Extrahepatic Collateral Artery Supply to the Tumor Thrombi of Hepatocellular Carcinoma Invading Inferior Vena Cava: The Prevalence and Determinant Factors

Materials and Methods 

Image Analysis 

Contrast medium–enhanced dynamic CT images of the liver obtained before chemoembolization and angiograms obtained in the chemoembolization session were retrospectively analyzed by two radiologists (I.J.L. and J.W.C.) in consensus.

The blood supply of IVC tumor thrombi was determined by angiographic findings of “thread and streaks” signs or hypervascular staining on selective angiography of hepatic arteries or extrahepatic collateral arteries (Fig 1) (14, 15). If angiograms of the extrahepatic collateral artery alone showed these findings, we considered that IVC tumor thrombi were exclusively supplied by extrahepatic collateral artery (Fig 2). Hepatic artery attenuation was also evaluated on angiography.

View Large Image Download to PowerPoint

Figure 1. Combined hepatic arterial and extrahepatic collateral supply for IVC tumor thrombi of HCC at initial presentation in a 48-year-old man. (a) On contrast medium–enhanced CT obtained in the arterial phase, the primary HCC shows a well enhancing tumor in the right posterior segment of the liver (arrowheads). (b) The tumor thrombi occupy more than half the IVC lumen and completely fill and distend the IVC lumen (arrow in the left image). The tumor thrombi also extend into the right atrium (arrow in the right image). (c) Celiac arteriography shows huge hypervascular tumor staining (arrowheads) and a thread-and-streaks sign (arrow) along the course of the right hepatic vein extending into the right atrium. In this patient, the left hepatic artery originates from the left gastric artery (not shown). (d) Right inferior phrenic arteriography shows primary tumor staining (arrowhead) and minor IVC tumor thrombi staining (arrow).

View Large Image Download to PowerPoint

Figure 2. Exclusive extrahepatic collateral supply of IVC tumor thrombi is seen in a 49-year-old man with a history of chemoembolization. (a) On contrast medium–enhanced dynamic CT at initial presentation, the primary tumor in the right posterior segment of the liver shows typical findings of HCC: strong enhancement at the arterial phase and rapid washout at the portal phase. There is no evidence of IVC invasion. (b) After two sessions of chemoembolization, follow-up CT shows compact iodized oil uptake at the primary HCC. However, tumor thrombi appeared in the IVC (arrow). The route of invasion is considered to be the inferior right hepatic vein. (c) Celiac arteriography performed at the third chemoembolization session does not show residual tumor staining. (d) Right adrenal arteriography shows hypervascular staining of IVC tumor thrombi (arrow) around the original mass with iodized oil uptake (arrowheads). Therefore, the IVC tumor thrombi are exclusively supplied by the right adrenal artery. (e) Radiograph obtained after chemoembolization through the right adrenal artery shows homogenous uptake of iodized oil in the tumor thrombi (arrowheads). (f) After 9 months, follow-up CT demonstrates shrinkage of the IVC tumor thrombi (arrow) with dense accumulation of iodized oil.

The items reviewed on the CT images included the tumor factors (size, number, and growth pattern), the pathway of IVC invasion, the distance from the primary tumor to IVC thrombi, portal vein invasion, extent of the IVC thrombi, and hypertrophy of the right inferior phrenic artery. The growth pattern of primary tumor was classified into diffuse and nodular patterns. The distance from primary tumor to IVC thrombi was classified into two categories: (i) direct contact between the primary tumor and IVC thrombi and (ii) propagation from the remote primary sites via hepatic veins. The extent of the IVC thrombi was assessed with use of the following classification according to whether the thrombi (i) occupied more than half of the IVC lumen on transverse CT image, (ii) completely filled and distended the IVC lumen, and (iii) extended into the right atrium. Hypertrophy of the right inferior phrenic artery was noted when the diameter of the right inferior phrenic artery was larger than that of the left inferior phrenic artery (17).

Results 

Fifty-four of the 82 patients (65.9%) had extrahepatic collateral artery supply to the IVC tumor thrombi: 47 from the right inferior phrenic artery (Fig 1), four from the right adrenal artery (Fig 2), two from the right internal mammary artery, and one from the right renal artery.

None of the 82 patients had proper hepatic artery occlusion; 20 showed hepatic artery attenuation on angiograms, and all these patients had a history of chemoembolization. On initial presentation, all patients had a patent hepatic artery without attenuation.

