Correspondence Dr Soon Koo Baik, Department of Internal Medicine, Yonsei University, Wonju College of Medicine, 162, Ilsan-dong, Wonju, Korea Tel: +82 33 741 1223 Fax: +82 33 741 1228 e-mail: firstname.lastname@example.org
Doppler ultrasonography (US) has an advantage of being non-invasive; therefore, several attempts have been made to investigate the haemodynamic alterations in cirrhosis and the response to medical treatment of portal hypertension. Doppler indices, which have been commonly used for the evaluation of portal hypertension, include the measurement of portal and splenic venous blood velocity and flows, and the resistive and pulsatility index at hepatic, splenic, renal, superior mesenteric artery. Although many positive evidences have been suggested, its clinical usefulness in portal hypertension remains unsettled because of being plagued by lack of reproducibility and accuracy characterized by intra- and interobserver variation. However, recently, Doppler's usefulness in assessment of severity of portal hypertension in terms of reproducibility, technical ease and accuracy and response to drugs that reduce the portal pressure has been proposed. In addition, because most of the patients with cirrhosis and portal hypertension have intrahepatic shunts, they show a decrease in intrahepatic circulatory time (IHCT). Doppler US using microbubble contrast agents allows measurement of IHCT. Therefore, application of contrast-enhanced Doppler US can be prospective for the assessment of the severity of portal hypertension. Several reports have demonstrated that colour Doppler endoscopic US enable haemodynamic study to assess the portal hypertension and has a role of guidance to measure the imaging-based variceal pressure. We have reviewed briefly the clinical usefulness of Doppler US in assessing the severity of portal hypertension and its response to treatment.
Portal hypertension leads to serious complications such as variceal bleeding and is responsible for significant morbidity and mortality in patients with cirrhosis (1–5). Precise assessment of the severity of portal hypertension will be very useful in the management of patients with portal hypertension and cirrhosis. Even though measurement of the hepatic venous pressure gradient (HVPG) has been accepted as the gold standard for assessing the degree of portal hypertension, because of its invasiveness, it is not suitable for widespread routine clinical use (6–8). Doppler ultrasonography (US) allows us to examine haemodynamics of abdominal vessels including the hepatic and portal system. Thus, many investigators have attempted to confirm the usefulness of Doppler US in assessing portal hypertension in patients with cirrhosis. In particular, it would be highly desirable to have any Doppler parameter be a suitable substitute for the invasive current gold standard of measuring HVPG for assessing portal hypertension (7–10). However, previous reports on the usefulness of Doppler US for assessing portal hypertension showed conflicting results in patients with cirrhosis (8–15). Recently, favourable results of Doppler US examination focusing on hepatic vein (HV) have been demonstrated (7, 16). Furthermore, there is a high hope of utilizing Doppler US using microbubble contrast agents to assess the severity of portal hypertension as well as to diagnose cirrhosis (17, 18).
The purpose of this article is to review the clinical usefulness of haemodynamic evaluation by Doppler US focusing on novel and conventional techniques through recently published data for Doppler US in portal hypertension.
Measurements of Doppler ultrasound indices for portal haemodynamics
Blood velocity and flow
Doppler examinations allow the measurement of blood velocity and flow in vessel. This method is simple and confers technical ease that its clinical application in portal hypertension has been attempted. During the measurement of velocity, the angle between the Doppler beam and the long axis of vessel should be <60° for accuracy (19–21). Applications of measuring blood velocity and flow are almost always possible at portal and splenic veins; however at the artery, it is impossible, except at the superior mesenteric artery. To obtain portal vein velocity (PVV) and flow, the portal vein is imaged longitudinally in the supine position, and the Doppler sample volume is set at its crossing point with the hepatic artery. When the sample point is adjusted to the centre of the portal vein, the PVV is recorded in a suspended expiration and is averaged over a few seconds (Fig. 1). Portal venous flow is determined by the formula, cross-sectional area × mean velocity × 60 (19–22). The mean PVV in cirrhotic patients is relatively low compared with that in healthy subjects because of increased intrahepatic vascular resistance (outflow resistance). Zironi and colleagues reported that the mean velocity of portal vein in cirrhosis and normal subjects were 13.0±3.2 vs. 19.6±2.6 cm/s respectively. The cut-off value of 15 cm/s showed a sensitivity and specificity of 88 and 96%, respectively (23). However, as portal hypertensive patients with cirrhosis have various portosystemic shunts, some show a high level of PVV similar to normal subjects. Indeed, we previously found that there was no significant correlation between PVV and HVPG (8). In other words, portal blood velocity and flow may differ between patients with similar portal pressures because of significant variability in portosystemic collateral patterns. This notion is supported by the study of Merkel and colleagues, who examined the correlation between HVPG and portal venous flow and velocity in 39 cirrhotic patients, in which no significant correlation was found (9). Variabilities in PVV measurement include equipment-related, intra- and interobserver variance. Acceptable levels, <8–10%, of intra- and interobserver variability have been reported previously (24–28). A co-operative training programme may reduce the intra- and interobserver variability in PVV measurement (29).
