change in hepatic arterial hemodynamics induced by hepatocellular carcinoma detected with doppler...

Change in Hepatic Arterial HemodynamicsInduced by Hepatocellular CarcinomaDetected with Doppler Sonography

Hitoshi Someda, MD,1 Fuminori Moriyasu, MD,2 Noriyuki Hamato, MD,1 Masazumi Fujimoto, MD,1

Minoru Okuma, MD1

1 First Department of Internal Medicine, Kyoto University School of Medicine, 54 Shogoin-Kawaharacho,Sakyo-ku, Kyoto 606, Japan2 Division of Gastroenterology and Hepatology, Kyoto University School of Medicine, 54 Shogoin-Kawaharacho,Sakyo-ku, Kyoto 606, Japan

Received 6 August 1996; accepted 12 February 1997

ABSTRACT: Purpose. To understand the hemody-namic differences between the hepatic arterialbranches that supply hepatocellular carcinomas(HCCs) and those that do not, we compared the veloc-ity waveforms of both types of arteries.

Methods. Using duplex color Doppler sonography,we examined 38 patients with HCC localized within asingle lobe of the liver and 34 patients with chronicliver disease but without HCC (controls). We mea-sured angle-corrected peak systolic velocity and thepulsatility index (PI) of color-coded hepatic arteriesalong the right anterior segmental portal branch andthe vertical segment of the left portal vein.

Results. There was no significant difference in peaksystolic velocity and PI between the arterial branchestested in the controls. In contrast, we found a signifi-cantly lower PI and a higher peak systolic velocity inthe arterial branches supplying the tumor than inthose not supplying the tumor in patients with an HCC3 cm or larger in diameter (p < 0.05). The degree ofthese hemodynamic changes correlated with the tu-mor size and the presence or absence of tumor throm-bus in the major portal branches.

Conclusions. These results indicate that the hepaticarterial branch supplying an HCC has a lower imped-ance than the branch not supplying the tumor. © 1997John Wiley & Sons, Inc. J Clin Ultrasound 25:359–365,1997.

Keywords: hepatocellular carcinoma; Doppler ultraso-nography; hepatic artery; hepatic artery pulsatility in-dex; liver circulation

Many studies on Doppler sonography of hepa-tocellular carcinoma (HCC) have reported

high-flow arterial signals within and around themass that reflected the characteristic tumor-related vascular changes demonstrated by angi-ography and microscopic examination.1–4 Most ofthese investigators thought that the high-flow ar-terial signals of HCC originated from arteriopor-tal shunts,2–4 but their pulsed Doppler studies didnot include a quantitative spectral analysis of thearterial waveforms and made no distinction be-tween the signals obtained from within the tumorand those from around it. Duplex color Dopplersonography makes it possible to depict a smallvessel and to analyze its velocity waveformsquantitatively. Using this modality, some inves-tigators have found that high peak systolic veloc-ity (PSV) correlates with arteriovenous shuntsand malignant portal vein thrombus and thatvarious degrees of resistance waveforms are ob-tained from intratumoral and peritumoral arte-rial Doppler signals.5–7 However, such arterialsignals obtained near a tumor, either internallyor peripherally, do not always reflect the entireperipheral vascular bed. In this study, we usedduplex color Doppler sonography to study thegross hemodynamics of the proximal arteries thatsupply HCCs and of those that do not.

PATIENTS AND METHODS

Of the 81 patients with HCC who underwentDoppler examination between January 1991 and

Correspondence to: Dr. Moriyasu

© 1997 John Wiley & Sons, Inc. CCC 0091-2751/97/070359-07

359VOL. 25, NO. 7, SEPTEMBER 1997

March 1994, 38 patients [29 men and 9 womenaged 40–84 years (mean age, 61 years)] had a soli-tary HCC located in either the anterior segmentof the right lobe (segment 8 or 5; 28 patients) orthe lateral inferior segment of the left lobe (seg-ment 3; 10 patients) of the liver. The patientswith HCC also had chronic liver disease caused byhepatitis C virus (HCV) in 34 patients and byhepatitis B virus (HBV) in 4. Pathologic proof ofHCC was obtained in 32 patients by surgical re-section or sonographically guided percutaneousbiopsy. In the remaining 6 patients, the diagnosiswas established on the basis of clinical data andimaging studies, including history of infectionwith either HBV or HCV, elevated serum level ofa-fetoprotein, absence of an extrahepatic primarymalignant tumor, hepatic angiography, computedtomography (CT), and sonography. All patientswith HCC underwent selective hepatic arteriog-raphy and arterioportography within 2 weeks af-ter Doppler sonography. The maximum sizes oftumors measured with CT and sonographyranged from 1.3 to 23.0 cm (mean, 6.7 cm).

