- Email: [email protected]
Earlier hepatic vein transit-time measured by contrast ultrasonography reflects intrahepatic hemodynamic changes accompanying cirrhosis
Journal of Hepatology 37 (2002) 578–583 www.elsevier.com/locate/jhep
Earlier hepatic vein transit-time measured by contrast ultrasonography reflects intrahepatic hemodynamic changes accompanying cirrhosis Hiroyuki Sugimoto*, Tetsuya Kaneko, Masashi Hirota, Ekmel Tezel, Akimasa Nakao Department of Surgery II, Nagoya University School of Medicine, Tsurumai-cho 65, Showa-ku, Nagoya 466-8550, Japan
Background/Aims: Non-invasive diagnosis of cirrhosis by transit-time analysis of an ultrasound contrast agent has been reported, even though the mechanism by which contrast arrives to the hepatic vein earlier in cirrhosis than in normal controls is unknown. The aim of this study is to assess whether the earlier appearance of contrast in the hepatic vein depends on intrahepatic or extrahepatic causes. Methods: There were 15 participants: six volunteers, three patients with hepatitis, and six with cirrhosis. The contrast agent was given intravenously, and transit-time analysis of the hepatic artery, portal vein and hepatic vein was performed. The time-acoustic intensity curves in the three vessels were analyzed by an image and cineloop display and quantification software package. Results: The hepatic artery and portal vein arrival times were not significantly different among the three groups. On the other hand, hepatic vein arrival times were significantly earlier in cirrhosis (median 18 seconds) compared with arrival times in hepatitis patients (median 30 seconds, P , 0:001) and in healthy volunters (median 31 seconds, P , 0:001). These results give support to a previous pilot study and indicate that most of the time delay in hepatic vein arrival time between cirrhosis and the other groups originated from intrahepatic circulation abnormalities. Conclusions: This study confirms that the earlier appearance of contrast in the hepatic vein observed in cirrhosis is due to intrahepatic, and not extrahepatic, hemodynamic changes. q 2002 European Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. Keywords: Ultrasonography; Cirrhosis; Hepatic vein transit-time; Pulse-inversion imaging
1. Introduction In cirrhosis, important changes occur in liver circulation. The increase in intrahepatic vascular resistance decreases the portal fraction of liver perfusion [1,2]. This decrease in portal perfusion is partially compensated by an increase in arterial inflow [1,3,4]. To evaluate cirrhosis, various methods exist for the determination of hepatic perfusion [5–7]. Most of them, however, are invasive or controversial [5]. Recently, microbubble contrast agents have been developed for ultrasonography [8]. Albrecht et al. [9] reported a technique for the diagnosis of hepatic cirrhosis by transit-time analysis of an ultrasound contrast agent. Their cirrhotic patients showed an
Received 28 May 2002; received in revised form 10 July 2002; accepted 17 July 2002 * Corresponding author. Tel.: 181-52-744-2245; fax: 181-52-744-2255. E-mail address: [email protected] (H. Sugimoto).
earlier arrival time of the contrast agent in their hepatic veins (HVs) than normal volunteers. Cirrhosis involves several intrahepatic and extrahepatic hemodynamic changes that contribute to the earlier arrival time such as arterialization of the liver, intrahepatic shunt, pulmonary arteriovenous shunt, or hyperdynamic circulation. However, to what degree such changes contributed to their results can not be assessed from their study. To investigate intrahepatic hemodynamic changes, not only must transit-time be measured in the HV but also in both the hepatic artery (HA) and portal vein (PV). However, the measurement of transit-time in the HA by spectral Doppler methods is very difficult because the sample volume cannot be kept within the small diameter of the HA long enough to complete the measurement. Bang et al. [10] reported that pulse-inversion imaging is simpler and has certain advantages over spectral Doppler quantification in assessing the transit-time of a bolus of an echo-enhancing agent. The major advantage of this method
0168-8278/02/$20.00 q 2002 European Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: S01 68- 8278(02)0026 4-7
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
is that the arrival of the contrast agent can be visually tracked. Thus, this pulse-inversion imaging can elucidate transit-time in both the HA and PV as it does in the HV. We conducted transit-time analysis simultaneously in the PV, HA, and HV using pulse-inversion imaging. The aim of this study is to assess whether the cause of an earlier HV transit-time depends on intrahepatic or extrahepatic causes. 2. Patients and methods 2.1. Patients There were 15 participants: six normal volunteers (four men, two women, median age 41 years, age range 34–69 years) with no history or clinical signs of liver disease; three patients (two men, one woman, median age 75 years, age range 62–76 years) with biopsy-proven non-cirrhotic diffuse liver disease (chronic hepatitis B); and six patients (two men, four women, median age 68 years, age range 65–74 years) with biopsy-proven cirrhosis (hepatitis C). Six patients with cirrhosis were classified as Child–Pugh A (three patients) and Child–Pugh B (three patients). On the basis of pathology in the biopsy, liver cirrhosis was defined as diffuse nodularity with fibrous septa, and chronic hepatitis was defined as the presence of bridging fibrosis, but without complete pseudoacini. Other characteristics of the subjects including histological grading and staging [11] are summarized in Table 1. This study protocol was in accordance with the guidelines of the Declaration of Helsinki and has been reviewed by the ethics committee of the Nagoya University School of Medicine. All patients gave their informed consent to inclusion in the study.
