Overview

Primary and secondary liver tumors cause tremendous neoplastic morbidity and mortality. Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third most common cause of cancer-related death in the world (Bosch et al, 2005). The highest incidences of HCC are found in Eastern Asia and sub-Saharan Africa, where hepatitis B is endemic; however, the incidence of HCC has been increasing steadily in the United States and Western Europe, largely owing to increased prevalence of viral hepatitis C in the past 2 decades (El-Serag et al, 2003). The majority of metastatic cancers involve the liver at some point as disease progresses. Despite the presence of extrahepatic spread, many patients with liver involvement die as a consequence of local tumor growth and hepatic parenchymal destruction.

Although surgery remains the best treatment in patients with most primary and secondary liver tumors, less than one third of patients are candidates for surgery when the diseases are initially found. This may be due to the extent or distribution of disease, underlying liver function, or general condition of the patient. Therefore, in the absence of effective systemic therapy, much effort has been put into developing and testing liver-directed therapies for local tumor control. Hepatic artery embolotherapy (HAE) has been outstanding among these methods as the most commonly used procedure to palliate symptoms and prolong survival.

Basic Principles of Hepatic Arterial Embolotherapy

Chemoembolization

In the past 3 decades, many papers have been published describing a myriad different techniques for performing HAE; combined with intraarterial chemotherapy (chemoembolization), HAE has been paramount among them as the most widely used procedure. The goal of chemoembolization is to combine the effects of targeted tumor ischemia by embolization with intraarterial chemotherapy. To date, the most common sole-agent chemotherapeutic drug has been doxorubicin, followed by cisplatin, epirubicin, mitoxantrone, mitomycin C, and a chemical conjugate of a synthetic copolymer of styrene maleic acid and neocarcinostatin (SMANCS) (Marelli et al, 2007). The chemotherapeutic drug is dissolved in water or water-soluble contrast agent and is then mixed with iodized oil (Lipiodol; Laboratoire Guerbet, France) and administered as an emulsion (Fig. 83.1). When injected into the hepatic artery, iodized oil is trapped selectively in the tumor because of the hemodynamic difference between the tumor and normal liver parenchyma and presumably the absence of Kupffer cells in the tumor (Kan et al, 1993; Nakajo et al, 1988). Thus mixing iodized oil with a chemotherapeutic drug allows slow release of the chemotherapeutic drug over a period of 6 to 12 weeks (Raoul et al, 1992). In contrast, in the normal liver parenchyma, the iodized oil does not occlude the hepatic artery; rather it accumulates in the peripheral portal vein through arterioportal communications and subsequently passes through sinusoids into the systemic circulation (Chung et al, 1993). HAE after infusion of chemotherapeutic drugs increases drug dwelling time in the tumor by slowing the rate of efflux from the hepatic circulation. Furthermore, ischemic damage by embolization potentiates absorption of chemotherapeutic drugs by impeding the function of transmembrane pumps in tumor cells (Yu & Keeffe, 2003).

FIGURE 83.1 Preparation of a mixture of iodized oil and doxorubicin hydrochloride. Every 10 mg of doxorubicin hydrochloride was dissolved in 0.5 mL of the water-soluble contrast medium; iodized oil and dissolved doxorubicin hydrochloride were drawn separately into syringes interconnected with a three-way stopcock (A) and emulsified by means of vigorous pushing of each syringe in alternation (B). C, Light photomicrograph shows the formation of oil in water in oil-type emulsion with variably sized (10 to 50 µm) water droplets containing doxorubicin hydrochloride in the oil base.

After infusion of the chemotherapeutic drug and iodized oil mixture, tumor-feeding peripheral hepatic arteries are embolized. Proximal arterial occlusion is not desirable, because it will not only induce development of intrahepatic and extrahepatic collateral vessels, it will also preclude repeat procedure. Thus it is important to select embolic materials of the proper size. The optimal size should be small enough to reach and occlude the capillaries to the tumor but bigger than the arteriovenous shunt and peribiliary plexus, to avoid the risk of pulmonary embolization and bile duct necrosis. To date, gelatin sponge particles have been the most frequently used agent, but polyvinyl alcohol (PVA) particle (Ramsey et al, 2002), absolute ethanol (Matsui et al, 1993), starch microspheres (Maluccio et al, 2008), cyanoacrylate (Berghammer et al, 1998), and even autologous blood clot (Gunji et al, 2002) have also been used.