To assess the hemodynamic changes inducedby HCC, we also studied 34 patients without HCCwho had chronic liver disease caused by HBV orHCV (control group). Sixteen patients [9 men and7 women aged 23–67 years (mean age, 49 years)]had chronic hepatitis, and the remaining 18 pa-tients [11 men and 7 women aged 44–73 years

(mean, 58 years)] had liver cirrhosis. Chronicliver disease was diagnosed by histologic analysisin 28 patients and by clinical examination in 6.

Doppler studies were performed with an SSA-270A color Doppler ultrasound unit (ToshibaMedical Systems, Tokyo, Japan) using a 3.75-MHz phased-array transducer operating at a fre-quency of 3.0 MHz in Doppler mode. The width ofthe sample volume was 2 mm, and a pulse repeti-tion frequency of 4.5–9.0 kHz was selected to ob-tain an adequate Doppler spectral waveform ofthe targeted blood vessels without aliasing. Thehigh-pass filter of the flowmeter was set as low aspossible in each patient.

Color Doppler imaging was used to identifyvessels of interest and serve as a guide for spec-tral analysis. To quantify the Doppler velocitywaveform of the hepatic artery, we used the pul-satility index (PI) and angle-corrected PSV. ThePI value was calculated by averaging 3–4 con-secutive measurements using built-in software.All measurements in this study were obtained atthe proximal site of the left hepatic arterialbranch (Figure 1) and the right hepatic arterialbranch (Figure 2). In a patient with an HCC lo-cated in the right lobe, the right and left arterialbranches corresponded to the arteries that didand did not supply the tumor, respectively. PSVwas not measured if the hepatic artery was tor-

FIGURE 1. Duplex color Doppler sonogram shows left hepatic arterial branch (coded red) along the verticalsegment (umbilical portion) of the left portal vein and its velocity waveform.

SOMEDA ET AL

360 JOURNAL OF CLINICAL ULTRASOUND

tuous. For all measurements of PSV and PI, theangle of insonation was less than 30°.

Arterioportal shunts were diagnosed when thehepatic angiogram showed the portal vein in thearterial phase and the thread-and-streaks sign.8

Tumor thrombi, which were located in the 3rd ormore proximal portal branches, were diagnosedby the thread-and-streaks sign on angiogramsand/or the presence on Doppler study of pulsatileflow in the thrombus (Figure 3).9,10

All data are expressed as means ± standarddeviation. The Wilcoxon signed-rank test wasused to compare the values obtained from the he-patic arterial branches in each patient. A p valueof less than 0.05 was considered to indicate sta-tistical significance.

RESULTS

Table 1 shows the measurements for PI and PSVof the hepatic arteries in the 34 patients withchronic liver disease (controls). In 8 of these pa-tients, it was impossible to perform angle correc-tion because color Doppler inadequately depictedthe hepatic arteries. We observed no significantdifference in the PI and PSV between the left andthe right hepatic arterial branches in controls.

Table 2 shows the measurements for PI andPSV in patients with HCC according to tumor

size. In 7 of these patients, it was impossible toperform angle correction because color Dopplerinadequately depicted the hepatic arteries. Ex-cept for patients with HCCs smaller than 3 cm,the PI and PSV of the arteries supplying HCCswere significantly lower and higher, respectively,than those of the arteries not supplying tumors(Figure 4). The HCC characteristics for the threetumor sizes (ie, <3 cm, 3–6 cm, and >6 cm in di-ameter) are summarized in Table 3. Angiographyshowed a hypervascular tumor in all patients andan arterioportal shunt in 5 of 13 patients withHCCs larger than 6 cm. Angiography and Dopplersonography showed tumor thrombus within theportal veins of 2 patients with HCCs smaller than3 cm, in 1 patient with an HCC 3–6 cm, and in 9patients with HCCs larger than 6 cm. In HCCslarger than 6 cm, the PI was significantly lower (p< 0.05 by Mann-Whitney U test) for those withthrombus than for those without thrombus, andthe PSV tended to be higher (p 4 0.052 by Mann-Whitney U test) for HCCs with thrombus (Ta-ble 4).