2.2. Transit-time analysis using the pulse-inversion harmonic method All subjects were tested in the morning after an overnight fast. All pulseinversion harmonic studies were performed with an ATL HDI 5000 (Philips, Bothell, WA) using a 2–5 MHz convex probe. All examinations were performed with patients in the supine position. In transit-time analysis using the contrast agent, the umbilical PV, HA, and middle or left HV were simultaneously scanned in a transverse section (Fig. 1). In four patients whose three intrahepatic vessels could not be simultaneously scanned, two injections of contrast agent were given at least 10 min apart to assess the transit-time analysis (one normal volunteer, one patient with noncirrhotic diffuse liver disease, and two patients with cirrhosis). Apparatus
579
settings for pulse-inversion harmonic mode were the following: focus 9 cm, mechanical index 1.5, and trigger mode with 300 ms delay. The microbubble contrast agent used was Levovist (Schering AG, Berlin, Germany). A bolus injection of 2.5 g Levovist 300 mg/ml was given via a 21 G cannula into an antecubital vein followed by 10 ml of saline at 2 ml/s. We asked the subjects to hold their breath, and we then recorded from 0 to 60 s after the contrast agent injection. Subjects who could not hold their breath were to breathe gently and were recorded for 60 s after the injection. The time delay from injection until the first echogenic bubbles of contrast agent could be subsequently seen in the HA, PV and HV was measured, and an analysis of time-acoustic intensity curves was performed using ATL HDI Lab (Philips, Bothell, WA). ATL HDI Lab is an image and cineloop display and quantification software package. Digital images and cineloop acquired by the ATL HDI 5000 can be transferred to a personal computer running ATL HDI Lab. One can trace the region of interest (ROI) on the HA, PV and HV, and then automatically obtain the time-acoustic intensity curves. The arrival time was defined as the time delay from injection until the first echogenic bubbles of contrast agent could be seen. The time to peak was defined as the time interval from the beginning of the injection to the peak of the time-acoustic intensity curve. The rising time was defined as the time interval from the arrival time to the peak of the time-acoustic intensity curve. The rising rate was defined as the average increase of the signal per second between the arrival time and the peak time. All examinations were performed by three of the authors (H.S., M.H., and T.K.) who were blinded to each patient’s clinical details. The interobserver and intraobserver variability were assessed in three normal subjects and two patients with cirrhosis. To assess the interobserver variability, the two operators examined each of five subjects on the same day. Operators were unaware of one another’s results. To assess the intraobserver variability, the one operator examined each of the five patients on different days (the second examinations were performed at least a week after the first examination). The coefficient of variation (CV) was calculated as the SD/mean value. Interobserver variability showed the following CVs: for the arrival time in the HA, 3.4%; the arrival time in the PV, 4.5%; and the arrival time in the HV, 3.5%. The difference in arrival time for HA, PV, and HV between two operators did not exceed 1, 1, and 2 s, respectively. Intraobserver variability showed the following CVs: for HA arrival time, 6.4%; PV arrival time, 2.7%; and HV arrival time, 3.4%. The difference in arrival time for HA, PV and HV between two measurements did not exceed 2, 1, and 2 s, respectively.
2.3. Statistics Data were expressed as median values with ranges. Differences among the three groups were compared by an analysis of variance (ANOVA).
Table 1 Clinical and biochemical data of subjects a
Age (years) Gender (male/female) BMI (kg/m 2) Bilirubin (mg/dl) Albumin (g/l) AST (IU/l) ALT (IU/l) PT (%) Ascites* Encephalopathy* Esophageal varices* Histological staging* (I/II/III/IV) Histological grading* (I/II/III/IV)
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Cirrhotic patients (n ¼ 6)
41 (34–69) 4/2 20.7 (19–24) 0.8 (0.7–0.9) 4.6 (4.4–4.8) 22 (16–30) 30 (22–35) 115 (100–120) 0 0 0 – –
75 (62–76) 2/1 22.6 (20–25) 0.9 (0.7–1.1) 3.9 (3.1–4.6) 5 (45–69) 28 (26–31) 96 (76–109) 0 0 0 0/2/1/0 1/1/1/0
68 (65–74) 2/4 24.1 (21–29) 1.1 (0.7–3.3) 3.5 (2.9–3.6) 70 (20–129) 52 (27–116) 74 (55–93) 2 1 3 0/0/0/6 1/2/3/0
a Values are the median (range); *, number of patients; BMI, body mass index; AST, aspartate transaminase; ALT, alanine transaminase; PT, prothrombin time.
580
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
Fig. 1. Illustration of the examination technique. (a) Before a bolus injection of Levovist. Both the middle hepatic vein (MHV, arrow) and the umbilical portion of the portal vein (PV, arrowhead) were scanned. (b) 18 s after a bolus injection of Levovist. First, the contrast agent arrived at the hepatic artery (HA, small arrow) 14 s after the injection. At 18 s after the injection, HA was markedly enhanced and the PV was slightly enhanced, but the HV was not. (c) 27 s after a bolus injection of Levovist. The contrast agent then arrived at both the HA and PV (arrowhead). The HV was not yet enhanced. (d) 53 s after a bolus injection of Levovist. Finally, the contrast agent arrived at the MHV 32 s after the injection. MHV was markedly enhanced 53 s after the injection (arrow). Differences between two groups were compared by the Mann–Whitney test. Statistical significance was considered as P , 0:05 for the ANOVA and P , 0:017 for the Mann–Whitney test (Holm–Bonferroni correction).
3. Results The time-acoustic intensity curves could be measured clearly in all participants. All HV time-acoustic intensity curves could be classified visually into two types, a gradual-rising curve (Fig. 2a) and a rapid-rising curve (Fig. 2b). The gradual-rising curves were seen in all controls and all non-cirrhotic patients. The rapid-rising curves, on the other hand, were seen in all cirrhotic patients. The HA and PV time-acoustic intensity curves were similar between cirrhotic patients and control subjects or non-cirrhotic patients (Fig. 2a,b). Table 2 shows the analysis of arrival time in the HA, PV, and HV. The HA and PV arrival times were not significantly different in the cirrhotic patients compared with the controls and non-cirrhotic patients. Moreover, the time intervals between the PV and HA arrival times were not significantly different in the cirrhotic patients compared with the controls
and non-cirrhotic patients. On the other hand, the HV arrival time was significantly earlier in cirrhotic patients compared with the other two groups (18 (14–21) vs. 31 (26–34) and 30 (26–31) s, respectively; P , 0:0001 and P , 0:0001, respectively). The clear separation between cirrhotic patients and the other groups gives supports to a previous pilot study of Albrecht et al. [9]. The HV arrival times in controls and noncirrhotic patients were fairly late, ranging from 26 to 34 s, compared with the HV arrival time in cirrhotic patients, with a range from 14 to 21 s. In a similar manner, the time interval between the arrival time in the HV and that in the HA in cirrhotic subjects was significantly different from the other two groups (7 (4–12) vs. 18 (15–21) and 16 (16) s, respectively; P , 0:0001 and P ¼ 0:0001, respectively). In cirrhosis, the rapid-rising form of the time-acoustic intensity curve in the HV was very similar to that in the PV (Fig. 2). The time-acoustic intensity curve characterizations in the HVs and PVs are shown in Table 3. The HV rising rates in cirrhotic patients were significantly larger compared with those in controls and non-cirrhotic patients (2.39 (1.14–2.57) vs. 0.61 (0.28–0.81) and 0.70 (0.39–0.75) dB/s, respectively; P , 0:0001 and P ¼ 0:0001, respec-
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
tively). On the other hand, the PV rising rates in cirrhotic patients showed no difference compared with the rates in controls and non-cirrhotic patients. Interestingly, in patients with cirrhosis, the HV rising rates were similar to the PV rising rates (2.39 (1.14–2.57) vs. 2.33 (1.89–4.00), respectively). Moreover, time intervals between the HV arrival time and PV arrival time in cirrhotic patients were very short, with a median value of 1 s, and significantly different from the other two groups (Table 2).