The velocity and cross-sectional area of splenic vein are measured at the splenic hilum.
Both splenic venous velocity and flow increase in portal hypertensive patients associated with dilated splenic vein and enlarged spleen (Fig. 2). It has been reported that the splenic venous flow exceeding portal venous flow is related to the formation of portosystemic varices and a high risk of variceal bleeding (30).
Measurement of blood velocity and flow by Doppler US is useful for the evaluation of patency of stent after transjugular intrahepatic shunt (TIPS). Patent shunts are characterized by stent velocity in excess of 70 cm/s and hepatofugal flow in portal circulation distal to shunt with the presence of cardiac pulsatility (31, 32). It is known that stent velocity between 50 and 60 cm/s is all that is required to diagnose shunt stenosis (33–36).
Resistance by measuring resistive and pulsatility index
Regardless of the incidence angle, the resistances in the hepatic, splenic and renal artery can be evaluated by measuring the resistive index (RI) and pulsatility index (PI) if the vessel is identified by colour Doppler (8, 11, 14, 37). For measuring RI and PI of the hepatic artery, under the right intercostal scanning of the liver, the branch of the hepatic artery around the portal hilus is identified using colour Doppler. After the Doppler sample volume is located in the branch of the hepatic artery, the time–velocity wave of the Doppler signal is recorded (Fig. 3). The peak systolic velocity, the end diastolic velocity and the mean velocity are measured. From these measurements, the hepatic RI [(peak systolic velocity−end diastolic velocity)/peak systolic velocity] and the hepatic PI [(peak systolic velocity−end diastolic velocity)/mean velocity] are determined (8, 14, 37) (Fig. 4). With higher arterial resistance, there is increase in the peak systolic velocity and a decrease in the peak diastolic velocity. Therefore, RI and PI increase with higher arterial resistance (38–40). PI is different from RI in that it uses mean velocity as its denominator instead of the peak velocity like RI. PI is superior to RI when arterial resistance is extremely high that the end diastolic velocity is close to 0 (8). Colour Doppler allows the identification of the main branches of the splenic artery at the splenic hilus. The time–velocity wave is recorded after the Doppler sample volume is placed inside these vessels and the RI and PI are determined using the same method used for the hepatic artery (11, 19). Similarly, the RI and PI of the renal artery are determined at the interlobar artery of the kidney (5, 38). It has been reported that the splenic RI and the hepatic PI increase parallel to the increase in the HVPG, that is, a Doppler ultrasonic determination of these indices may contribute to a non-invasive evaluation of portal hypertension (11, 37). However, other studies have reported that the RI of the hepatic artery did not correlate with portal hypertension (8, 13, 14). Even though the kidney is an extrahepatic organ, measurement of RI and PI in kidney can be useful in the diagnosis of portal hypertension and cirrhosis. Renal RI and PI are increased in patients with portal hypertension and cirrhosis, specifically those in advanced states, because renal vasoconstriction is modulated by a decrease in effective circulatory volume and increase in sympathetic tone in cirrhosis (Fig. 5) (5, 38–40).
In terms of measuring the RI and PI, which is known to have an advantage in measuring the vascular resistance regardless of the incidence angle, acquiring the same arterial branch by colour Doppler in each patient is difficult. Therefore, it is difficult to evaluate the RI and PI under the same conditions for the enrolled patients. Hence, the accuracy and reproducibility of arterial RI and PIs have been questioned (8, 41).