On CT imaging analysis, 47 of the 82 patients with IVC tumor thrombi had multiple tumors (57.3%), 24 had a diffuse growth pattern (29.3%), and 39 had portal vein invasion (47.6%). The primary tumors ranged from 3 cm to 18 cm in maximum diameter (mean, 8.9 cm ± 3.3). Thirty-three patients (40.2%) had distant metastases (to the lung in 27 patients, lymph nodes in five, adrenal glands in four, and bones in two). In 35 patients, the primary tumor was in direct contact with IVC thrombi without intervening thrombosed hepatic veins. In the other 47 patients (57.3%), the tumor thrombi of HCCs extended from the remote primary sites into the IVC through the drainage veins, including 22 right hepatic veins, eight inferior right hepatic veins, seven middle hepatic veins, six phrenic veins, one accessory hepatic vein, one left hepatic vein, one caudate vein, and one right renal vein. Fifty-four of the 82 IVC tumor thrombi (65.9%) occupied more than half the IVC lumen. Twenty-five thrombi (30.5%) caused luminal expansion of the IVC and 14 (17.1%) extended to the right atrium.

Table 1 summarizes the results of demographic, imaging, and statistical analyses in the two groups IVC tumor thrombi with or without extrahepatic collateral artery supply. On univariate analysis between the two groups, the P value was lower than .25 for the following items: age, Child-Pugh classification, history of chemoembolization, primary tumor size, distance between primary tumor and IVC, and extent of IVC thrombi (Table 1). On multiple logistic regression analysis, the presence of extrahepatic collateral artery supply to IVC tumor thrombi showed a significant relationship with history of chemoembolization (P = .001) and distension of the IVC by tumor thrombi (P = .005). In the patients who had undergone chemoembolization, the odds of the presence of extrahepatic collateral artery supply to IVC tumor thrombi was 22.4 times higher than in patients who had not undergone chemoembolization. When the IVC was completely filled and distended by tumor thrombi, the OR was 9.1 times higher than when it was only partially filled by tumor thrombi. A trend toward a greater incidence of occupancy of more than half of the IVC lumen by tumor thrombi was seen in the group with extrahepatic collateral artery supply to IVC tumor thrombi, but the P value did not meet our criteria of statistical significance (P = .079, OR = 4.71).

Table 1. Determinant Factors for Extrahepatic Collateral Supply to IVC Tumor Thrombi of HCC

	Variable	ExCA Supply for IVC Tumor Thrombi		Univariate P Value	Multiple Logistic Regression
		Present (n = 54)	Absent (n = 28)		P Value	OR	95% CI
	Sex
	Male	46(85.2)	24(85.7)	1.000
	Female	8(14.8)	4(14.3)
	Age (y)
	≥50	29(53.7)	20(71.4)	.156	.248	0.41	0.09–1.84
	<50	25(46.3)	8(28.6)
	Child-Pugh Class
	B/C	12(22.2)	2(7.1)	.123	.587	1.87	0.20–17.8
	A	42(77.8)	26(92.9)
	History of chemoembolization
	Repeated chemoembolization	32(59.3)	3(10.7)	<.001	.001	22.41	3.67–136.82
	Initial presentation	22(40.7)	25(89.3)
	Size of primary tumor (cm)
	≥10	19(35.2)	17(60.7)	.036	.164	0.35	0.08–1.54
	<10	35(64.8)	11(39.3)
	Primary tumor(s)
	Multiple	32(59.3)	15(53.6)	.645
	Single	22(40.7)	13(46.4)
	Growth pattern of primary tumor
	Diffuse	17(31.5)	7(25.0)	.615
	Nodular	37(68.5)	21(75.0)
	Distance between primary tumor and IVC
	Direct contact	27(50.0)	8(28.6)	.099	.447	1.76	0.41–7.57
	Separate	27(50.0)	20(71.4)
	Tumor thrombi in PV
	Present	28(51.9)	11(39.3)	.353
	Absent	26(48.1)	17(60.7)
	IVC tumor thrombi extent
	More than half of the IVC lumen	41(75.9)	13(46.4)	.013	.079	4.71	0.834–26.58
	Less than half of the IVC lumen	13(24.1)	15(53.6)
	Distension of the IVC by tumor thrombi
	Yes	22(40.7)	3(10.7)	.005	.012	9.05	1.61–50.91
	No	32(59.3)	25(89.3)
	Involvement of right atrium
	Yes	10(18.5)	4(14.3)	.762
	No	44(81.5)	24(85.7)
	Size of right inferior phrenic artery
	Hypertrophic	35(64.8)	18(64.3)	1.000
	Equal to the left	19(35.2)	10(35.7)

Note.—Values in parentheses are percentages. ExCA = extrahepatic collateral artery.