Gelfoam is used as 500- to 1000-µm cubes, which only occludes the artery temporarily, with recanalization taking place within 2 weeks (Coldwell et al, 1994). It was proved not to cause serious hepatic damage in patients with good hepatic functional reserve (Kan et al, 1993; Sonomura et al, 1997); however, gelfoam powder should not be used, as it may cause biliary stenosis and biloma resulting from embolization and necrosis of small arteriolar branches, which may cause biliary damage (Makuuchi et al, 1985). PVA particles cause a permanent or semipermanent arterial occlusion and can achieve more distal obstruction because of their smaller size (50 to 250 µm in diameter; Coldwell et al, 1994). Autologous blood clot achieves the same temporary artery occlusion as a gelatin sponge. Because the clot is lysed faster after embolization, there might be less chance of arterial thrombosis after several sessions of chemoembolization (Gunji et al, 2002). In a recent comparative nonrandomized study, no difference in patient survival was found between chemoembolization using gelatin sponge particles and that using PVA particles (Brown et al, 2005).

It is important to evaluate tumor-related arteriovenous shunt for safe procedures. In patients with HCC, angiographic incidence of shunting has been reported in 31.2%, shunting into the portal vein has been reported in 28.8%, and hepatic vein shunting has been reported in 2.4% (Ngan & Peh, 1997). The more liver parenchyma is occupied by the tumor, the more arteriovenous shunts exist; these are associated with poor prognosis (Granov et al, 1992) and are frequently problematic in chemoembolization of liver tumors (Izaki et al, 2004). Severe arterioportal shunt can cause hepatofugal portal flow, ascites, and variceal bleeding. In patients with a prominent arteriovenous shunt, embolization of the shunt is recommended prior to HAE (Huang et al, 2004). After embolization for massive arterioportal shunt, hepatofugal portal flow may convert to hepatopetal flow, and performance status and ascites may be improved (Izaki et al, 2004).

Patient Selection

Chemoembolization is indicated for the treatment of unresectable hypervascular liver tumors, which receive more blood supply from the hepatic artery than normal liver parenchyma, which makes the tumor suitable to HAE. The hemodynamic characteristics of a tumor may be documented on contrast-enhanced ultrasonography (Suzuki et al, 2003), computed tomography (CT) (Tajima et al, 2002b), and magnetic resonance imaging (MRI) (Levy, 2002).

The most common indication of chemoembolization is unresectable HCC. The determination of resectability of HCC should be based on either the extent of tumor involvement, underlying liver function, or a combination of these (see Chapters 2 and 90A). The majority of patients with HCC have underlying liver cirrhosis, which often requires a larger remnant liver after surgery to maintain hepatic function than in patients without underlying liver disease; therefore tumors that might be resectable in patients with normal liver parenchyma may not be resectable in patients with cirrhosis. Hepatic dysfunction also acts as a limitation in selecting patients for chemoembolization. Patients with poor hepatic function are not likely to tolerate extensive arterial embolization, because their livers are more dependent on arterial blood supply than normal liver, and moreover, patients with severe cirrhosis are more likely to die of their underlying liver disease than of HCC. Thus chemoembolization is typically recommended in patients with reasonably preserved liver function (Child-Turcotte-Pugh [CTP] class of A or B) and performance status (Eastern Cooperative Oncology Group [ECOG] 0 or 1). In general, patients with class C cirrhosis should not undergo chemoembolization as a primary treatment.

The extent of tumor is another important consideration in selecting patients for chemoembolization. Several methods have been proposed to weigh both hepatic function and the extent of tumor. In Okuda staging, patients are given points for presence of ascites, serum bilirubin and albumin levels, and the extent of tumor, and the patient is staged as I, II, or III (Okuda et al, 1985). The Cancer of the Liver Italian Program (CLIP) group proposed a more complex scoring system that includes the CTP score, tumor morphology and extension, presence of portal vein thrombosis, and serum α-fetoprotein (AFP) levels. The CLIP score was validated worldwide in multiple clinical trials and was proven to give more accurate prognostic information and greater survival predictive power than the Okuda staging system (CLIP, 2000; Ueno et al, 2001); however, those staging systems for predicting prognosis of patients with HCC did not indicate which patients would benefit from HAE. Recently, the Barcelona Clinic Liver Cancer (BCLC) staging system was developed to allow for the indication of the best therapy for each stage of HCC (Llovet et al, 1999a). In this staging system, chemoembolization is recommended as first-line therapy for intermediate stage HCC (Okuda staging 1 to 2, performance score 0, and large or multinodular disease) and some advanced-stage HCC (Okuda 1 to 2, performance score 1, no extrahepatic disease).