Hepatic vein waveform analysis
The Doppler HV waveform in healthy subjects is triphasic (two negative waves and one positive), and this pattern is the consequence of variations in the central venous pressure because of the cardiac cycle. In patients with cirrhosis, the presence of abnormal biphasic or monophasic HV waveforms has been incontrovertibly demonstrated by a number of studies (7, 42–44). In addition, previous work has shown that the monophasic waveform is correlated with higher Child–Pugh scores and decreased survival rates (45). For Doppler HV examination, HV can be easily visualized along its longitudinal axis by colour flow mapping at the supine position. The flow in HV displays the blue colour in colour flow mapping because it is away from the ultrasonic probe. Thereafter, Doppler shift signals are obtained from the hepatic vein at a distance of 3–6 cm from the junction of the vein with the inferior vena cava (Fig. 6). To determine whether HV waveform analysis might be useful in the assessment of portal hypertension, Baik et al. (7) prospectively examined the relationship between waveforms and the severity of portal hypertension measured by HVPG in 78 cirrhotic patients who experienced variceal bleeding. A correlation was found between abnormalities in HV waveforms and HVPG, i.e. with increasing HVPG, the HV waveform tended to flatten. Furthermore, the monophasic waveform was associated with severe portal hypertension (HVPG>15 mmHg) with relatively high sensitivity and specificity in that study population. Hence, flattening of the HV waveform observed in the cirrhotic patients indicates a high likelihood of severe portal hypertension. In addition, the change in the HV waveform following vasoactive agent administration, which reduces the portal pressure, also significantly correlated with that of HVPG.
Although the above HV waveform analysis is useful, evaluation with lack of a quantitative value reduces the clinical value in the assessment of portal hypertension and in response to drug treatment. In this view, assessment of damping index (DI) allows the quantification of the extent of the abnormal HV waveform (loss of pulsatility).
Kim and colleagues prospectively evaluated the correlation between the extent of abnormal Doppler HV waveforms, expressed as DI, and the HVPG, and response to propranolol in patients with cirrhosis. DI is calculated by dividing the minimum velocity over the maximum velocity of the HV waveform (Fig. 7) (16). Abnormal HV waveforms were seen in 66 out of 76 patients (86.8%). DI significantly correlated with the grade of HVPG, i.e. with higher HVPG, an increase in DI was observed (P<0.01). By logistic regression analysis, DI>0.6 was significantly more likely to be severe portal hypertension (odds ratio: 14.19, 95% CI: 4.07–49.55). The ROC curve according to the value of 0.6 of DI showed a sensitivity of 75.9% and a specificity of 81.8% for the presence of severe portal hypertension. The positive and negative predictive values according to a DI of 0.6 were 91.1 and 58.1% respectively.
Regarding the evaluation of response to drug, change of DI following propranolol treatment also significantly correlated with that of HVPG (P<0.01) (Fig. 8). In the responder group, which showed a decrease in HVPG of more than 20% compared with baseline or a decrease in value to below 12 mmHg after propranolol treatment, the mean decreased value of DI was 0.23, corresponding to a 33.6% reduction from baseline. Hence, these results suggest that the evaluation of Doppler HV waveform could be a valuable supplementary tool to assess the therapeutic response to vasoactive drugs used to treat portal hypertension, when HVPG measurement is unfeasible or unavailable.
The exact cause of these changes in the Doppler hepatic vein waveform remains unclear. Some investigators have suggested that the hepatic vein wall is thin and surrounded by liver parenchyma, and so its compliance can be easily reduced by parenchymal fibrosis and fat infiltration (44, 45). However, our vasoactive agent-induced improvement in the waveforms suggests that a haemodynamic effect of high portal pressure, rather than a fixed structural abnormality, is the pathogenic mechanism responsible for the abnormal waveforms (7, 16). It is supported by a recent preliminary study that reports that abnormal HV waveform in cirrhotic patients is associated with earlier hepatic vein transit time using microbubble contrast agents, which means the presence of intrahepatic shunts (17). In other words, flattening of the HV wave can be attributed to an increase in HV inflow from intrahepatic shunts implicated in portal hypertension, which results in haemodynamically blunting the effect of variations in central venous pressure during the cardiac cycle, rather than lack of liver compliance. Alterations of haemodynamic parameters in portal hypertension with cirrhosis found on Doppler US are summarized in Table 1.
