (0.33-0.83)	0.42 (0.01-0.71)
HBV + ALD	0.83 (0.63-0.93)	0.22 (0-0.71)	0.26 (0-0.75)

L-HVPG: Left hepatic venous pressure gradient; M-HVPG: Middle hepatic venous pressure gradient; R-HVPG: Right hepatic venous pressure gradient; HBV: Hepatitis B virus; ALD: Alcoholic-related liver disease.

In the patients with HBV, the ICC was 0.70 (0.51-0.82) with a substantial agreement between R-HVPG and M-HVPG, 0.56 (0.31-0.74) with a moderate agreement between R-HVPG and L-HVPG, 0.55 (0.31-0.72) with a moderate agreement between M-HVPG and L-HVPG (Table 4, Figure 3B). In the patients with ALD, the ICC was 0.63 (0.26-0.84) with a substantial agreement between R-HVPG and M-HVPG, 0.65 (0.33-0.83) with a substantial agreement between R-HVPG and L-HVPG, 0.42 (0.01-0.71) with a moderate agreement between M-HVPG and L-HVPG (Table 4, Figure 3C). In the patients with HBV + ALD, the ICC was 0.83 (0.63-0.93) with an almost-perfect between R-HVPG and M-HVPG, 0.22 (0-0.71) with a fair agreement between R-HVPG and L-HVPG, 0.26 (0-0.75) with a fair agreement between M-HVPG and L-HVPG (Table 4, Figure 3D).

Correlation and agreement between M-HVPG and PPG

In sub-study 2, the mean M-HVPG and PPG were 23.44 (4.29) and 23.63 (4.22), respectively. An ICC between M-HVPG and PPG of 0.80 (95%CI: 0.52-0.92) was obtained, suggesting a substantial agreement, with one case (6.3%) outside the 1.96 SD according to the Bland-Altman plot. The ICC for absolute agreement (0.799) and consistency (0.801) remained the same, suggesting the absence of systematic error (Figure 4). Major discrepancies between the M-HVPG and PPG measurements were found in only 1/16 patients (6.3%). Diagnosis performance of M-HVPG, L-HVPG and their average for identifying R-HVPG ≥ 10, 12, and 16.20 mmHg.

Figure 4 Middle-hepatic venous pressure gradient and portal pressure gradient measurement in the pilot cohort. A: Correlation between middle-hepatic venous pressure gradient and portal pressure gradient; B: Agreement between middle-hepatic venous pressure gradient and portal pressure gradient according to Bland-Altman plots. M-HVPG: Middle hepatic venous pressure gradient; PPG: Portal pressure gradient.

The average of M-HVPG and L-HVPG showed the best performance than single M-HVPG and L-HVPG on determining R-HVPG ≥ 10 mmHg, with a cutoff of 12.75, AUC of 0.85 (95%CI: 0.67-1.00), specificity of 0.89 (95%CI: 0.67-1.00) and sensitivity of 0.86 (95%CI: 0.76-0.95); R-HVPG ≥ 12 mmHg, with a cutoff of 14.25, AUC of 0.83 (95%CI: 0.71-0.94), specificity of 0.71 (95%CI: 0.47-0.88) and sensitivity of 0.82 (95%CI: 0.70-0.92); R-HVPG ≥ 16 mmHg, with a cutoff of 17.75, AUC of 0.89 (95%CI: 0.81-0.97), specificity of 0.89 (95%CI: 0.79-0.97) and sensitivity of 0.76 (95%CI: 0.59-0.90); R-HVPG ≥ 20 mmHg, with a cutoff of 17.75, AUC of 0.90 (95%CI: 0.82-0.99), specificity of 0.74 (95%CI: 0.63-0.85) and sensitivity of 0.92 (95%CI: 0.77-1.00) (Supplementary Table 1, Supplementary Figure 1A). The AUCs of M-HVPG and L-HVPG were as follows: R-HVPG ≥ 10 mmHg, 0.78 (95%CI: 0.60-0.95) and 0.82 (95%CI: 0.62-1.00); R-HVPG ≥ 12 mmHg, 0.78 (95%CI: 0.65-0.91) and 0.79 (95%CI: 0.66-0.92); R-HVPG ≥ 16 mmHg, 0.87 (95%CI: 0.79-0.96) and 0.82 (95%CI: 0.71-0.92); R-HVPG ≥ 20 mmHg, 0.86 (95%CI: 0.75-0.98) and 0.83 (95%CI: 0.73-0.93), respectively (Supplementary Table 1, Supplementary Figure 1A).