DISCUSSION

To clarify the change in arterial hemodynamicsinduced by HCC, we analyzed Doppler velocitywaveforms of arteries supplying tumors using the

FIGURE 2. Duplex color Doppler sonogram shows right hepatic arterial branch (coded red) and its velocitywaveform.

HEPATIC ARTERIAL HEMODYNAMICS


PI and PSV. Compared with other indices, suchas the resistance index (RI), the PI has the ad-vantage that its calculated formula includes themean velocity, which reflects information aboutthe entire velocity waveform during 1 cardiaccycle.11,12 In contrast, the RI cannot distinguishbetween Doppler waveforms that differ in shapebut not in PSV and end-diastolic velocity. The PIwas initially used as an index of proximal arterialstenosis in the aortoiliac area.13,14 Later stud-ies15,16 showed that the PI was in good accordancewith the total peripheral resistance. However,when absolute PI values are interpreted in differ-ent patients, various individual conditions thataffect the PI, such as heart rate, distensibility of a

vessel wall, and, particularly, total peripheral re-sistance, must be considered. Our comparison be-tween the PIs for the right and the left hepaticarterial branches in the same patient was not af-fected by these limitations.

The Doppler sampling points were chosen atproximal sites of the hepatic arterial branchesthat were thought to supply 2 or 3 segments inthe left lobe and 2 segments in the right lobe. Atthose sites, signals were obtained with a smallDoppler angle and without interference from

FIGURE 3. Duplex color Doppler sonogram shows the tumor thrombus within the right portal vein and thedilated arteries along the thrombus in a patient with hepatocellular carcinoma. Pulsatile velocity waveformsflowing hepatofugally are identified in the thrombus.

TABLE 1

Pulsatility Index and Peak Systolic Velocity of the Right

and Left Hepatic Arterial Branches in Patients with Chronic

Liver Disease (Controls)

N LHA RHA p Value

Pulsatility index 34 1.29 ± 0.24 1.29 ± 0.26 NSPeak systolic

velocity (m/sec) 26 0.45 ± 0.11 0.44 ± 0.09 NS

Abbreviations: N, number of hepatic arterial branches examined;LHA, proximal site of hepatic arterial branch along the vertical seg-ment of the left portal vein; RHA, proximal site of hepatic arterialbranch along the right anterior segmental portal branch; NS, not sig-nificant.

TABLE 2

Pulsatility Index and Peak Systolic Velocity of Arteries

That Do and Do Not Supply Tumors of Various Sizes

Tumor Size/Parameter N

HA(HCC+)

HA(HCC−)

pValue

<3.0 cmPI 13 1.22 ± 0.23 1.23 ± 0.19 NSPSV (m/sec) 9 0.56 ± 0.16 0.51 ± 0.12 NS

3.0–6.0 cmPI 12 1.19 ± 0.32 1.41 ± 0.34 <0.05PSV (m/sec) 10 0.72 ± 0.26 0.47 ± 0.16 <0.05

>6.0 cmPI 13 0.73 ± 0.25 1.29 ± 0.32 <0.05PSV (m/sec) 12 1.45 ± 0.59 0.57 ± 0.28 <0.05

Abbreviations: N, number of hepatic arterial branches examined;HA (HCC+), hepatic arterial branches supplying the tumor; HA (HCC−),hepatic arterial branches not supplying the tumor; PI, pulsatility in-dex; NS, not significant; PSV, peak systolic velocity.

SOMEDA ET AL


bowel gas. In control patients with chronic liverdisease, we found no significant differences in thePIs and PSVs between the examined left andright segmental arteries in the same patient.Therefore, as shown in Table 2, HCC must cause

significant changes to the hepatic arterial hemo-dynamics that affect the PI and the PSV.

