Hepatic vein arrival time as assessed by contrast-enhanced ultrasonography is useful for the assessment of portal hypertension in compensated cirrhosis

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

The measurement of the hepatic venous pressure gradient (HVPG) for the estimation of portal hypertension (PH) in cirrhosis has some limitations, including its invasiveness. Hepatic vein arrival time (HVAT), as assessed by microbubble contrast-enhanced ultrasonography (CEUS), is negatively correlated with the histological grade of liver fibrosis because of the associated hemodynamic abnormalities. Anatomical and pathophysiological changes in liver microcirculation are the initial events leading to PH. However, the direct relationship between HVAT and PH has not been evaluated. The present study measured both HVPG and HVAT in 71 consecutive patients with compensated cirrhosis and analyzed the relationship between the two parameters (i.e., the derivation set). Results were validated in 35 compensated patients with cirrhosis at another medical center (i.e., the validation set). The derivation set had HVPG and HVAT values of 11.4 ± 5.0 mmHg (mean ± standard deviation; range, 2-23) and 14.1 ± 3.4 seconds (range, 8.4-24.2), respectively; there was a statistically significant negative correlation between HVPG and HVAT (r2 = 0.545; P < 0.001). The area under the receiver operating characteristic curve (AUROC) was 0.973 for clinically significant PH (CSPH; HVPG, ≥10 mmHg), and the sensitivity, specificity, positive predictive value, negative predictive value, and positive and negative likelihood ratios for CSPH for an HVAT cut-off value of 14 seconds were 92.7%, 86.7%, 90.5%, 89.7%, 6.95, and 0.08, respectively. In addition, a shorter HVAT was associated with worse Child-Pugh score (P < 0.001) and esophageal varices (P = 0.018). In the validation set, there was also a significant negative correlation between HVAT and HVPG (r2 = 0.538; P < 0.001), and AUROC = 0.953 for CSPH. HVAT was significantly correlated with PH. These results indicate that measuring HVAT is useful for the noninvasive prediction of CSPH in patients with compensated cirrhosis. (HEPATOLOGY 2012;56:1053–1062)

Portal hypertension (PH) leads to serious complications, such as variceal bleeding, and is responsible for significant morbidity and mortality in patients with cirrhosis.1 In compensated cirrhosis in particular, the presence of clinically significant PH (CSPH) makes it possible to predict a poor prognosis, such as progression to decompensated cirrhosis or mortality.2 This makes the precise grading of PH essential for the treatment and follow-up of patients with compensated cirrhosis,3, 4 but PH estimation has not been commonly used in clinical practice because of its associated measurement limitations. The measurement of the hepatic venous pressure gradient (HVPG) has been accepted as the gold standard for assessing the degree of PH.3-6 However, the widespread routine clinical use of this method is limited by its invasiveness and its requirement for skilled expertise and special equipment. This has led to many trials being conducted in attempts to develop an alternative noninvasive method of assessing the severity of PH.

Doppler ultrasonography (US) is one of the potential candidates for the noninvasive investigation of PH, and many investigators have attempted to confirm the usefulness of Doppler US in the assessment of PH in patients with cirrhosis.4, 7, 8 Several Doppler-based methods and indices have been developed and proposed, most of which are related to the measurement of relative flow and velocity in the abdominal splanchnic vessels; however, none of the currently available Doppler-based methods are regarded as being sufficiently sensitive or specific to replace HVPG in clinical practice.9, 10 Microbubble contrast agents that enhance the signal when they are administered intravenously (IV) have been developed for the US evaluation of the vasculature or masses.11-13 It has been reported recently that analysis of the transit time between the hepatic vein (HV) and hepatic artery (HA) or portal vein (PV), using microbubble contrast-enhanced US (CEUS), can be useful for assessing the severity of liver fibrosis.14-16 The HV arrival time (HVAT) is the time (in seconds) taken for the microbubble contrast agent to arrive at the HV after injection. There is an inverse relationship between HVAT and the severity of liver histological grade, such that a reduction in HVAT is accompanied by an increase in the severity of liver disease resulting from the presence of intrahepatic hemodynamic changes in the cirrhotic liver, which can result from capillarization of the sinusoids or arteriovenous or portovenous shunts.10, 17 In this context, we hypothesized that the HVAT value achieved by microbubble CEUS represents the degree of portal pressure; that is, HVAT decreases as HVPG increases. Hence, the present study prospectively evaluated the relationship between HVPG and HVAT and also determined the usefulness of HVAT as a noninvasive method of assessing the severity of PH.