In the 54 patients with extrahepatic collateral artery supply, 32 had a history of chemoembolization and 22 patients were at initial presentation. We further investigated the influence of previous repeated chemoembolization sessions on the pattern of extrahepatic collateral artery supply to IVC tumor thrombi. Table 2 summarizes the results and demonstrates that, in patients who developed IVC invasion during repeated chemoembolization sessions, extrahepatic collateral artery supply appeared at an early stage of IVC invasion and that IVC tumor thrombi can be exclusively supplied by the extrahepatic collateral artery (Fig 2). Thirteen of 32 patients who had undergone repeated chemoembolization sessions (40.6%) showed tumor thrombi occupation of less than half of the IVC lumen, whereas all patients at initial presentation showed tumor thrombi occupation of more than half of the IVC lumen (P < .001). Seven of 32 patients who had undergone repeated chemoembolization sessions (21.9%) showed distension of the IVC, compared with 15 of 22 patients at initial presentation (68.2%; P = .002). All patients at initial presentation had additional blood supply from hepatic arteries. However, 12 of 33 patients who had undergone repeated chemoembolization sessions had exclusive extrahepatic collateral artery supply (P = .001).

Table 2. Influence of History of Chemoembolization on the Prevalence and Pattern of Extrahepatic Collateral Supply to IVC Tumor Thrombi Stratified by Extent

		ExCA Supply (n = 54)			No ExCA Supply (n = 28)
	Variable	Initial Presentation (n = 22)	History of Chemoembolization (n = 32)	P Value	Initial Presentation (n = 25)	History of Chemoembolization (n = 3)	P Value
	IVC tumor thrombi extent
	More than half of IVC lumen	22(100.0)	19(59.4)	<.001	11(44.0)	2(66.7)	.457
	Less than half of IVC lumen	0	13(40.6)		14(56.0)	1(33.3)
	IVC distension by tumor thrombi
	Yes	15(68.2)	7(21.9)	.002	3(12.0)	0	.525
	No	7(31.8)	25(78.1)		22(88.0)	3(100.0)
	Supply
	Combined from hepatic artery	22(100.0)	20(62.5)	.001	—	—	—
	Exclusive from ExCA	0	12(37.5)

Note.—Values in parentheses are percentages. ExCA = extrahepatic collateral artery.

In the 28 patients without extrahepatic collateral artery supply at the chemoembolization session just after the initial CT diagnosis of IVC invasion, 18 patients underwent several sessions of follow-up chemoembolization (mean follow-up interval, 14.5 months; range, 3–50 months). Thirteen of the 18 patients (72.2%) showed progression of IVC tumor thrombi, and all developed extrahepatic collateral artery supply from the right inferior phrenic artery. In four patients, IVC tumor thrombi regressed after chemoembolization via hepatic arteries, and all had negative results on angiographic investigation of extrahepatic collateral arteries. The remaining patient showed no interval change in the extent of IVC tumor thrombi and also had negative results on angiographic investigation of extrahepatic collateral artery supply to IVC tumor thrombi.

The follow-up period ranged from 1 to 125 months (mean, 12.9 months; median, 6 months). In August 2008, 10 patients were alive and being followed up. Overall cumulative survival rates, calculated from times of the detection of IVC tumor thrombi, were 62.1% at 6 months, 35.8% at 1 year, 24.3% at 2 years, and 17.2% at 3 years (Fig 3).

View Large Image Download to PowerPoint

Figure 3. Overall cumulative survival rates calculated from time of detection of IVC tumor thrombi.

Discussion 

HCCs commonly invade intrahepatic vasculatures such as the portal and hepatic veins (13, 14). As tumor thrombi in hepatic veins grow, they could extend to the IVC. Therefore, we believe IVC invasion of HCC should be regarded as a severe form of vascular invasion. According to our experience, its prevalence was 3.8% during the course of chemoembolization.

Chemoembolization for advanced HCC with IVC invasion is still considered a contraindication at many institutions. However, in a recent review of 26 cases, Chern et al (18) reported that chemoembolization is a safe and effective treatment for advanced HCC with IVC invasion. Based on our experience, we agree with their opinion. To continue chemoembolization in these patients for the effective control of IVC tumor thrombi, it is essential to understand their blood supply, including extrahepatic artery supply. Until now, to our knowledge, the pattern of blood supply of IVC tumor thrombi has not been investigated. This study demonstrated that the angiographic prevalence of extrahepatic collateral artery supply to IVC tumor thrombi was 65.9% at initial diagnosis and higher during sequential follow-up studies. Among the 28 patients without extrahepatic collateral artery supply at initial diagnosis, 18 underwent subsequent chemoembolization and 13 of these 18 (72.2%) developed extrahepatic collateral artery supply with the progression of IVC tumor thrombi.