Spontaneous rupture of HCC is an indication for emergency HAE, regardless of underlying liver function (Kirikoshi et al, 2009). Even in patients with advanced liver cirrhosis, nodular HCC showing exophytic growth can be managed safely by selective embolization to prevent tumor rupture without diminishing liver function. Chemoembolization plays additional neoadjuvant roles as a downstaging therapy before resection or as a bridge therapy for patients awaiting liver transplantation (see Chapter 97D).

In general, the selection criteria for patients with HCC apply to patients with metastatic liver tumors; patients with unresectable hypervascular tumors will benefit from HAE. Typical hypervascular metastatic liver tumors include neuroendocrine tumors (NETs), gastrointestinal stromal tumors (GISTs), ocular melanoma, islet cell tumors, carcinoids, and sarcomas. The goal of HAE is symptomatic relief and improvement of survival. Most common indications are rapid progression of liver tumor with stable or absent extrahepatic disease, symptoms related to tumor bulk, and symptoms related to hormonal excess, especially in metastasis from neuroendocrine tumors.

Although no absolute contraindications to HAE exist, patients with two or three major poor prognostic indicators should not be treated. Generally accepted relative contraindications are extensive tumor involvement of more than 50% to 75% of the liver and class C cirrhosis. Major portal vein invasion by the tumor has been considered another relative contraindication, but this can be safely and effectively managed by an adjustment of the embolization protocol to reduce the amount of chemoembolic agents and the extent of the embolization, especially in patients with a limited parenchymal tumor and adequate liver function (Chung et al, 1995). Recent evidence demonstrates that HAE can be safely performed, even in patients with main portal vein occlusion, if hepatopetal collateral flow is present (Lee et al, 1997; Tazawa et al, 2001; Georgiades et al, 2005). In addition, active gastrointestinal bleeding, refractory ascites, extrahepatic spread, hepatic encephalopathy, and biliary obstruction are also considered relative contraindications. Regardless of the liver tumors, anaphylactoid reaction to contrast media or renal insufficiency, uncorrectable coagulopathy, and severe peripheral vascular disease can preclude embolotherapy.

Procedure

Before any chemoembolization procedure, laboratory tests should be done that include a complete blood cell count, prothrombin time, creatinine levels, and liver function testing. The baseline serum levels of tumor markers should be measured before treatment to monitor changes after treatment. By means of cross-sectional imaging studies, the size and extent of the tumor, its growth pattern (expansile vs. replacing or infiltrating), and macroscopic angioinvasion into the hepatic or portal veins are evaluated. For successful chemoembolization, accurate segmental localization of the tumor is crucial, and cross-sectional images are useful for that purpose (Yoon et al, 2008). In addition, imaging studies of the chest, abdomen, and pelvis are recommended to assess comorbid disease and ensure the presence or absence of metastatic disease.

Patients are fasted overnight and hydrated with normal saline at a rate of 200 to 300 mL/h. Antiemetics are administered intravenously, and patients with contrast allergies receive oral prednisone 1 hour before the procedure. The use of prophylactic antibiotics is controversial except in patients with a bilioenteric anastomosis or biliary stent (Geschwind et al, 2002; Ong et al, 2004; Tarazov, 1995). After infiltration of local anesthetic, the Seldinger technique is used to gain access to the common femoral artery, and initial diagnostic visceral arteriography is performed to determine arterial anatomy to the liver and patency of the portal vein. For complete angiography, all hepatic arteries should be adequately opacified, and all feeding arteries of the tumor should be identified.