Abbreviations

AUROC, area under ROC curve; CEUS, contrast-enhanced ultrasonography; 95% CI, 95% confidence interval; CSPH, clinically significant PH; CT, computed tomography; CVs, coefficients of variation; HA, hepatic artery; HBV, hepatitis B virus; HV, hepatic vein; HVAT, hepatic vein arrival time; HVPG, hepatic venous pressure gradient; IV, intravenous; –LR, negative likelihood ratio; +LR, positive likelihood ratio; NPV, negative predictive value; PBF, portal blood flow; PH, portal hypertension; PPV, positive predictive value; PV, portal vein; ROC, receiver operating characteristic; ROI, region of interest; SD, standard deviation; Se, sensitivity; Sp, specificity; TICs, time-intensity curves; US, ultrasonography.

Patients and Methods

Study Population and Design.

This study was performed between October 2009 and July 2011. The procedure was divided into two sequential phases. In phase I, 131 consecutive patients with cirrhosis who satisfied the inclusion and exclusion criteria were enrolled between October 2009 and March 2011. HVAT measurement failed in 15 (11.5%) of these patients because of a poor echo window that was related to obesity (n = 4; 3.1%), rib shadow (n = 1; 0.8%), air shadow (n = 2; 1.5%), poor respiratory cooperation (n = 3; 2.3%), and severe right liver lobe atrophic changes related to advanced decompensated cirrhosis (n = 5; 3.8%). Therefore, a final total of 116 patients with cirrhosis (95 men and 21 women; age, 49.4 ± 8.6 years [mean ± standard deviation; SD]; range, 29-69) underwent measurement of both HVPG and HVAT serially on the same day. Of these 116 patients, 71 with compensated cirrhosis (59 men and 12 women; age, 49.2 ± 9.2 years; range, 29-69) were defined as the derivation set for the estimation of the predictive value of HVAT for CSPH. In the phase I study, diagnosis of cirrhosis was confirmed by biopsy in 76 patients as well as by the presence of varices, laboratory data, and imaging results, including US and computed tomography (CT) scans in the remaining patients. Compensated cirrhosis was defined as Child-Pugh class A or B and no episodes of recent gastrointestinal bleeding, including variceal hemorrhage, ascites or peripheral edema, jaundice, or hepatic encephalopathy.1

The phase II study was performed between October 2010 and July 2011 for external validation of the predictive value of HVAT for CSPH in 35 patients with compensated cirrhosis (30 men and 5 women; age, 51.3 ± 6.4 years; range, 40-66; the validation set). The phase II study population was also enrolled using the same inclusion criteria as for the phase I study, which resulted in 35 consecutive patients with compensated cirrhosis who satisfied the inclusion criteria undergoing measurements for both HVPG and HVAT. Biopsy-confirmed cirrhosis was found in 21 patients. The phase I and phase II studies were performed at different medical centers (Yonsei University Wonju College of Medicine, Wonju Christian Hospital, Wonju, Republic of Korea; and Hallym University College of Medicine, Chuncheon Sacred Heart Hospital Chuncheon, Republic of Korea; respectively) and were conducted regardless of the etiology in all enrolled patients. The primary outcome was the relationship between HVPG and HVAT. HVPG and HVAT measurements were performed serially on the same day. Esophagogastroduodenoscopy was performed to confirm the presence of varices, and radiological imaging studies, including US, CT, and laboratory studies, were applied to each patient to determine the clinical stage of their cirrhosis and their Child-Pugh score. Patients who did not provide informed consent to participate were excluded from the study. In addition, patients with hepatocellular carcinoma, severe liver failure (i.e., serum bilirubin level >5 mg/dL or hepatic encephalopathy), spontaneous bacterial peritonitis, hepatorenal syndrome, acute renal impairment, underlying severe cardiac illness, or noncirrhotic PH, or who were receiving nonselective β-blockers, nitrates, or any other pharmacotherapy for the prevention of variceal bleeding, were excluded. Patients whose HVs could not be convincingly identified on US were also excluded. In cases of cirrhosis related to alcohol abuse, those combined with alcoholic hepatitis or recent alcohol intakes (within 4 weeks) were excluded from the present study. In addition, we measured HVAT in 30 healthy controls (23 male and 7 female volunteers; age, 45.6 ± 6.8 years; range, 34-62) and compared the results with the population from the phase I study. Healthy controls were included according to their age and sex if they had no previous medical history. This study was designed according to the ethical guidelines issued by the 2000 revision (Edinburgh) of the 1975 Declaration of Helsinki. The ethics committees of the involved hospitals approved the protocol, and the patients and healthy controls provided written informed consent to participate (Fig. 1).