4. Discussion Albrecht et al. [9] first reported that the assessment of the arrival time of a peripherally injected microbubble bolus in a
Fig. 2. The HA, PV, and HV time-acoustic intensity curves of a normal volunteer (a) and a patient with cirrhosis (b). There were no differences in HA and PV time-acoustic intensity curves between a cirrhotic patient and a normal volunteer. On the other hand, the HV time-acoustic intensity curves between a cirrhotic patient and a normal volunteer were clearly distinguishable. The HV time-acoustic curve in a cirrhotic patient demonstrated an earlier arrival time than that in a normal volunteer.
581
HV allowed discrimination between patients with biopsyproven cirrhosis and normal controls and patients with noncirrhotic diffuse liver disease; an arrival time of less than 24 s was 100% sensitive and 96% specific for the diagnosis of cirrhosis. It may, therefore, prove useful for screening cirrhosis less invasively. The earlier transit-times seen in patients with cirrhosis could have a hepatosplanchnic origin, and be caused by hyperdynamic central circulation, the presence of pulmonary shunts, or a combination of these, but Albrecht and coworkers were unable to elucidate these aspects. In the present study, we analyzed the HV transit-time using pulse-inversion imaging. HV arrival times were earlier in cirrhotic patients than in both control subjects and noncirrhotic patients. In all cirrhotic patients, HV arrival times were less than 24 s, while in all normal volunteers they were more than 24 s, which is the cut-off time in Albrecht’s study [9]. Moreover, we showed a median HV arrival time of 18 s for cirrhotic patients, which was in accordance with the mean HV arrival time of 18.3 s for cirrhotic patients reported by Albrecht et al. [9]. These findings indicate that the transittime analysis using pulse-inversion imaging is as useful as the spectral Doppler method for the diagnosis of cirrhosis. Bang et al. [10] also confirmed identical results between the spectral Doppler and pulse-inversion methods. We analyzed the HA and PV transit-times to investigate the cause of the earlier HV arrival time. The HA and PV arrival times were not different among control subjects, noncirrhotic patients, and cirrhotic patients. Systemic hemodynamic changes such as hyperdynamic circulation or the presence of pulmonary shunts were reported in cirrhotic patients [12–15]. Although these factors could shorten the HV arrival time, they would simultaneously affect the earlier arrival time in both the HA and PV. However, our results indicated no difference between the HA and PV arrival times in controls, non-cirrhotic patients and cirrhotic patients. This factor would thus have little effect on the early HV transit-time in cirrhosis. Furthermore, we analyzed the time interval between the HV arrival time and HA arrival time (DHV–HA), and between HV and PV (DHV–PV). These parameters mainly indicate intrahepatic circulation. Then, we showed a significantly earlier DHV–HA and DHV–PV in cirrhotic patients than those in control subjects and non-cirrhotic subjects. Although the time difference in HV arrival time between cirrhotic patients and control subjects (median HV arrival times were 18 and 31 s, respectively; Table 2) was 13 s, the time difference in DHV–HA between cirrhotic patients and control subjects (median DHV–HA were 7 and 18 s, respectively; Table 2) was 11 s, showing that the time difference in HV arrival time between cirrhotic patients and control subjects was almost entirely caused by intrahepatic hemodynamic changes. In cirrhosis, such changes occur with the formation of a regenerative nodule. Vascular casts have demonstrated that fibrous septa have a dense network of vessels with some anastomoses between branches of the PV and HV and
582
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
Table 2 Arrival time differences among HA, PV, and HV in control, hepatitis, and cirrhosis groups a
HA arrival time (s) PV arrival time (s) HV arrival time (s) DPV–HA (s) DHV–HA (s) DHV–PV (s)
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Patients with cirrhosis (n ¼ 6)
P value
12 (11–14) 17 (15–19) 31 (26–34) 5 (4–6) 18 (15–21) 13 (11–16)
14 (10–15) 20 (16–20) 30 (26–31) 6 (5–6) 16 (16) 10 (10–11)
12 17 18 6 7 1
0.4120 0.3402 , 0.0001 0.4832 , 0.0001 , 0.0001
(6–14) (12–20) (14–21)* ,† (4–8) (4–12)* ,† (0–4)* ,†
a Values are the median (range); DHV–HA, time delay between arrival in HV and that in HA; DHV–PV, time delay between arrival in HV and that in PV; DPV–HA, time delay between arrival in PV and that in HA. *P , 0:001 vs. normal; †P , 0:001 vs. hepatitis.