Table 1. Alterations of haemodynamic parameter in cirrhosis found on Doppler ultrasound
Cirrhosis with PH (compared with normal)
Accuracy and reproducibility
CV, coefficient of variation (calculated by dividing the standard deviation by the mean and multiplying by 100); DI, damping index; HA, hepatic artery; HV, hepatic vein; PH, portal hypertension; PVV, portal venous velocity; PVF, portal venous flow; RA, renal artery; SA, splenic artey; SMA, superior mesenteric artery; SPI, splenoportal index; SVF, splenic venous flow; SVV, splenic venous velocity; –, never been reported.
PVV <15 cm/s is associated with a sensitivity and a specificity of 88 and 96% for PH
SPI threshold of 3.0 predict presence of EV in 92% of patients
HA resistance (RI, PI)
Increased or no change
SA resistance (RI, PI)
Increased or no change
RA resistance (RI, PI)
Renal PI>1.14 is associated with poor prognosis. Higher than normal renal RI and PI have a high PPV (84–100%) for detection of severe PH.
SMA resistance (RI, PI)
With liver dysfunction and cirrhosis progress, SMA resistance decrease while SMA flow increase.
Monophasic wave form is associated with severe PH, with a sensitivity of 74% and a specificity of 95%.
DI of HV
DI>0.6 predict severe PH (HVPG>12 mmHg) with a PPV of 91%
Colour Doppler endoscopic ultrasound
Colour Doppler endoscopic ultrasound (CD EUS) can provide significant information regarding haemodynamics as well as morphological change in varix. Morphological and haemodynamic changes of the azygos vein and the left gastric vein occur in patients with portal hypertension. Haemodynamic study and visualization of the azygos vein and the left gastric vein can be performed well with CD EUS to assess portal hypertension. It has been suggested that a higher hepatofugal flow of the left gastric vein with CD EUS is associated with the development of oesophageal varix (46). Maximal blood velocity of the azygos vein is increased in patients with portal hypertension. Azygos vein flow has been found to be four to six times higher in patients with portal hypertension and cirrhosis than in normal subjects and is directly related to pressure in the portal system. CD EUS is also useful in assessing azygos blood flow and in monitoring the effect of vasoactive agents in portal hypertension (47).
Oesophageal variceal pressure assessed by CD EUS-guided manometry significantly correlated with portal pressure by HVPG in vitro study. Therefore, CD EUS-guided manometry of oesophageal varices for the assessment of portal haemodynamics appears promising; however, further validation studies are required (48). Hence, CD EUS and imaging-based variceal pressure measurement can be useful as a supplementary tool for the assessment of portal haemodynamics and risk of variceal bleeding.
Future opportunity and challenges: Doppler ultrasound using microbubble contrast agents
Opportunity: intrahepatic circulatory time and hepatic vein transit time using microbubble contrast agents
Several reports have suggested that analysis of intrahepatic circulatory time (IHCT) is useful for the diagnosis of cirrhosis and can assess the severity of chronic liver disease (49–51). IHCT can be calculated from the difference between the arrival time of the microbubbles in the hepatic vein and that in the portal vein or hepatic artery. IHCT and severity of liver disease have an inverse relation, where a decrease in IHCT is accompanied by an increase in the severity of liver disease, which results from the presence of arteriovenous and portovenous shunts in cirrhotic liver. Similarly, measurement of hepatic vein transit time (HVTT) using a quantification software package is also useful in diagnosing cirrhosis, which is calculated as the time from injection to a sustained increase in Doppler signal to more than 10% above baseline. An HVTT value of <21 s has a sensitivity of 100% and a specificity of 80% for the diagnosis of cirrhosis in 78 patients with chronic liver disease (51–53). In this regard, we assume that the HVTT value by Doppler US using microbubble contrast agent may represent the degree of portal pressure, i.e. with increasing HVPG, HVTT decreases (Fig. 9).