Figure 1.

Study design and patient enrollment.

Measurement of HVPG.

HVPG was measured after an overnight fast. The right HV was catheterized percutaneously through the femoral vein, and pressure was recorded in both the wedged position and the free position with a 7-F balloon-tipped catheter (Arrow International, Erding, Germany). All measurements were performed at least in triplicate, and permanent tracings were obtained on a multichannel recorder.18 HVPG was determined by subtracting the free HV pressure from the wedged HV pressure.5, 19 During HVPG measurement, VISIPAQUE (iodixanol injection; GE Healthcare Canada Inc., Mississauga, Ontario, Canada) was used as a radiographic contrast medium; VISIPAQUE is isotonic with blood (290 mOsm/kg of water) and causes fewer, less-severe osmolality-related disturbances. CSPH was defined as an HVPG value of equal or above 10 mmHg, according to a consensus definition.20 An examiner with 14 years of experience with HVPG measurement (Y.J.K.) performed all HVPG procedures in the derivation set, whereas HVPG measurements for the validation set were conducted by an examiner with 5 years of experience (G.H.B.). The coefficient of variation of HVPG measurement at both institutions was 7%.

Measurement of HVAT.

All enrolled patients underwent measurement of HVAT immediately after the measurement of HVPG on the same day. All CEUS procedures were performed by an examiner (derivation set by M.Y.K. and validation set by K.T.S.) at each institute using a 2.5-μm (range, 2-8) second-generation sulfur hexafluoride microbubble-based contrast agent (SonoVue; Bracco S.P.A., Milan, Italy); the operators were blind to the HVPG results. After conventional B-mode and color Doppler US examinations, the sonographer placed the 3.5-MHz convex probe (Prosound α10; Aloka, Tokyo, Japan) in an appropriate intercostal position to delineate the right or middle HV. A real-time, dual-frame, contrast-enhanced sonographic technique was applied with a low mechanical index (0.1). Baseline signals from the HV were first recorded for 10 seconds. A 2.4-mL bolus injection of SonoVue was then administered for 1 second and was immediately followed by a rapid flush of normal saline (5 mL) through a three-way tap for 2 seconds through a 20-gauge IV catheter that had been inserted into the cubital vein at the level of the left antecubital fossa. HV imaging was continued for 60 seconds after completing the saline-flush injection. During the test, patients were asked to hold their breath in end-expiration for 20 seconds from 5 seconds after the injection of contrast agent.

In most cases, the regions of interest (ROIs) were set on the HV 3-5 cm from the inferior vena cava and on the first or second branch of the right or middle HV. Time-intensity curves (TICs) for enhancement intensity of the HV included in the ROIs were elaborated by the scanner software. The baseline intensity value was defined as the highest value reached during the first 10 seconds of the recording (before the SonoVue injection). The time taken for the contrast agent to reach the HV (i.e., HVAT) was then measured from the initiation of the contrast-agent injection (after 10 seconds of the recording). HVAT was defined as the interval between this injection point and the second point on the curve representing the signal intensity that exceeded the baseline intensity by 10% (Fig. 2).15, 16

Figure 2.