In our study, the arteries supplying the HCCshad lower PIs and higher PSVs than did those notsupplying the HCCs. Previous histopathologic

FIGURE 4. Comparison of Doppler velocity waveforms between the artery supplying the tumor and the one not supplying the tumor in a patientwith hepatocellular carcinoma (14 cm in diameter). The tumor was located mainly in the right anterior segment of the liver, with tumor thrombusin the right portal vein. (A) Duplex color Doppler sonogram shows the left hepatic arterial branch corresponding to the host artery not supplyingthe tumor in this patient. (B) The velocity waveform of the right arterial branch has a higher peak systolic velocity (1.9 versus 0.5 m/sec) and a lowerpulsatility index (0.49 versus 1.77) than that of the left arterial branch.



study of the tumor vessels in HCCs showed en-larged neovascular vessels, very thin walls, ascanty smooth muscle layer, and few elastic fiberscompared with normal hepatic arteries of thesame diameter.17 Therefore, the peripheral vas-cular bed of the hepatic arteries that includes theHCC has decreased vascular impedance and alower PI. High distensibility of the arteries sup-plying an HCC also may cause both impedanceand the PI to decrease.

As shown in Table 4, another major factor lead-ing to low peripheral impedance is tumor throm-bus, seen in our study in the 3rd or more proximalportal branches. Nakashima and Kojiro18 re-ported that a tumor thrombus tends to producearterioportal shunts by the following mecha-nisms: (1) blood flows from the periportal arterialbranches into the arterial vessels of the tumorthrombus and its blood spaces, which drain intothe portal lumen; or (2) dilated periportal arterialbranches are directly destroyed by the rapidly ex-panding tumor thrombus; this results in directcommunication between the interlobular arteryand the portal vein.18 The former phenomenoncorresponds to the thread-and streaks sign on an-giograms.8 In 5 of 9 patients in our study who hadHCCs larger than 6 cm with arterioportal shuntsand tumor thrombus, the dilated artery supplyingthe HCC had both an extremely high velocity(>1.4 m/sec) and a low PI (<0.9). In the remaining4 patients, low impedance waveforms like thosementioned above were found. The poor portal in-

flow caused by a thrombus results in dilatation ofthe normal arterial vessels.19

Earlier studies5,20 showed no correlation be-tween tumor size and amplitude of the Dopplersignals, but in our series, as tumor size increased,flow velocity increased, probably because the sig-nals we obtained were from arteries located apartfrom the mass. We considered our waveformanalysis of the host artery supplying the HCC torepresent precise characteristics of the entire pe-ripheral vascular bed, which included neovascu-lar and tumor-related alterations of the hepaticcirculation (eg, arterioportal shunts and tumorthrombus) as the tumor grew. In HCCs smallerthan 3 cm, we saw no significant difference be-tween the relatively proximal arterial branchessupplying the tumor and those not supplying it.However, we expect that when more peripheralbranches can be visualized and compared, a sig-nificant difference will be found between arteriesthat do and do not supply HCCs.

REFERENCES

1. Taylor KJW, Ramos I, Carter D, et al: Correlationof Doppler US tumor signals with neovascular mor-phologic features. Radiology 1988;166:57.

2. Ohnishi K, Nomura F: Ultrasonic Doppler studiesof hepatocellular carcinoma and comparison withother hepatic focal lesions. Gastroenterology 1989;97:1489.

3. Taylor KJW, Ramos I, Morse SS, et al: Focal livermasses: differential diagnosis with pulsed DopplerUS. Radiology 1987;164:643.

4. Yasuhara K, Kimura K, Ohto M, et al: PulsedDoppler in the diagnosis of small liver tumors. Br JRadiol 1988;61:898.

5. Tochio H, Minowa K, Tomita S, et al: Differentialdiagnosis of hepatic tumors using color Dopplerflow imaging: value of the Doppler spectral analy-sis. Japanese Journal of Medical Ultrasonics 1992;19:277.

6. Shimamoto K, Sakuma S, Ishigaki T, et al: Hepa-tocellular carcinoma: evaluation with color DopplerUS and MR imaging. Radiology 1992;182:149.