In a preliminary study with 41 cirrhotic patients, we found a significant negative correlation between HVTT measured by a Doppler microbubble contrast agent and HVPG (P<0.001, r2=0.543) (Fig. 10). The AUROC for the prediction of severe portal hypertension (HVPG≥12 mmHg) was 0.948, and the sensitivity, specificity, positive predictive value and negative predictive value according to an HVTT cutoff value of 17.0 s were 78.6, 83.3, 91.7 and 62.5% respectively. HVTT also showed significant correlation with the grade of oesophageal varices (F0–F1: 18.0±3.3 s vs. F2–F3: 15.8±2.5 s, P<0.05), Child–Pugh's score and MELD score (P<0.05) (unpublished data). A large study dealing with the relationship between the measurement of HVTT by contrast Doppler US and HVPG is expected to strongly determine the value of HVTT in assessment of severity of portal hypertension. Therefore, microbubble contrast agents may be potentially a supplementary adjunct to Doppler US for the diagnosis of portal hypertension and cirrhosis.
Challenges: measurement of intravascular pressure using microbubble contrast agents
A new experimental method for noninvasive intravascular pressure measurement, based on the disappearance time of free gas bubble, was proposed. By transmitting a low frequency, high acoustic amplitude US burst, encapsulated bubbles ruptured and free gas bubbles were released into the desired region, i.e. portal vein where the local pressure can be measured. Higher value of local intravascular pressure results in a quicker disappearance time of free gas bubbles (54, 55). Microbubble contrast agents are capable of enhancing Doppler US signals. Therefore, changes in ambient pressure affects the reflectivity of contrast microbubbles, leading to a significant correlation between subharmonic signals and hydrostatic pressure (56, 57). This is another candidate of new techniques for noninvasive intravascular pressure measurement using microbubble contrast agents. Although these methods are still in investigative stages, they are nonetheless potential alternatives to Doppler in measuring the severity of liver diseases, especially portal hypertension.
Conventional Doppler US indices including the portal and splenic venous blood velocity and flow, and arterial RI and PI at hepatic, splenic, superior mesenteric and renal arteries do not seem to be sensitive enough for the accurate diagnosis of portal hypertension because of the conflicting results it yields.
There are two important reasons why clinical usefulness of the above conventional Doppler US indices is still questioned.
Firstly, most cirrhotic patients have portosystemic shunts arising from portal hypertension, and the shunt patterns are not unique but vary in complexity in each patient. Therefore, the value of Doppler index may differ even between patients with similar portal pressures because of significant variability in portosystemic collateral patterns.
In contrast, the presence of intrahepatic shunts resulting from portal hypertension can be helpful to diagnose portal hypertension through novel characteristic, such as Doppler US index. It has been recently suggested that the extent of abnormality in Doppler HV waveform is associated with the degree of HVPG, and a change in the HV waveform is closely correlated with that of HVPG. Abnormal flattened HV waveform is thought to be because of haemodynamically blunting the effect of variations in central venous pressure during the cardiac cycle, which arises from increased HV inflow via intrahepatic shunts secondary to portal hypertension. Moreover, under the microbubble contrast enhancement, Doppler US detection of decreased hepatic vein transit time (similar to IHCT) because of the presence of intrahepatic shunts can be useful to assess the severity of portal hypertension.
Secondly, Doppler US has relatively poor reproducibility and accuracy during measurements of Doppler US indices. To reproduce and improve accuracy, cooperative training programmes for operators with strict examination protocols would prove to be helpful in reducing the rate of intra-and inter-observer variation.
Colour Doppler endoscopic ultrasound focusing on haemodynamics and visualization of the azygos vein and the left gastric vein have the potential to be adjunctive methods for assessment of portal haemodynamics and risk of variceal bleeding.
Consequently, developments of novel Doppler US indices and techniques are ongoing in order to overcome the limitations of conventional Doppler US methods in the diagnosis of portal hypertension and some of them have the potential to be clinically useful.
In conclusion, haemodynamic evaluation by Doppler US would be of value as a supplementary tool for the assessment of portal hypertension and response to treatment, specifically when HVPG is not feasible or adequate to potentially allow widespread clinical use.
This work was supported by a grant from the Ministry for Health, Welfare and Family Affairs, Republic of Korea (no.A050021).