HV enhancement with microbubble CEUS and measurement of HVAT. (A and B) US images showing the HV (white arrows) before the contrast injection (A) and the arrival of microbubble contrast agent in the HV after contrast enhancement (B). (C) After a 10-second lead time to estimate baseline value, a 2.4-mL bolus of SonoVue was injected into the left antecubital vein, and the TIC of the signal was recorded from the right HV. HVAT was calculated as the time (in seconds) from injection to a sustained increase in signal in the TIC to more than 10% above baseline. The recorded TIC profile shows early HVAT (11.0 seconds; the 10-second lead time was subtracted from 21.0 seconds) in a patient with cirrhosis with HVPG of 20 mmHg. (D) The recorded TIC profile shows an HVAT of 27.0 seconds (37.0 minus 10 seconds) in a healthy control.

Statistical Analysis.

Quantitative variables are expressed as mean ± SD values, and qualitative data are expressed as percentages. The independent t test or analysis of variance was applied for comparisons of normally distributed variables. For nonnormally distributed data, Kruskal-Wallis' test or Wilcoxon's rank-sum (Mann-Whitney's) test was used to analyze the statistical significance of intergroup differences. Pearson's correlation for normally distributed variables and Spearman's rank-correlation coefficient for non-normally distributed data were used, as appropriate. Linear regression analyses were calculated according to the least-squares methods. To assess the performance of HVAT in predicting CSPH, receiver operating characteristic (ROC) curves with the area under the ROC curve (AUROC) were calculated, and optimal HVAT cut-off values were selected on the basis of sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and positive and negative likelihood ratio (+LR and –LR, respectively) in the derivation set. These values were also tested in the validation set, and comparisons of the AUROC values were made using the method suggested by Hanley and McNeil.21, 22 The coefficient of variation (calculated by dividing the SD by the mean and multiplying by 100) for HVAT measurement was estimated to evaluate intraobserver reproducibility. For interobserver variability, a kappa value was also estimated. The level of statistical significance was set at P < 0.05. Statistical analysis was performed using the Statistical Package for the Social Sciences software package (version 13.0 for Windows; SPSS, Inc., Chicago, IL).

Results

The general characteristics of the phase I and phase II study populations are summarized in Table 1. The general characteristics did not differ between the derivation and validation sets, including the proportion of patients with CSPH (i.e., HVPG ≥10 mmHg; n = 41 [57.7%] and n = 20 [57.1%], respectively; P > 0.05). The etiology of cirrhosis was predominantly alcohol intake (derivation set: n = 51 [71.8%]; validation cohort: n = 25 [71.4%]).

Table 1. General Characteristics
 Phase IPhase II
 Total Patients (n = 116)Derivation Set (n = 71, compensated)Decompensated Cirrhosis (n = 45)Validation Set (n = 35)
  1. Abbreviations: HCV, hepatitis C virus; AST, aspartate aminotransferase; ALT, alanine aminotransferase; INR, international normalized ratio.