We performed a subgroup analysis to explore the value of the M-HVPG and L-HVPG in detecting R-HVPG ≥ 10, 12, 16, and 20 mmHg in patients with HBV- and alcohol-related cirrhosis. In HBV patients, the AUCs of M-HVPG, L-HVPG and their average were as follows: R-HVPG ≥ 10 mmHg, 0.76 (95%CI: 0.41-1.00), 0.59 (95%CI: 0.08-1.00) and 0.84 (95%CI: 0.56-1.00); R-HVPG ≥ 12 mmHg, 0.91 (95%CI: 0.81-1.00), 0.90 (95%CI: 0.80-1.00) and 0.89 (95%CI: 0.78-1.00); R-HVPG ≥ 16 mmHg, 0.85 (95%CI: 0.73-0.97), 0.79 (95%CI: 0.64-0.94) and 0.88 (95%CI: 0.79-0.98); R-HVPG ≥ 20 mmHg, 0.92 (95%CI: 0.84-1.00), 0.75 (95%CI: 0.54-0.96) and 0.94 (0.86-1.00) (Supplementary Table 2, Supplementary Figure 1B). In ALD patients, the AUCs of M-HVPG, L-HVPG and their average were as follows: R-HVPG ≥ 10 mmHg, 0.78 (95%CI: 0.56-0.94), 0.96 (95%CI: 0.86-1.00) and 0.97 (95%CI: 0.90-1.00); R-HVPG ≥ 12 mmHg, 0.79 (95%CI: 0.51-1.00), 0.82 (95%CI: 0.59-1.00) and 0.79 (95%CI: 0.55-1.00); R-HVPG ≥ 16 mmHg, 0.84 (95%CI: 0.65-1.00), 0.80 (95%CI: 0.59-1.00) and 0.86 (95%CI: 0.69-1.00); R-HVPG ≥ 20 mmHg, 0.85 (95%CI: 0.70-1.00), 0.85 (95%CI: 0.68-1.00) and 0.91 (0.77-1.00) (Supplementary Table 3, Supplementary Figure 1C).

DISCUSSION

Measurement of HVPG has become the cornerstone for managing advanced chronic liver disease. Following the Baveno VII consensus and landmark trials such as PREDESCI, HVPG-guided stratification should be encouraged in clinical trials investigating novel therapies[1,18-20]. Its prognostic value extends to predicting variceal hemorrhage, hepatocellular carcinoma risk, and post-resection outcomes[21,22]. However, the widespread application of this “gold standard” relies heavily on the technical feasibility of the procedure. While strict cut-offs (e.g., 10, 12, and 16 mmHg) are well-established[23-28], they depend on obtaining accurate pressure readings, which can be challenging in real-world clinical settings.

Classically, the R-HVPG is the preferred access site for HVPG measurements because of its larger caliber and accessible trajectory. However, relying solely on the R-HVPG is not always feasible. Technical failure occurs in a significant proportion of patients owing to anatomical constraints, such as ostial stenosis or lobe atrophy. Furthermore, hemodynamic distortions, specifically intrahepatic venovenous shunts or arterialization of the hepatic vasculature, can compromise the accuracy of R-HVPG readings, leading to underestimation of the true portal pressure[29,30]. These irregularities often stem from the heterogeneous regional perfusion and structural remodeling inherent to cirrhotic livers. Consequently, when R-HVPG catheterization is technically unsuccessful or yields unreliable traces, alternative access routes via the middle or left hepatic veins are necessary. However, the evidence confirming the interchangeability of measurements across these veins remains insufficient.

To date, literature addressing this issue is limited. A study by Keiding and Vilstrup[29] highlighted significant segmental variation, noting that relying on a single vein could misclassify risk in 13% of cases. More recently, Wang et al[31] reported poor correlation between HVPG and direct PPG across all three veins, although they noted that M-HVPG performed marginally better than the others. Unlike these previous reports, our study is the first prospective investigation to systematically compare HVPG measurements across all three hepatic veins within the same patient cohort, while also validating M-HVPG against direct PPG. This design allows for a more definitive assessment of whether the M-HVPG and L-HVPG are valid alternatives for risk stratification when the R-HVPG approach fails.