between branches of the HA and PV [16,17]. Our results suggest that this intrahepatic shunt in cirrhosis was the major cause of an earlier HV arrival time in cirrhotic patients. Thus, an earlier HV transit-time is thought to be based on histological changes in cirrhosis. All non-cirrhotic patients in our study and all but one of the 12 non-cirrhotic patients in Albrecht’s study [9] showed results with very similar curves to normal volunteers, indicating that no intrahepatic hemodynamic change (such as forming a regenerative nodule) occurred before the onset of cirrhosis. However, our small study could not entirely prove the relation between histological change and HV transit-time. Further large-scale studies are required. Interestingly, this test could obviate the need for biopsy because the changes in HV arrival time are based on intrahepatic hemodynamic factors in accordance with histological change. When used as a test for cirrhosis, this technique has an important limitation. Blomley et al. [18] reported that many patients with liver metastases show a left shift of HV timeintensity curves similar to that in patients with cirrhosis. In the present study, we excluded patients with liver tumors. Liver tumors are known to be associated with both arterialization of the liver blood supply and the presence of vascular shunts. Our study indicated that intrahepatic hemodynamic change was a major cause of the earlier HV arrival time. Thus, liver tumors with vascular shunts would cause the earlier HV arrival time. Conversely, when used as a test for liver metastases, the arrival time analysis using pulse-inversion imaging has an advantage. In transit-time analysis using pulse-inversion methods, more than one HV can be visualized in the same scan plane. Bang et al. [10] reported that the different arrival
times in different HVs were observed in patients with liver tumors. This phenomenon also suggests that intrahepatic hemodynamic change is the major cause of the earlier HV arrival time. The difference in the profile of the HV time-acoustic intensity curves between normal volunteers and cirrhotic patients is an interesting aspect of this study. In normal volunteers, the HV time-acoustic curve gradually rises. First, after bolus injection, the contrast agents reach the sinusoid from the HA. At this point, the contrast agents are diluted by the portal venous flow into the HV, so the vein shows virtually no enhancement. Next, the contrast agents reach the PV at which point the agents concentration rises in the HV. However, some of the Kuppfer cells take up the contrast agent [19]. Thus, in the normal liver the effectiveness of the agent in the HV increases only gradually. On the other hand, in cirrhosis, there is a sharp rise in the HV time-acoustic intensity curves. In cirrhosis, the contrast agents reach the HV via the HA through the intrahepatic shunt. Thus, the rising rate is large and similar to that in the PV with a certain kind of venous blood flow system. Therefore, a rapid enhancement of the HV would prove that the agent is bypassing the sinusoid. In conclusion, this study confirms that in cirrhotic patients there is an earlier HV arrival time of contrast and it shows that this earlier HV arrival time is due to intrahepatic and not extrahepatic hemodynamic changes. Our pilot study thus suggests that HV transit-time analysis using pulse-inversion imaging may be a simpler and less invasive technique compared with biopsy, and may be useful for the diagnosis of cirrhosis. A further large-scale study will be necessary to assess its diagnostic value.
Table 3 Time-acoustic intensity curve characterization in HV and PV a
HV rise time (s) HV rising rate (dB/s) PV rise time (s) PV rising rate (dB/s) a
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Patients with cirrhosis (n ¼ 6)
P value
19 (16–23) 0.61 (0.28–0.81) 10 (3–14) 2.65 (1.43–4.00)
12 (10–18) 0.70 (0.39–0.75) 8 (5–14) 2.37 (1.21–4.60)
7 (6–7)* ,† 2.39 (1.14–2.57)* ,† 7 (5–9) 2.33 (1.89–4.00)
, 0.0001 , 0.0001 0.5280 0.9948
Values are the median (range). *P , 0:001 vs. normal; †P , 0:001 vs. hepatitis.
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
References [1] Leen E, Goldberg JA, Anderson JR, Robertson J, Moule B, Cooke TG, et al. Hepatic perfusion changes in patients with liver metastases: comparison with those patients with cirrhosis. Gut 1993;34: 554–557. [2] Tsushima Y, Blomley MJ, Kusano S, Endo K. The portal component of hepatic perfusion measured by dynamic CT: an indicator of hepatic parenchymal damage. Dig Dis Sci 1999;44:1632–1638. [3] Lautt WW. Mechanism and role of intrinsic regulation of hepatic arterial blood flow: hepatic buffer response. Am J Physiol 1985;249:G549–G556. [4] Kleber G, Steudel N, Behrmann C, Zipprich A, Hubner G, Lotterer E, et al. Hepatic arterial flow volume and reserve in patients with cirrhosis: use of intra-arterial Doppler and adenosine infusion. Gastroenterology 1999;116:906–914. [5] Johnson DJ, Muhlbacher F, Wilmore DW. Measurement of hepatic blood flow. J Surg Res 1985;39:470–481. [6] Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993;188:405–411. [7] Reichen J. Assessment of hepatic function with xenobiotics. Semin Liver Dis 1995;15:189–201. [8] Blomley MJK, Cosgrove DO. Microbubble echo-enhancers: a new direction for ultrasound? Lancet 1997;349:1855–1856. [9] Albrecht T, Blomley MJK, Cosgrove DO, Taylor-Robinson SD, Jayaram V, Eckersley R, et al. Non-invasive diagnosis of hepatic cirrhosis by transit-time analysis of an ultrasound contrast agent. Lancet 1999;353:1579–1583.
583
[10] Bang N, Nielsen MB, Rasmussen AN, Osterhammel PA, Pedersen JF. Hepatic vein transit time of an ultrasound contrast agent: simplified procedure using pulse inversion imaging. Br J Radiol 2001;74:752–755. [11] Hu¨ bscher SG. Histological grading and staging in chronic hepatitis: clinical applications and problems. J Hepatol 1998;29:1015–1022. [12] Murray JF, Dawson AM, Sherlock S. Circulatory changes in chronic liver disease. Am J Med 1958;24:358–367. [13] Kontos HA, Shapiro W, Mauck HP, Patterson JL. General and regional circulatory alterations in cirrhosis of the liver. Am J Med 1964;37:526–535. [14] Wolfe JD, Tashkin DP, Holly FE, Brachman MB, Genovesi MG. Hypoxemia of cirrhosis: detection of abnormal small pulmonary vascular channels by a quantitative radionuclide method. Am J Med 1977;63:746–754. [15] Agusti AGN, Roca J, Bosch J, Rodriguez-Roisin R. The lung in patients with cirrhosis. J Hepatol 1990;10:251–257. [16] Popper H, Elias H, Petty DE. Vascular pattern of cirrhotic liver. Am J Clin Pathol 1952;22:717–729. [17] Hales MR, Allan JS, Hall EM. Injection of corrosion studies of normal and cirrhotic liver. Am J Pathol 1959;35:909–941. [18] Blomley MJK, Albrecht T, Cosgrove DO, Jayaram V, Eckersley RJ, Patel N, et al. Liver vascular transit time analysed with dynamic hepatic venography with bolus injections of an US contrast agent: early experience in seven patients with metastases. Radiology 1998;209:862–866. [19] Blomley MJK, Albrecht T, Cosgrove DO, Jayaram V, Butler-Brames J, Eckersley RJ. Stimulated acoustic emission in liver parenchyma with Levovist. Lancet 1998;351:568.