Age, years (range)/sex (M:F)49.4 ± 8.6 (29-69)/95:2149.2 ± 9.2 (29-69)/59:1249.8 ± 7.6 (30-66)/36:951.3 ± 6.4 (40-66)/30:5
Etiology (%)    
 Alcohol85 (73.3)51 (71.8)34 (75.6)25 (71.4)
 HBV28 (24.1)18 (25.4)10 (22.2)9 (25.7)
 HCV2 (1.7)1 (1.4)1 (2.2)1 (2.9)
 Cryptogenic1 (0.9)1 (1.4)0 (0)0 (0)
Albumin (g/dL)3.5 ± 0.5 (2.2-4.7)3.6 ± 0.5 (2.4-4.7)3.3 ± 0.4 (2.2-4.0)3.3 ± 0.4 (2.5-4.0)
AST (U/L)35.5 ± 11.3 (8-59)33.8 ± 12.7 (8-57)38.2 ± 10.6 (16-59)34.5 ± 11.2 (11-54)
ALT (U/L)34.2 ± 13.3 (9-63)31.5 ± 13.4 (9-56)36.7 ± 12.8 (11-63)33.6 ± 13.1 (9-61)
Total bilirubin (mg/dL)1.5 ± 1.3 (0.2-4.7)1.2 ± 1.0 (0.2-3.9)1.9 ± 1.5 (0.3-4.7)1.2 ± 1.0 (0.3-4.1)
Prothrombin time (INR)1.2 ± 0.2 (0.9-1.8)1.1 ± 0.1 (0.9-1.4)1.3 ± 0.2 (0.9-1.8)1.2 ± 0.2 (0.9-1.5)
Platelet count (/mm3)102.5 ± 50.4 (37.3-174.2)106.4 ± 50.8 (65.7-174.2)98.9 ± 50.2 (37.3-147.7)105.3 ± 49.5 (71.1-165.6)
Child-Pugh score6.4 ± 1.5 (5-12)5.9 ± 1.0 (5-8)7.2 ± 1.7 (5-12)6.5 ± 1.0 (5-9)
 Class A (%)77 (66.4)56 (78.9)21 (46.7)21 (60.0)
 Class B (%)34 (29.3)15 (21.1)19 (42.2)14 (40.0)
 Class C (%)5 (4.3)0 (0)5 (11.1)0 (0)
Esophageal varices (%)    
 None21 (18.1)17 (23.9)4 (8.9)10 (28.6)
 Small (<5 mm)43 (37.1)35 (49.3)8 (17.8)15 (42.9)
 Large (≥5 mm)52 (44.8)19 (26.833 (73.3)10 (28.5)
HVPG (mmHg)13.5 ± 5.6 (2-27)11.4 ± 5.0 (2-23)16.5 ± 4.6 (6-27)11.3 ± 5.4 (4-21)
HVPG ≥10 mmHg (%)84 (72.4)41 (57.7)43 (95.6)20 (57.1)
HVAT (seconds)13.3 ± 3.2 (8.2-24.2)14.1 ± 3.4 (8.4-24.2)12.1 ± 2.3 (8.2-17.3)13.9 ± 2.9 (8.7-20.2)

Comparison of HVAT Data Between Patients With Cirrhosis and Healthy Controls.

The HVAT was significantly longer for the healthy control group (30.5 ± 3.3 seconds; range, 25.1-37.1) than for the population with cirrhosis from the phase I study (13.3 ± 3.2 seconds; range, 8.2-24.2; P < 0.001; Fig. 3).

Figure 3.

Comparison of HVATs between a healthy control group and patients with cirrhosis. HVAT is significantly shorter among the patients with cirrhosis, compared to the healthy controls (30.5 ± 3.3 versus 13.3 ± 3.2 seconds; P < 0.001). Horizontal lines indicate mean and SD values.

Relationship Between HVPG and HVAT in the Phase I Study Population.

There was a statistically significant negative correlation between HVPG and HVAT in the total patient population of the phase I study (linear regression analysis: r2 = 0.558; P < 0.001). A subgroup analysis revealed that compensated patients with cirrhosis (i.e., the derivation set) had mean HVPG and HVAT values of 11.4 ± 5.0 mmHg (range, 2-23) and 14.1 ± 3.4 seconds (range, 8.4-24.2), respectively; there was a statistically significant negative correlation between HVPG and HVAT (r2 = 0.545; P < 0.001). In the decompensated group, mean HVPG and HVAT values were 16.5 ± 4.6 mmHg (range, 6-27) and 12.1 ± 2.3 seconds (range, 8.2-17.3), respectively. Although there was also a statistically significant negative correlation between these values (linear regression analysis: r2 = 0.464; P < 0.001), it was weaker than in the compensated group (Fig. 4). The derivation set was composed mainly of alcohol (n = 85) and hepatitis B virus (HBV)-related (n = 28) cirrhosis, and HVAT did not differ significantly between these etiologies (P = 0.325).

Figure 4.

Relationship between HVPG and HVAT in the phase I study. HVAT was significantly linearly correlated with HVPG in the total cirrhosis group (r2 = 0.558; P < 0.001) (A), in the patients with compensated cirrhosis (r2 = 0.545; P < 0.001) (B), and in the patients with decompensated cirrhosis (r2 = 0.464; P < 0.001) (C).

Noninvasive Prediction of CSPH (HVPG ≥10 mmHg) in Patients With Compensated Cirrhosis: The Derivation Set.

