Assessment of intrahepatic blood flow by Doppler ultrasonography: Relationship between the hepatic vein, portal vein, hepatic artery and portal pressure measured intraoperatively in patients with portal hypertension

The hepatic vein (HV) is the only draining vessel in the liver, which has two supplying vessels, from the liver sinus to the inferior vena cava (IVC). The thin-walled veins are anechoic under ultrasonography, do not have valves, and can be distinguished from the portal vein. The spectrum of HVs can reflect changes in blood flow through the tricuspid valve during the cardiac cycle, leading to pulsatile changes in the spectrum during ultrasonography. The normal waveform of HVs is triphasic with two hepatofugal phases related to atrial and ventricular diastole, and a short phase of retrograde (hepatopetal) flow caused by the pressure increase in the right atrium at atrial systole. With increased stiffness in the liver parenchyma (especially around the HVs), the hepatic waveform becomes less pulsatile with no retrograde flow, and can eventually lead to a flat waveform [1‐3]. The ability to detect a hepatic parenchymal abnormality by hepatic venous Doppler studies has been discussed by several authors: their opinions vary considerably. Hence, a semi-quantitative parameter DI was introduced and data showed that evaluation of the HV DI might be a useful supplemental method for assessing the therapeutic response to portal hypertensive drugs when portal pressure examination is not available [4].

The exact cause of abnormalities in HVs is controversial. One study suggested that terlipressin-induced improvement in the waveforms is evidence that a hemodynamic effect of high portal pressure rather than a fixed structural abnormality is the pathogenic mechanism responsible [5]. In the study, we evaluated the relationship between changes in intrahepatic blood flow and portal pressure (PP) measured intraoperatively in patients with portal hypertension (PHT), aiming to discuss the indicative value of HV waveform and its quantitative index DI firstly, then the cause of changes in HV waveforms both from the histological and hemodynamic angles.

The mean PP measured directly from the right gastroepiploic vein was 30.02 ± 3.81 mmHg. The main clinical and pathological data for the patients at the beginning of the study are presented in Table 1.

The flow pattern was triphasic in 26 (86.7%), biphasic in 3 (10.0%), and monophasic in 1 (3.3%) of control subjects. Among 60 patients, the Doppler flow waveform in the middle hepatic vein was triphasic in 19 (31.6%), biphasic in 28 (46.7%), monophasic in 13 (21.6%) (Figure 1). The distribution of the HV flow pattern in the patient group according to the PP is summarized in Table 2. PP in patients with a monophasic HV waveform was much higher than in those with biphasic and triphasic waveforms (p = 0.004 and p = 0.003, respectively), but no significant difference was found between patients with biphasic and triphasic flow patterns (29.39 ± 3.40 vs. 28.94 ± 3.36, p = 0.673).

Although PVVel was slightly lower in patients with PHT compared with healthy controls (15.08 ± 4.12 cm/s vs.17.14 ± 3.46 cm/s), a significant difference was not observed (p = 0.114), and no significant correlation was observed between PVVel and PP (p = 0.597). However, PVVel in patients with a monophasic pattern (11.82 ± 3.91 cm/s) decreased significantly compared with biphasic and triphasic patterns (15.33 ± 3.63 cm/s, p = 0.007 and 16.96 ± 3.76 cm/s, p < 0.0001) (Figure 2).

On average, the HAPI in the patient group was much higher than that in the healthy group (1.63 ± 0.48 vs. 1.17 ± 0.18, p < 0.0001). The HAPI in patients with monophasic and biphasic waveforms was significantly higher than that in patients with triphasic waveforms (p = 0.001, p = 0.022, respectively) (Figure 3). All the parameters about healthy subjects were summarized in Table 3.

Significant linear correlations with PP were found only in HV waveforms with r = 0.579 (p < 0.0001) and with the HAPI with r = 0.427 (p = 0.001). The correlation according to the changes in HV waveforms was also observed between PVVel and the HV waveform (r = -0.44, p < 0.0001), HAPI and HV waveforms (r = 0.438, p < 0.0001). Bloodflow parameters, HAPI, PVVel and HV-waveform changes showed no significant correlations with Child-Pugh scores. The Child-Pugh score showed a significant correlation with PP (r = 0.589, p = 0.044). This seemed to indicate that the weaker the liver function, the higher is the PP. However, there was no significant difference among the A, B, and C Child-Pugh scores (p = 0.249). HV DI was found significantly correlated with PP, namely, with higher PP, increased DI was observed(r = 0.473, p < 0.0001) (Figure 4). The same correlation (with an increase of DI, the liver function weakened) was also observed between DI and Child-Pugh scores (r = 0.411, p = 0.001).

In PHT patients, 68.3% of the livers had small nodules. PP was statistically different with respect to nodule size (p = 0.038, small/mixed; p = 0.009, small/large). However, changes in the HV waveform were not significantly correlated with this pathological change.