Earlier hepatic vein transit-time measured by contrast ultrasonography reflects intrahepatic hemodynamic changes accompanying cirrhosis Hiroyuki Sugimoto*, Tetsuya Kaneko, Masashi Hirota, Ekmel Tezel, Akimasa Nakao Department of Surgery II, Nagoya University School of Medicine, Tsurumai-cho 65, Showa-ku, Nagoya 466-8550, Japan
Background/Aims: Non-invasive diagnosis of cirrhosis by transit-time analysis of an ultrasound contrast agent has been reported, even though the mechanism by which contrast arrives to the hepatic vein earlier in cirrhosis than in normal controls is unknown. The aim of this study is to assess whether the earlier appearance of contrast in the hepatic vein depends on intrahepatic or extrahepatic causes. Methods: There were 15 participants: six volunteers, three patients with hepatitis, and six with cirrhosis. The contrast agent was given intravenously, and transit-time analysis of the hepatic artery, portal vein and hepatic vein was performed. The time-acoustic intensity curves in the three vessels were analyzed by an image and cineloop display and quantification software package. Results: The hepatic artery and portal vein arrival times were not significantly different among the three groups. On the other hand, hepatic vein arrival times were significantly earlier in cirrhosis (median 18 seconds) compared with arrival times in hepatitis patients (median 30 seconds, P , 0:001) and in healthy volunters (median 31 seconds, P , 0:001). These results give support to a previous pilot study and indicate that most of the time delay in hepatic vein arrival time between cirrhosis and the other groups originated from intrahepatic circulation abnormalities. Conclusions: This study confirms that the earlier appearance of contrast in the hepatic vein observed in cirrhosis is due to intrahepatic, and not extrahepatic, hemodynamic changes. q 2002 European Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. Keywords: Ultrasonography; Cirrhosis; Hepatic vein transit-time; Pulse-inversion imaging
1. Introduction In cirrhosis, important changes occur in liver circulation. The increase in intrahepatic vascular resistance decreases the portal fraction of liver perfusion [1,2]. This decrease in portal perfusion is partially compensated by an increase in arterial inflow [1,3,4]. To evaluate cirrhosis, various methods exist for the determination of hepatic perfusion [5–7]. Most of them, however, are invasive or controversial [5]. Recently, microbubble contrast agents have been developed for ultrasonography [8]. Albrecht et al. [9] reported a technique for the diagnosis of hepatic cirrhosis by transit-time analysis of an ultrasound contrast agent. Their cirrhotic patients showed an
Received 28 May 2002; received in revised form 10 July 2002; accepted 17 July 2002 * Corresponding author. Tel.: 181-52-744-2245; fax: 181-52-744-2255. E-mail address: [email protected] (H. Sugimoto).
earlier arrival time of the contrast agent in their hepatic veins (HVs) than normal volunteers. Cirrhosis involves several intrahepatic and extrahepatic hemodynamic changes that contribute to the earlier arrival time such as arterialization of the liver, intrahepatic shunt, pulmonary arteriovenous shunt, or hyperdynamic circulation. However, to what degree such changes contributed to their results can not be assessed from their study. To investigate intrahepatic hemodynamic changes, not only must transit-time be measured in the HV but also in both the hepatic artery (HA) and portal vein (PV). However, the measurement of transit-time in the HA by spectral Doppler methods is very difficult because the sample volume cannot be kept within the small diameter of the HA long enough to complete the measurement. Bang et al. [10] reported that pulse-inversion imaging is simpler and has certain advantages over spectral Doppler quantification in assessing the transit-time of a bolus of an echo-enhancing agent. The major advantage of this method
0168-8278/02/$20.00 q 2002 European Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: S01 68- 8278(02)0026 4-7
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
is that the arrival of the contrast agent can be visually tracked. Thus, this pulse-inversion imaging can elucidate transit-time in both the HA and PV as it does in the HV. We conducted transit-time analysis simultaneously in the PV, HA, and HV using pulse-inversion imaging. The aim of this study is to assess whether the cause of an earlier HV transit-time depends on intrahepatic or extrahepatic causes. 2. Patients and methods 2.1. Patients There were 15 participants: six normal volunteers (four men, two women, median age 41 years, age range 34–69 years) with no history or clinical signs of liver disease; three patients (two men, one woman, median age 75 years, age range 62–76 years) with biopsy-proven non-cirrhotic diffuse liver disease (chronic hepatitis B); and six patients (two men, four women, median age 68 years, age range 65–74 years) with biopsy-proven cirrhosis (hepatitis C). Six patients with cirrhosis were classified as Child–Pugh A (three patients) and Child–Pugh B (three patients). On the basis of pathology in the biopsy, liver cirrhosis was defined as diffuse nodularity with fibrous septa, and chronic hepatitis was defined as the presence of bridging fibrosis, but without complete pseudoacini. Other characteristics of the subjects including histological grading and staging [11] are summarized in Table 1. This study protocol was in accordance with the guidelines of the Declaration of Helsinki and has been reviewed by the ethics committee of the Nagoya University School of Medicine. All patients gave their informed consent to inclusion in the study.