In the noninvasive prediction of CSPH using the HVAT, the AUROC was 0.973 (95% confidence interval [95% CI]: 0.944-0.997). Different cut-off values for HVAT were determined based on the ROC curve. An HVAT value of <14 seconds resulted in Se, Sp, PPV, NPV, +LR, and –LR values of 92.7% (95% CI: 80.6-97.5), 86.7% (95% CI: 70.3-94.7), 90.5% (95% CI: 83.7-97.3), 89.7% (95% CI: 82.6-96.7), 6.95 (95% CI: 2.78-17.38), and 0.08 (95% CI: 0.03-0.25), respectively, for the prediction of CSPH. Table 2 summarizes the best results for PPV, NPV, +LR, and –LR at different HVAT cut-off values. Only 3 (10.3%) of 29 patients with an HVAT value of ≥14 seconds had CSPH (Fig. 5).

Figure 5.

(A) ROC curve showing the prediction of CSPH (i.e., HVPG ≥10 mmHg) using HVAT in the derivation set. (B) Relationship between HVPG (≥10 mmHg) and HVAT. The optimal HVAT cutoff is indicated with a line.

Table 2. Diagnostic Accuracy of HVAT for CSPH (HVPG ≥10 mmHg)
Accuracy (seconds)Se (%)Sp (%)PPV (%)NPV (%)+LR−LRAccuracy
HVAT ≤1363.4 (48.1-76.4)100 (88.7-100.0)100 (100-100)66.7 (82.6-96.7)0.37 (0.25-0.55)0.80 (70.7-89.3)
HVAT ≤1492.7 (80.6-97.5)86.7 (70.3-94.7)90.5 (83.7-97.3)89.7 (82.6-96.7)6.95 (2.78-17.38)0.08 (0.03-0.25)0.90 (83.0-97.0)
HVAT ≤1597.6 (87.4-99.6)73.3 (55.6-85.8)83.3 (74.7-92.0)95.7 (90.9-100.0)3.66 (2.02-6.64)0.03 (0.01-0.23)0.87 (79.2-94.8)

Relationship Between HVAT, Child-Pugh Score, and Esophageal Varices in Patients With Cirrhosis: The Derivation Set.

There was also a significant correlation between HVAT and Child-Pugh score (r2 = 0.162; P < 0.001). Similarly, HVAT was significantly shorter in patients with large esophageal varices than in those with small esophageal varices (10.8 ± 2.0 versus 14.3 ± 3.5 seconds; P = 0.018).

Relationship Between HVPG and HVAT in Patients With Cirrhosis: The Validation Set.

A significant negative correlation was found between HVAT and HVPG in the validation study (linear regression analysis: r2 = 0.538; P < 0.001). With regard to the prediction of CSPH, an HVAT value of <14 seconds resulted in an AUROC of 0.953 as well as Se, Sp, PPV, NPV, and +LR and –LR values of 90.0%, 86.7%, 90.0%, 86.7%, 6.77, and 0.12, respectively (Fig. 6).

Figure 6.

(A) ROC curve showing the prediction of CSPH using HVAT in the validation set. (B) Relationship between HVPG (≥10 mmHg) and HVAT. The optimal HVAT cutoff is indicated with a line.

Reproducibility and Safety of HVAT Measurements.

Reproducibility was calculated for the entire HVAT measurement process, including the performance of CEUS and the drawing and interpretation of the TIC. Furthermore, reproducibility of the drawing/interpretation of TICs was calculated separately because this step is crucial in the measurement of HVAT. As for intraobserver variability of the entire HVAT measurement process, we observed day-to-day variability for 5 consecutive days in 10 healthy control subjects at each institute; the coefficients of variation (CVs) were 3.7% and 3.9%, respectively. Regarding intraobserver variability of the drawing and interpretation of the TIC, each observer (M.Y.K. and K.T.S.) independently repeated this process in 20 randomly selected patients from their own study-population set (day-to-day variability was performed on 5 consecutive days) using recorded CEUS videos; the CVs were 2.7% and 3.2%, respectively. Interobserver variability between the two observers (M.Y.K. and K.T.S.) of the drawing and interpretation of the TIC, expressed as a kappa value, was analyzed in all subjects (a total of 106 patients: 71 in the derivation set and 35 in the validation set) using the recorded CEUS video by confirming it as positive or negative based on an HVAT value of 14 seconds as the cutoff. The kappa value was calculated to be 0.87, which indicates excellent reproducibility and concordance between the two observers. The SonoVue injections were well tolerated by all patients, and no adverse events were noted.