Our results demonstrated that M-HVPG was a superior alternative to R-HVPG compared with L-HVPG. The M-HVPG showed the highest technical success rate. That might benefit from the broader drainage territory in the middle hepatic vein and less affected by right posterior lobe atrophy or shunts. Furthermore, the agreement between the R-HVPG and M-HVPG was stronger than that between the R-HVPG and L-HVPG. While combining M-HVPG and L-HVPG measurements yielded the highest precision, the M-HVPG alone proved sufficiently robust for clinical use. The finding contradicted one previous study that suggested significant segmental variation. Crucially, in our pilot validation cohort, we observed substantial agreement between the M-HVPG and direct PPG, with major discrepancies occurring in only 6.3% of cases. This strongly suggests that the M-HVPG accurately reflects the actual portal pressure, overcoming the concerns of the poor correlation raised in prior studies.

Subgroup analysis stratified by etiology demonstrated that the consistency of HVPG measurements across the three hepatic veins remained irrelevant to the underlying liver disease. We observed no statistically significant differences in mean pressures among the R-HVPG, M-HVPG, and L-HVPG in patients with HBV, ALD, or HBV + ALD (P > 0.05). This suggests that the hemodynamic uniformity of the hepatic venous system was generally preserved across these major cirrhotic causes, validating the interchangeability of measurements in diverse patient populations. Given that HBV infection constituted the predominant etiology (58.9%), we further evaluated the diagnostic performance of the M-HVPG in this specific cohort. The M-HVPG demonstrated excellent diagnostic accuracy for identifying the key prognostic thresholds defined by the standard R-HVPG. The predictive value was particularly strong for severe portal hypertension (≥ 12 mmHg) and high-risk stages (≥ 20 mmHg), with AUCs exceeding 0.90. Notably, M-HVPG consistently outperformed L-HVPG in predicting these critical cut-offs (e.g., AUC 0.92 vs 0.75 for ≥ 20 mmHg). These findings strongly support the clinical utility of the M-HVPG as a reliable surrogate for risk stratification in HBV-related cirrhosis when R-HVPG assessment is technically challenging.

Our study had several limitations inherent to its observational and cross-sectional design. First, the recruitment of patients from a single tertiary center who underwent complete hemodynamic evaluation might have introduced selection bias, potentially excluding those with subclinical portal hypertension. However, this bias was partially mitigated by our institution’s strict protocol involving routine liver catheterization for cirrhotic workup. Second, the overall sample size was relatively small, particularly for the ALD subgroup. The cohort size was further constrained by the exclusion of patients with intrahepatic shunts, which precluded accurate HVPG measurement. Third, R-, M- and L-HVPG were measured in fixed order. The effect of fixed order for procedural duration, respiratory variation, contrast injection, or hemodynamic drift couldn’t be assessed. Then, radiation dose and contrast volume were not recorded. It potentially led to more complications. Last but not the least, the etiological distribution was skewed, with a predominance of HBV and ALD. Consequently, our findings may not be generalizable to cirrhosis caused by other etiologies. Future multicenter studies with larger cohorts and longitudinal follow-ups are needed to validate these findings and confirm their clinical utility.

CONCLUSION

In conclusion, our study demonstrated that the M-HVPG achieved the highest technical success rate, which supports the priority of technical feasibility of HVPG proposed in Baveno VII consensus. We observed no significant differences among the R-HVPG, M-HVPG, and L-HVPG and superior agreement with the R-HVPG compared to the L-HVPG, a pattern consistent in both HBV- and alcohol-related cirrhosis. Furthermore, the strong correlation between the M-HVPG and PPG suggests that the M-HVPG potentially reflected accurate clinical portal venous pressure. Consequently, we recommend that M-HVPG may be considered a preferred alternative when R-HVPG is unfeasible or unreliable, pending larger multicenter validation. Our results have important implications for individualizing hemodynamic assessments in patients with cirrhosis.