2.2. Transit-time analysis using the pulse-inversion harmonic method All subjects were tested in the morning after an overnight fast. All pulseinversion harmonic studies were performed with an ATL HDI 5000 (Philips, Bothell, WA) using a 2–5 MHz convex probe. All examinations were performed with patients in the supine position. In transit-time analysis using the contrast agent, the umbilical PV, HA, and middle or left HV were simultaneously scanned in a transverse section (Fig. 1). In four patients whose three intrahepatic vessels could not be simultaneously scanned, two injections of contrast agent were given at least 10 min apart to assess the transit-time analysis (one normal volunteer, one patient with noncirrhotic diffuse liver disease, and two patients with cirrhosis). Apparatus
579
settings for pulse-inversion harmonic mode were the following: focus 9 cm, mechanical index 1.5, and trigger mode with 300 ms delay. The microbubble contrast agent used was Levovist (Schering AG, Berlin, Germany). A bolus injection of 2.5 g Levovist 300 mg/ml was given via a 21 G cannula into an antecubital vein followed by 10 ml of saline at 2 ml/s. We asked the subjects to hold their breath, and we then recorded from 0 to 60 s after the contrast agent injection. Subjects who could not hold their breath were to breathe gently and were recorded for 60 s after the injection. The time delay from injection until the first echogenic bubbles of contrast agent could be subsequently seen in the HA, PV and HV was measured, and an analysis of time-acoustic intensity curves was performed using ATL HDI Lab (Philips, Bothell, WA). ATL HDI Lab is an image and cineloop display and quantification software package. Digital images and cineloop acquired by the ATL HDI 5000 can be transferred to a personal computer running ATL HDI Lab. One can trace the region of interest (ROI) on the HA, PV and HV, and then automatically obtain the time-acoustic intensity curves. The arrival time was defined as the time delay from injection until the first echogenic bubbles of contrast agent could be seen. The time to peak was defined as the time interval from the beginning of the injection to the peak of the time-acoustic intensity curve. The rising time was defined as the time interval from the arrival time to the peak of the time-acoustic intensity curve. The rising rate was defined as the average increase of the signal per second between the arrival time and the peak time. All examinations were performed by three of the authors (H.S., M.H., and T.K.) who were blinded to each patient’s clinical details. The interobserver and intraobserver variability were assessed in three normal subjects and two patients with cirrhosis. To assess the interobserver variability, the two operators examined each of five subjects on the same day. Operators were unaware of one another’s results. To assess the intraobserver variability, the one operator examined each of the five patients on different days (the second examinations were performed at least a week after the first examination). The coefficient of variation (CV) was calculated as the SD/mean value. Interobserver variability showed the following CVs: for the arrival time in the HA, 3.4%; the arrival time in the PV, 4.5%; and the arrival time in the HV, 3.5%. The difference in arrival time for HA, PV, and HV between two operators did not exceed 1, 1, and 2 s, respectively. Intraobserver variability showed the following CVs: for HA arrival time, 6.4%; PV arrival time, 2.7%; and HV arrival time, 3.4%. The difference in arrival time for HA, PV and HV between two measurements did not exceed 2, 1, and 2 s, respectively.
2.3. Statistics Data were expressed as median values with ranges. Differences among the three groups were compared by an analysis of variance (ANOVA).
Table 1 Clinical and biochemical data of subjects a
Age (years) Gender (male/female) BMI (kg/m 2) Bilirubin (mg/dl) Albumin (g/l) AST (IU/l) ALT (IU/l) PT (%) Ascites* Encephalopathy* Esophageal varices* Histological staging* (I/II/III/IV) Histological grading* (I/II/III/IV)
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Cirrhotic patients (n ¼ 6)
41 (34–69) 4/2 20.7 (19–24) 0.8 (0.7–0.9) 4.6 (4.4–4.8) 22 (16–30) 30 (22–35) 115 (100–120) 0 0 0 – –
75 (62–76) 2/1 22.6 (20–25) 0.9 (0.7–1.1) 3.9 (3.1–4.6) 5 (45–69) 28 (26–31) 96 (76–109) 0 0 0 0/2/1/0 1/1/1/0
68 (65–74) 2/4 24.1 (21–29) 1.1 (0.7–3.3) 3.5 (2.9–3.6) 70 (20–129) 52 (27–116) 74 (55–93) 2 1 3 0/0/0/6 1/2/3/0
a Values are the median (range); *, number of patients; BMI, body mass index; AST, aspartate transaminase; ALT, alanine transaminase; PT, prothrombin time.
580
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
Fig. 1. Illustration of the examination technique. (a) Before a bolus injection of Levovist. Both the middle hepatic vein (MHV, arrow) and the umbilical portion of the portal vein (PV, arrowhead) were scanned. (b) 18 s after a bolus injection of Levovist. First, the contrast agent arrived at the hepatic artery (HA, small arrow) 14 s after the injection. At 18 s after the injection, HA was markedly enhanced and the PV was slightly enhanced, but the HV was not. (c) 27 s after a bolus injection of Levovist. The contrast agent then arrived at both the HA and PV (arrowhead). The HV was not yet enhanced. (d) 53 s after a bolus injection of Levovist. Finally, the contrast agent arrived at the MHV 32 s after the injection. MHV was markedly enhanced 53 s after the injection (arrow). Differences between two groups were compared by the Mann–Whitney test. Statistical significance was considered as P , 0:05 for the ANOVA and P , 0:017 for the Mann–Whitney test (Holm–Bonferroni correction).
3. Results The time-acoustic intensity curves could be measured clearly in all participants. All HV time-acoustic intensity curves could be classified visually into two types, a gradual-rising curve (Fig. 2a) and a rapid-rising curve (Fig. 2b). The gradual-rising curves were seen in all controls and all non-cirrhotic patients. The rapid-rising curves, on the other hand, were seen in all cirrhotic patients. The HA and PV time-acoustic intensity curves were similar between cirrhotic patients and control subjects or non-cirrhotic patients (Fig. 2a,b). Table 2 shows the analysis of arrival time in the HA, PV, and HV. The HA and PV arrival times were not significantly different in the cirrhotic patients compared with the controls and non-cirrhotic patients. Moreover, the time intervals between the PV and HA arrival times were not significantly different in the cirrhotic patients compared with the controls
and non-cirrhotic patients. On the other hand, the HV arrival time was significantly earlier in cirrhotic patients compared with the other two groups (18 (14–21) vs. 31 (26–34) and 30 (26–31) s, respectively; P , 0:0001 and P , 0:0001, respectively). The clear separation between cirrhotic patients and the other groups gives supports to a previous pilot study of Albrecht et al. [9]. The HV arrival times in controls and noncirrhotic patients were fairly late, ranging from 26 to 34 s, compared with the HV arrival time in cirrhotic patients, with a range from 14 to 21 s. In a similar manner, the time interval between the arrival time in the HV and that in the HA in cirrhotic subjects was significantly different from the other two groups (7 (4–12) vs. 18 (15–21) and 16 (16) s, respectively; P , 0:0001 and P ¼ 0:0001, respectively). In cirrhosis, the rapid-rising form of the time-acoustic intensity curve in the HV was very similar to that in the PV (Fig. 2). The time-acoustic intensity curve characterizations in the HVs and PVs are shown in Table 3. The HV rising rates in cirrhotic patients were significantly larger compared with those in controls and non-cirrhotic patients (2.39 (1.14–2.57) vs. 0.61 (0.28–0.81) and 0.70 (0.39–0.75) dB/s, respectively; P , 0:0001 and P ¼ 0:0001, respec-
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
tively). On the other hand, the PV rising rates in cirrhotic patients showed no difference compared with the rates in controls and non-cirrhotic patients. Interestingly, in patients with cirrhosis, the HV rising rates were similar to the PV rising rates (2.39 (1.14–2.57) vs. 2.33 (1.89–4.00), respectively). Moreover, time intervals between the HV arrival time and PV arrival time in cirrhotic patients were very short, with a median value of 1 s, and significantly different from the other two groups (Table 2).