Discussion

There exists a need for a reliable, noninvasive method of assessing the severity of PH in cirrhosis; in particular, any method that could be a suitable alternative for the current invasive gold standard for assessing PH (i.e., measurement of HVPG) would be highly desirable.10, 17 CEUS, using a microbubble contrast agent, has been widely used to improve the detection and specificity of liver tumors, because some lesions can be characterized by their enhancement patterns.23 Microbubble contrast agents increase the signal intensity from the blood after their injection, which improves poor Doppler US examinations. In addition, the recently developed methodology of CEUS has expanded the potential of US hemodynamic studies. Several previous studies have suggested that the severity of hepatic fibrosis in patients with chronic liver diseases is strongly correlated with early enhancement of HV using a microbubble contrast agent with spectral Doppler analysis.15, 16 Earlier arrival of the microbubble contrast agent at the HV in patients with cirrhosis is considered to be secondary to intrahepatic hemodynamic changes, such as arteriovenous shunting and arterialization of capillary beds in the liver.14-16, 24-26 Basically, the formation of intrahepatic shunts contributes to the high portal pressure arising from cirrhosis. Hence, we assumed that early enhancement of the HVs reflects the grade of PH as well as the stage of hepatic fibrosis.

In the present study, we evaluated HVAT as a method of assessing the severity of PH, corresponding to the arrival time of microbubbles at the HV after peripheral venous injection, as calculated using a software program displaying the TICs for signal-intensity enhancement. As expected, HVAT was significantly and negatively correlated with HVPG in the total population with cirrhosis (r2 = 0.558; P < 0.001). The present results have demonstrated that CEUS is an accurate method for determining the severity of PH with a reduced HVAT corresponding to an increasing severity of PH, a finding attributable to the abundant formation of intrahepatic shunts that develop in the setting of high portal pressure. An HVAT value of <14 seconds can be used to discriminate CSPH with good values of the AUROC, Se, Sp, PPV, MPV, and +LR and –LR. The presence of CSPH could be predicted with a high PPV of 90.5% and its exclusion could be identified with a high NPV of 89.7% for an HVAT value of <14 seconds. Based on this cutoff, 42 patients (59.2%) from the derivation set and 20 (57.1%) from the validation set might avoid HVPG.

There is currently just one report in the literature of a similar result regarding the relationship between PH and HA-HV interval time; however, the population in that study was too small and heterogeneous to enable the drawing of firm conclusions. The researchers measured all hepatic vessels simultaneously in one cross-sectional scan, and so it was difficult to estimate the TIC accurately; this method has many limitations in practice.14 In contrast, the present study involved a homogeneous population, and a delicate analysis was performed through subgroup analysis. In addition, the methodology used in this study was relatively simple to apply repeatedly at the bedside and showed a high degree of reproducibility.

The correlation between PH and HVAT was stronger for the compensated group than for the decompensated group in the present study. Although there are several possible reasons for this, however, the difference in composition of HA-HV and PV-HV shunt flow may be the main cause of this phenomenon. Sugimoto et al. demonstrated that intrahepatic (not extrahepatic) shunts cause an earlier HVAT.25 Intrahepatic shunts consist of two components: HA-HV and PV-HV shunts.27 After the injection of the contrast agent, flow through the HA-HV shunt arrives earlier than through the PV-HV shunt and induces early-phase TIC elevation. The PV-HV shunt flow then follows because it takes longer for the PV flow to circulate through the splanchnic vessels, resulting in a late-phase TIC elevation. In decompensated cirrhosis, PH is mainly produced and maintained by the increase of portal blood flow (PBF). PBF is influenced by both the systemic vascular resistance and increased shunt flow or cardiac output.28 Hence, in decompensated cirrhosis, HVAT, which reflects intrahepatic hemodynamic changes, is not as fast as in compensated cirrhosis and has a relatively weak correlation with PH. In addition, with increased PBF, the contrast-enhancement intensity is dependent mainly on PV-HV shunt flow, rather than on HA-HV shunt flow; thus, the shortening of HVAT is weaker in decompensated cirrhosis than in compensated cirrhosis. In contrast, as portal pressure increases, occlusive hyperemia of the PV system can reduce the PBF and perfusion, and this change leads to a compensatory increase in HA-HV shunt flow.29, 30 Therefore, in decompensated cirrhosis, two opposing intrahemodynamic conditions can develop, leading to less-consistent HVAT results.