Various studies have been designed for evaluating the severity of liver abnormalities by Doppler ultrasound in patients with chronic liver disease [10‐12]. The pattern of blood flow in the HV is one of the parameters to be evaluated. Three grades of hepatic waveforms have been described by Bolondi et al. [13] to indicate changes from the normal triphasic pattern to the flat pattern widely used in older studies of chronic liver disease [13‐15]. Nowadays, more studies are carried out to determine if analyses of HV waveforms may be useful in the assessment of PHT [9, 16]. In the present study, we suggest that an abnormal HV Doppler curve and quantitative index DI may be non-specific indicators of liver abnormality as well as of PP in PHT patients. Furthermore, by comparing the hemodynamic changes in the portal vein and hepatic artery and histological changes in liver parenchyma, we attempted to discuss the mechanism of abnormal HV waveform.

Even though measurement of the hepatic vein pressure gradient (HVPG) has been accepted as the 'gold standard' for assessing the degree of PHT, it is not suitable for widespread routine use because of its invasiveness [17‐19]. Additionally, in the presence of increased pre-sinusoidal resistance during the cirrhotic process, the portal venous pressure can be higher than the wedged hepatic venous pressure [20]. There still was a lot of studies indicated that a good correlation between the HVPG and portal venous pressure either with a wedge catheter or a balloon catheter measured, especially in patients with alcoholic cirrhosis and hepatitis B virus (HBV) cirrhosis [21]. Considering that limitations of HVPG, and the chance we can get the directly measured portal pressure, the present study compared HV waveforms with PP measured directly in the portal venous system for the first time.

In contrast with other results [15, 22], in the present study, the HV waveforms in the healthy control group also presented three types of flow patterns, whereas the triphasic HV waveform was observed in 88% of subjects. Only one person had a monophasic waveform in the HV. However, this subject had a much higher triglyceride (TG) level (8.56 mmol/L) in a subsequent blood lipid test, which may have been the cause. The two other two subjects with a biphasic HV waveform had a normal lipid profile. In patients with PHT, a monophasic HV waveform meant a higher PP, a lower PVVel, and higher hepatic artery resistance. DI in patients with PHT indicated a high likelihood of higher portal pressure and severe liver function as Child-Pugh score showed.

The exact causes of changes in Doppler HV waveforms remain unclear. Some investigators have suggested that parenchymal fibrosis and fat infiltration surrounding the wall of the HV compress the wall and reduce its compliance [23]. Other authors think that the pathogenic mechanism causing intrahepatic shunts is responsible for the abnormal waveforms [5, 24]. To explore the mechanism of changes in HV waveforms, we assessed the intrahepatic changes from two aspects. Firstly, two hemodynamic parameters, PPVel and HAPI, were also assessed to evaluate the mechanism of changes in HV waveforms. The results showed that the more flattened the HV waveform, the lower the PVVel and the higher the HAPI became. Intrahepatic shunts from the hepatic artery to HVs or from the portal vein to HVs would cause increased inflow in the HVs. However, draining by HVs through vascular shunts could also decrease the vascular resistance of the supplying vessels, consequently leading to increased velocity in the portal vein and decreased resistance in the hepatic artery. However, values of PVVel and HAPI measured in the present study were not in accordance with this theory.

Secondly, with respect to pathology, we attempted to find correlations between architectural distortion in the liver parenchyma and changes in HV waveforms. Nagula et al. showed that small nodule size in the liver is also indicative of greater damage and architectural distortion of liver tissue, and further increases intrahepatic resistance. Patients with small nodules have higher PP [25]. In the present study, the relationship between PP and histological changes was consistent with the findings of Nagula et al., but direct evidence about histological changes being responsible for the changes in HV waveforms was inadequate.

In conclusion, changes in HV waveforms plus the change tendency of DI in patients with PHT can indicate the severity of PHT to a certain degree. Taking all the intrahepatic hemodynamic changes under PHT into account may provide information for investigation of the mechanism of abnormal HV waveforms. From the hemodynamic angle, we speculated that histological stiffness in the liver parenchyma around the HVs was believed to contribute to changes in HV waveforms.

The present study had several limitations. Firstly, we did not assess the effect of cardiac output and systemic vascular resistance on the hyperdynamic syndrome, which could also be a determinant of abnormalities of HV Doppler flow. Secondly, the confirmation of liver disease by liver biopsy was not carried out in healthy subjects. Hence, we could not explain exactly why there were abnormal HV waveforms in control group, we could only attribute the abnormality to abnormal blood lipid profiles. Thirdly, the histological evidence was not sufficient enough to support our hypothesis that abnormal HV waveform was caused by pathological abnormality. If the present results can be confirmed in further studies, indicative value of HV waveform and its cause can be illustrated adequately.

The study highlighted the hemodynamic changes and correlations of liver blood vessels with portal pressure, Child-Pugh scores and liver morphological changes. We put more emphasis on the relationships between HV waveform changes and other parameters, aiming to assess the indicative value of HV waveform and the cause of HV abnormal changes in patients with PHT. Results showed that flattening of HV waveform was accompanied by higher PP, lower PVVel and higher HAPI. DI indicated higher portal pressure and poorer liver function. The correlations between the parameters support the hypothesis that histological changes reducing HV compliance may be the cause of abnormalities of Doppler HV waveforms in a direct and indirect way.