4. Discussion Albrecht et al. [9] first reported that the assessment of the arrival time of a peripherally injected microbubble bolus in a
Fig. 2. The HA, PV, and HV time-acoustic intensity curves of a normal volunteer (a) and a patient with cirrhosis (b). There were no differences in HA and PV time-acoustic intensity curves between a cirrhotic patient and a normal volunteer. On the other hand, the HV time-acoustic intensity curves between a cirrhotic patient and a normal volunteer were clearly distinguishable. The HV time-acoustic curve in a cirrhotic patient demonstrated an earlier arrival time than that in a normal volunteer.
581
HV allowed discrimination between patients with biopsyproven cirrhosis and normal controls and patients with noncirrhotic diffuse liver disease; an arrival time of less than 24 s was 100% sensitive and 96% specific for the diagnosis of cirrhosis. It may, therefore, prove useful for screening cirrhosis less invasively. The earlier transit-times seen in patients with cirrhosis could have a hepatosplanchnic origin, and be caused by hyperdynamic central circulation, the presence of pulmonary shunts, or a combination of these, but Albrecht and coworkers were unable to elucidate these aspects. In the present study, we analyzed the HV transit-time using pulse-inversion imaging. HV arrival times were earlier in cirrhotic patients than in both control subjects and noncirrhotic patients. In all cirrhotic patients, HV arrival times were less than 24 s, while in all normal volunteers they were more than 24 s, which is the cut-off time in Albrecht’s study [9]. Moreover, we showed a median HV arrival time of 18 s for cirrhotic patients, which was in accordance with the mean HV arrival time of 18.3 s for cirrhotic patients reported by Albrecht et al. [9]. These findings indicate that the transittime analysis using pulse-inversion imaging is as useful as the spectral Doppler method for the diagnosis of cirrhosis. Bang et al. [10] also confirmed identical results between the spectral Doppler and pulse-inversion methods. We analyzed the HA and PV transit-times to investigate the cause of the earlier HV arrival time. The HA and PV arrival times were not different among control subjects, noncirrhotic patients, and cirrhotic patients. Systemic hemodynamic changes such as hyperdynamic circulation or the presence of pulmonary shunts were reported in cirrhotic patients [12–15]. Although these factors could shorten the HV arrival time, they would simultaneously affect the earlier arrival time in both the HA and PV. However, our results indicated no difference between the HA and PV arrival times in controls, non-cirrhotic patients and cirrhotic patients. This factor would thus have little effect on the early HV transit-time in cirrhosis. Furthermore, we analyzed the time interval between the HV arrival time and HA arrival time (DHV–HA), and between HV and PV (DHV–PV). These parameters mainly indicate intrahepatic circulation. Then, we showed a significantly earlier DHV–HA and DHV–PV in cirrhotic patients than those in control subjects and non-cirrhotic subjects. Although the time difference in HV arrival time between cirrhotic patients and control subjects (median HV arrival times were 18 and 31 s, respectively; Table 2) was 13 s, the time difference in DHV–HA between cirrhotic patients and control subjects (median DHV–HA were 7 and 18 s, respectively; Table 2) was 11 s, showing that the time difference in HV arrival time between cirrhotic patients and control subjects was almost entirely caused by intrahepatic hemodynamic changes. In cirrhosis, such changes occur with the formation of a regenerative nodule. Vascular casts have demonstrated that fibrous septa have a dense network of vessels with some anastomoses between branches of the PV and HV and
582
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
Table 2 Arrival time differences among HA, PV, and HV in control, hepatitis, and cirrhosis groups a
HA arrival time (s) PV arrival time (s) HV arrival time (s) DPV–HA (s) DHV–HA (s) DHV–PV (s)
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Patients with cirrhosis (n ¼ 6)
P value
12 (11–14) 17 (15–19) 31 (26–34) 5 (4–6) 18 (15–21) 13 (11–16)
14 (10–15) 20 (16–20) 30 (26–31) 6 (5–6) 16 (16) 10 (10–11)
12 17 18 6 7 1
0.4120 0.3402 , 0.0001 0.4832 , 0.0001 , 0.0001
(6–14) (12–20) (14–21)* ,† (4–8) (4–12)* ,† (0–4)* ,†
a Values are the median (range); DHV–HA, time delay between arrival in HV and that in HA; DHV–PV, time delay between arrival in HV and that in PV; DPV–HA, time delay between arrival in PV and that in HA. *P , 0:001 vs. normal; †P , 0:001 vs. hepatitis.