In addition to the onset of PH, HVAT reflects the deterioration of hepatic function, which is caused by impairment of liver-tissue oxygenation and hepatic perfusion secondary to sinusoidal capillarization and intrahepatic shunting. HVAT is therefore correlated with hepatic function, as expressed by the Child-Pugh score and the Model for End-Stage Liver Disease score.31-33 In addition, a close correlation with clinical manifestations of PH (i.e., the grade of esophageal varices) was also observed. However, this correlation was relatively weak and can be attributed to the extrahepatic component of PH, which can influence the formation of esophageal varices.

The first-generation contrast agent, Levovist (Schering, Berlin, Germany), is taken up and mostly destroyed in the parenchyma of the liver. This results in a short half-life and hence a low clinical usefulness.17, 24 The second-generation microbubble agent, SonoVue, is more stable, stays mainly within the liver vasculature, and allows better real-time contrast enhancement. Thus, HVAT calculated in cirrhosis is shorter when using SonoVue than when using Levovist. SonoVue therefore facilitates both the differentiation of the severity of PH and hepatic fibrosis staging.17, 34

In the present study, derivation and validation set findings were obtained at two independent centers. The validation set was screened according to the same inclusion and exclusion criteria used for the derivation set. There was a very strong agreement between the results for the derivation and validation sets. In addition, through this study, HVAT measurement with CEUS showed no contrast-agent–related adverse effect in terms of safety as well as accuracy, ease of performance, and the possibility as a potential noninvasive technique to assess the severity of PH.

The measurement of HVAT does have some limitations, and many issues need to be resolved before it can be used clinically. The failure rate of HVAT was approximately 11.5%; HVAT could not be measured in cases of a poor echo window related to obesity, poor respiratory cooperation, or severe atrophic changes of the right lobe of the liver in decompensated cirrhosis. Moreover, the kind of software or US equipment used can influence HVAT, to some degree. These issues must be resolved, and more large-scale studies must be conducted in various clinical conditions before the measurement of HVAT can be used conventionally for the prediction of CSPH. Furthermore, technical training of the operators will be needed to avoid potential human errors in the use of this technique. Fortunately, some researchers and companies are trying to develop software that can be applied clinically with ease, and with a high degree of reproducibility, irrespective of the US equipment. Therefore, we expect that many similar studies and trials will follow this pioneering study, the results from which should increase the clinical application of HVAT.

In spite of the aforementioned limitations, HVAT shares a strong histological and pathophysiologic base with the progression of fibrosis and cirrhosis and shows promising clinical significance. However, HVPG remains the most reliable surrogate marker for PH, and so should still be chosen as the standard method for evaluating PH. HVPG is an independent prognostic marker for detecting the clinical progression from compensated cirrhosis to decompensation in patients,2 and as shown in the present study, a large portion of patients with compensated cirrhosis have CSPH. Thus, the estimation of PH using HVPG could be more essential in compensated cirrhosis. Our suggestion is that HVAT measurement can be a supplementary or screening tool when HVPG cannot be done or before the decision to perform HVPG, especially in cases of compensated cirrhosis; clinicians frequently face situations in which they must decide whether or not to measure HVPG in cases of compensated cirrhosis. However, the well-known limitations of HVPG, such as its invasiveness and requirement for a particular skill set, make us hesitate; HVAT can be helpful in decision making in situations such as these.

In conclusion, the benefit of less invasiveness and the strong correlation with HVPG makes HVAT a potential alternative for HVPG.

Ancillary