7. Numata K, Tanaka K, Mitsui K, et al: Flow char-acteristics of hepatic tumors at color Doppler so-nography: correlation with arteriographic findings.Am J Roentgenol 1993;160:515.

8. Okuda K, Musha H, Yoshida T, et al: Demonstra-tion of growing casts of hepatocellular carcinoma inthe portal vein by celiac angiography: the threadand streaks sign. Radiology 1975;117:303.

9. Wang LY, Lin ZY, Chang WY, et al: Duplex pulsedDoppler sonography of portal vein thrombosis inhepatocellular carcinoma. J Ultrasound Med 1991;10:265.

10. Tanaka K, Numata K, Okazaki H, et al: Diagnosis

TABLE 3

Sonographic and Angiographic Characteristics of HCCs of

Different Sizes

Tumor Size NLocation

(Left/Right)Diameter

(cm)AP

ShuntTumor

Thrombus*

<3.0 cm 13 4/9 2.3 ± 0.7 0 23.0–6.0 cm 12 4/8 4.7 ± 1.0 0 1>6.0 cm 13 2/11 12.9 ± 4.5 5 9

(5 withAP shunt)

Abbreviations: N, number of patients; AP, arterioportal.*Portal vein involvement by the HCC.

TABLE 4

Pulsatility Index and Peak Systolic Velocity of Hepatic

Arteries Supplying Large Tumors (>6 cm) with and

without Portal Vein Tumor Thrombus

TumorThrombus N

No TumorThrombus N

pValue

Pulsatility index 0.63 ± 0.23 9 0.94 ± 0.15 4 <0.05Peak systolic

velocity (m/sec) 1.64 ± 0.49 9 0.87 ± 0.48 3 0.052

Abbreviation: N, number of hepatic arterial branches examined.

SOMEDA ET AL


of portal vein thrombosis in patients with hepato-cellular carcinoma: efficacy of color Doppler sonog-raphy compared with angiography. Am J Roent-genol 1993;160:1279.

11. Gosling RG, Dunbar G, King DH, et al: The quan-titative analysis of occlusive peripheral arterialdisease by a non-intrusive ultrasonic technique.Angiology 1971;22:52.

12. Bolognesi M, Sacerdoti D, Merkel C, et al: Rela-tionship between portal blood flow measured byimage-directed Doppler ultrasonography and he-patic blood flow measured by indocyanine greenconstant infusion in patients with cirrhosis. J ClinUltrasound 1995;23:297.

13. Harris PL, Taylor LA, Cave FD, et al: The relation-ship between Doppler ultrasound assessment andangiography in occlusive arterial disease of thelower limbs. Surg Gynecol Obstet 1974;138:911.

14. Johnston KW, Kassam M, Cobbold RSC: Relation-ship between Doppler pulsatility index and directfemoral pressure measurements in the diagnosis ofaortoiliac occlusive disease. Ultrasound Med Biol1983;9:271.

15. Legarth J, Thorup E: Characteristics of Dopplerblood-velocity waveforms in a cardiovascular invitro model. II. The influence of peripheral resis-tance, perfusion pressure and blood flow. Scand JClin Lab Invest 1989;49:459.

16. Legarth J, Nolsoe C: Doppler blood velocity wave-forms and the relation to peripheral resistance inthe brachial artery. J Ultrasound Med 1990;9:499.

17. Suzuki M, Takahashi T, Sato T: Medial regressionand its functional significance in tumor-supplyinghost arteries: a morphometric study of hepatic ar-teries in human livers with hepatocellular carci-noma. Cancer 1987;59:444.

18. Nakashima T, Kojiro M: Tumor thrombus of thehepatic and portal veins. In Nakashima T, KojiroM, eds: Hepatocellular Carcinoma. Tokyo,Springer-Verlag, 1987, p. 117.

19. Lautt WW, Greenway CV: Conceptual review ofthe hepatic vascular bed. Hepatology 1987;7:952.

20. Reinhold C, Hammers L, Taylor CR, et al: Charac-terization of focal hepatic lesions with duplex so-nography: findings in 198 patients. Am J Roent-genol 1995;164:1131.



change in hepatic arterial hemodynamics induced by hepatocellular carcinoma detected with doppler...

Documents