between branches of the HA and PV [16,17]. Our results suggest that this intrahepatic shunt in cirrhosis was the major cause of an earlier HV arrival time in cirrhotic patients. Thus, an earlier HV transit-time is thought to be based on histological changes in cirrhosis. All non-cirrhotic patients in our study and all but one of the 12 non-cirrhotic patients in Albrecht’s study [9] showed results with very similar curves to normal volunteers, indicating that no intrahepatic hemodynamic change (such as forming a regenerative nodule) occurred before the onset of cirrhosis. However, our small study could not entirely prove the relation between histological change and HV transit-time. Further large-scale studies are required. Interestingly, this test could obviate the need for biopsy because the changes in HV arrival time are based on intrahepatic hemodynamic factors in accordance with histological change. When used as a test for cirrhosis, this technique has an important limitation. Blomley et al. [18] reported that many patients with liver metastases show a left shift of HV timeintensity curves similar to that in patients with cirrhosis. In the present study, we excluded patients with liver tumors. Liver tumors are known to be associated with both arterialization of the liver blood supply and the presence of vascular shunts. Our study indicated that intrahepatic hemodynamic change was a major cause of the earlier HV arrival time. Thus, liver tumors with vascular shunts would cause the earlier HV arrival time. Conversely, when used as a test for liver metastases, the arrival time analysis using pulse-inversion imaging has an advantage. In transit-time analysis using pulse-inversion methods, more than one HV can be visualized in the same scan plane. Bang et al. [10] reported that the different arrival
times in different HVs were observed in patients with liver tumors. This phenomenon also suggests that intrahepatic hemodynamic change is the major cause of the earlier HV arrival time. The difference in the profile of the HV time-acoustic intensity curves between normal volunteers and cirrhotic patients is an interesting aspect of this study. In normal volunteers, the HV time-acoustic curve gradually rises. First, after bolus injection, the contrast agents reach the sinusoid from the HA. At this point, the contrast agents are diluted by the portal venous flow into the HV, so the vein shows virtually no enhancement. Next, the contrast agents reach the PV at which point the agents concentration rises in the HV. However, some of the Kuppfer cells take up the contrast agent [19]. Thus, in the normal liver the effectiveness of the agent in the HV increases only gradually. On the other hand, in cirrhosis, there is a sharp rise in the HV time-acoustic intensity curves. In cirrhosis, the contrast agents reach the HV via the HA through the intrahepatic shunt. Thus, the rising rate is large and similar to that in the PV with a certain kind of venous blood flow system. Therefore, a rapid enhancement of the HV would prove that the agent is bypassing the sinusoid. In conclusion, this study confirms that in cirrhotic patients there is an earlier HV arrival time of contrast and it shows that this earlier HV arrival time is due to intrahepatic and not extrahepatic hemodynamic changes. Our pilot study thus suggests that HV transit-time analysis using pulse-inversion imaging may be a simpler and less invasive technique compared with biopsy, and may be useful for the diagnosis of cirrhosis. A further large-scale study will be necessary to assess its diagnostic value.
Table 3 Time-acoustic intensity curve characterization in HV and PV a
HV rise time (s) HV rising rate (dB/s) PV rise time (s) PV rising rate (dB/s) a
Control subjects (n ¼ 6)
Non-cirrhotic patients (n ¼ 3)
Patients with cirrhosis (n ¼ 6)
P value
19 (16–23) 0.61 (0.28–0.81) 10 (3–14) 2.65 (1.43–4.00)
12 (10–18) 0.70 (0.39–0.75) 8 (5–14) 2.37 (1.21–4.60)
7 (6–7)* ,† 2.39 (1.14–2.57)* ,† 7 (5–9) 2.33 (1.89–4.00)
, 0.0001 , 0.0001 0.5280 0.9948
Values are the median (range). *P , 0:001 vs. normal; †P , 0:001 vs. hepatitis.
H. Sugimoto et al. / Journal of Hepatology 37 (2002) 578–583
References [1] Leen E, Goldberg JA, Anderson JR, Robertson J, Moule B, Cooke TG, et al. Hepatic perfusion changes in patients with liver metastases: comparison with those patients with cirrhosis. Gut 1993;34: 554–557. [2] Tsushima Y, Blomley MJ, Kusano S, Endo K. The portal component of hepatic perfusion measured by dynamic CT: an indicator of hepatic parenchymal damage. Dig Dis Sci 1999;44:1632–1638. [3] Lautt WW. Mechanism and role of intrinsic regulation of hepatic arterial blood flow: hepatic buffer response. Am J Physiol 1985;249:G549–G556. [4] Kleber G, Steudel N, Behrmann C, Zipprich A, Hubner G, Lotterer E, et al. Hepatic arterial flow volume and reserve in patients with cirrhosis: use of intra-arterial Doppler and adenosine infusion. Gastroenterology 1999;116:906–914. [5] Johnson DJ, Muhlbacher F, Wilmore DW. Measurement of hepatic blood flow. J Surg Res 1985;39:470–481. [6] Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993;188:405–411. [7] Reichen J. Assessment of hepatic function with xenobiotics. Semin Liver Dis 1995;15:189–201. [8] Blomley MJK, Cosgrove DO. Microbubble echo-enhancers: a new direction for ultrasound? Lancet 1997;349:1855–1856. [9] Albrecht T, Blomley MJK, Cosgrove DO, Taylor-Robinson SD, Jayaram V, Eckersley R, et al. Non-invasive diagnosis of hepatic cirrhosis by transit-time analysis of an ultrasound contrast agent. Lancet 1999;353:1579–1583.
583
[10] Bang N, Nielsen MB, Rasmussen AN, Osterhammel PA, Pedersen JF. Hepatic vein transit time of an ultrasound contrast agent: simplified procedure using pulse inversion imaging. Br J Radiol 2001;74:752–755. [11] Hu¨ bscher SG. Histological grading and staging in chronic hepatitis: clinical applications and problems. J Hepatol 1998;29:1015–1022. [12] Murray JF, Dawson AM, Sherlock S. Circulatory changes in chronic liver disease. Am J Med 1958;24:358–367. [13] Kontos HA, Shapiro W, Mauck HP, Patterson JL. General and regional circulatory alterations in cirrhosis of the liver. Am J Med 1964;37:526–535. [14] Wolfe JD, Tashkin DP, Holly FE, Brachman MB, Genovesi MG. Hypoxemia of cirrhosis: detection of abnormal small pulmonary vascular channels by a quantitative radionuclide method. Am J Med 1977;63:746–754. [15] Agusti AGN, Roca J, Bosch J, Rodriguez-Roisin R. The lung in patients with cirrhosis. J Hepatol 1990;10:251–257. [16] Popper H, Elias H, Petty DE. Vascular pattern of cirrhotic liver. Am J Clin Pathol 1952;22:717–729. [17] Hales MR, Allan JS, Hall EM. Injection of corrosion studies of normal and cirrhotic liver. Am J Pathol 1959;35:909–941. [18] Blomley MJK, Albrecht T, Cosgrove DO, Jayaram V, Eckersley RJ, Patel N, et al. Liver vascular transit time analysed with dynamic hepatic venography with bolus injections of an US contrast agent: early experience in seven patients with metastases. Radiology 1998;209:862–866. [19] Blomley MJK, Albrecht T, Cosgrove DO, Jayaram V, Butler-Brames J, Eckersley RJ. Stimulated acoustic emission in liver parenchyma with Levovist. Lancet 1998;351:568.