Extrahepatic collateral supply to HCC develops in the advanced stages with a patent hepatic artery as well as with hepatic arterial injury by surgical ligation or repeated chemoembolization (5, 6, 7, 8, 9, 10, 11, 12). According to a prospective study on the prevalence of extrahepatic collateral artery supply to HCC (7), 17% of primary parenchymal tumors had extrahepatic collateral artery supply at initial presentation and tumor size was the most important determining factor for extrahepatic collateral artery supply. The results of the present study also demonstrate that the major determining factors for extrahepatic collateral artery supply to IVC tumor thrombi are a history of chemoembolization and the extent of IVC tumor thrombi. Extrahepatic collateral supply to IVC tumor thrombi was determined by the extent of invasion. Among the 47 patients with IVC tumor thrombi on initial presentation with HCC, 22 (46.8%) had extrahepatic collateral artery supply to IVC tumor thrombi. Among the 22 patients with extrahepatic collateral artery supply to IVC tumor thrombi at initial presentation of HCC, all 22 had IVC tumor thrombi that occupied more than half the IVC lumen (Table 2). None of the patients with IVC tumor thrombi occupying less than half the IVC lumen had obvious extrahepatic collateral artery supply.

Considering that repeated chemoembolization associated with the obliteration of peripheral hepatic arteries is a well known cause of extrahepatic collateral artery supply to primary HCCs (6, 7), it can be readily expected that a history of chemoembolization predisposes IVC tumor thrombi to the development of collateral supply. In our study, although no patients had proper hepatic artery occlusion, 20 patients had hepatic artery attenuation on angiography, which suggested hepatic artery injury, and all of these patients had a history of chemoembolization. However, to avoid multiple collinearity on multivariate analysis, hepatic artery attenuation was not included the analysis of determining factors. In the present study, 32 of the 35 patients with a history of chemoembolization at the diagnosis of IVC invasion (91.4%) had extrahepatic collateral artery supply to IVC tumor thrombi (Table 2). In addition, patients with a history of repeated chemoembolization developed extrahepatic collateral artery supply to IVC tumor thrombi at an early stage of IVC invasion occupying less than half the IVC lumen. Then, the IVC tumor thrombi that developed during repeated chemoembolization were frequently supplied by extrahepatic collateral arteries alone (Fig 2).

The most common extrahepatic collateral artery was the right inferior phrenic artery. The right inferior phrenic artery supplies most of the right hemidiaphragm, including the area in contact with the bare area of the liver (ie, subphrenic area or caudate lobe) (5, 6, 7). Considering that the IVC is located around the caudate lobe and penetrates the diaphragm, it is not surprising that IVC tumor thrombi are almost always supplied by a feeder vessel from the right inferior phrenic artery except in exceptional situations, such as IVC invasion via the inferior right hepatic veins or obliteration of the right inferior phrenic artery by previous chemoembolization. When HCC invades the IVC via the inferior right hepatic vein, the adrenal arteries from the aorta and the right renal artery should be regarded as potential feeder vessels for the thrombi. When the right inferior phrenic artery is obliterated by previous chemoembolization, adjacent collateral arteries including the right internal mammary artery, left inferior phrenic artery, and intercostal arteries can supply the IVC tumor thrombi.

There are several limitations to our study. Because of the retrospective nature of this study, 87 patients with IVC invasion were excluded because angiographic investigation for extrahepatic collateral arteries was not performed. Therefore, the overall prevalence of extrahepatic collateral artery supply to IVC tumor thrombi may be overestimated or underestimated. However, in our routine clinical practice of chemoembolization, angiographic investigation for extrahepatic collateral arteries was usually performed when the hepatic functional reserve was sufficient and aggressive complete chemoembolization of extrahepatic collateral arteries was considered to be safe and beneficial. That is, most patients excluded from this study had poor hepatic functional reserve and extensive tumors. In addition, in the majority of patients without extrahepatic collateral artery supply to IVC tumor thrombi at initial diagnosis, the IVC tumor thrombi eventually progressed and subsequently developed extrahepatic collateral artery supply. Considering the high percentage of tumor progression after hepatic arterial chemoembolization, there is a possibility that extrahepatic collateral artery supply to IVC tumor thrombi was missed at initial diagnosis because it was too small to be detected angiographically. Therefore, we believe the overall prevalence of extrahepatic collateral arteries for IVC tumor thrombi in this study may be underestimated. However, the angiographic prevalence stratified by the extent of IVC tumor thrombi may not be affected by the retrospective nature of this study.

In conclusion, IVC tumor thrombi of HCCs are frequently supplied by extrahepatic collateral arteries, the most common of which is the right inferior phrenic artery. The determining factors are a history of chemoembolization and the extent of IVC tumor thrombi.