Anatomic variations of celiac trunk and hepatic arteries are commonly encountered. With the introduction of multidetector row CT, these anatomic variations can be predicted before the procedure by careful review of the arterial phase scan (Song et al, 2010). Most common hepatic artery variations are the right hepatic artery arising from the superior mesenteric artery and the left hepatic artery arising from the left gastric artery, thus celiac and superior mesenteric arteriography is mandatory to identify the variations. Frequently, additional views in different angles and magnifications and selective contrast injection with use of a microcatheter are necessary to identify small feeding arteries (Yoon et al, 2008). To avoid nontargeted embolization, it is important to recognize the origin of the cystic artery and the right gastric artery, presence of a large falciform artery, and the accessory left gastric artery originating from the left hepatic artery (Song et al, 2006).

The treatment protocol should be individualized according to the hepatic functional reserve, extent of the tumor, and major portal vein invasion. Every effort should be made to preserve nontumorous liver parenchyma from ischemic damage by embolization. The best way to maximize treatment effect and to minimize procedure-related complications is to perform embolization of all tumor feeding arteries as selectively as possible. Patients with focal lesions should undergo segmental or, ideally, subsegmental embolization, in which a catheter or microcatheter is placed more selectively into the feeding artery in closer proximity to the tumor.

After the catheter is positioned appropriately, the mixture of iodized oil and chemotherapeutic drug is injected (Figs. 83.2 and 83.3). If there is a prominent arteriovenous shunt that precludes chemoembolic agents reaching the tumor vasculature, it is recommended first to embolize the shunt before injection of chemotherapeutic drugs (Fig. 83.4). The dose of iodized oil and chemotherapeutic agent depends on the size and vascularity of the tumor. The generally accepted upper limit of iodized oil is 15 mL. For a small tumor, enough iodized oil to saturate the entire tumor neovasculature could be administered. The end point for the mixture administration is stasis in tumor-feeding arteries or appearance of iodized oil in portal vein branches (see Fig. 83.3). If flow is a persistent in the tumor-feeding arteries after infusion of chemotherapeutic drug, HAE should be performed. Gelatin sponge particles are the embolic agents most frequently used, because they degrade in a few weeks, allowing repeat treatment. PVA particle is another commonly used embolic agent, which causes a more permanent arterial occlusion (Ramsey et al, 2002).

FIGURE 83.2 The concept of subsegmental or segmental chemoembolization using iodized oil. A, Exclusive arterial supply for encapsulated nodular hepatocellular carcinoma (HCC) and mixed arterial and portal venous supply for the portion of extracapsular invasion and small HCC without capsule formation. B, If a sufficient amount of a mixture of iodized oil and anticancer drug is injected through a tumor-feeding artery, not only the tumor neovasculature but also the peripheral portal veins around the tumor are filled with the emulsion. Subsequent hepatic artery embolization (arrow) may result in the effect of combined arterial and portal blockage; tumor fraction with mixed arterial and portal venous supply can be treated effectively by the combined effect of high-dose chemotherapy and ischemia.

FIGURE 83.3 Subsegmental chemoembolization in a 43-year-old man with multinodular hepatocellular carcinoma. A, Celiac arteriogram shows two hypervascular tumor nodules (arrows) in the liver. B, The larger mass is supplied by a feeding artery (arrow) from the right posterior segmental branch. C, Subsegmental chemoembolization was performed by selective catheterization of the tumor-feeding artery and injection of an emulsion of 5 mL of iodized oil and 30 mg of doxorubicin hydrochloride. Gelfoam embolization was not performed because sufficient stagnation of blood flow was achieved after injecting the emulsion. Note the filling with iodized oil of the peripheral portal branch (arrow) accompanying the tumor-feeding artery. D, Subsegmental chemoembolization also was performed for the tumor located in hepatic segment IV by selective catheterization of the tumor-feeding artery (arrow). E and F, Iodized oil computed tomography (CT) reveals compact retention of iodized oil in the nodules. G and H, Follow-up CT scan 2 years later shows shrunken tumors without evidence of any viable portion.

FIGURE 83.4 Severe arteriovenous shunt into the middle hepatic vein managed by chemoembolization after balloon occlusion of the middle hepatic vein in a 66-year-old man with hepatocellular carcinoma. A, Celiac arteriogram shows huge hypervascular mass with severe arteriovenous shunt into the middle hepatic vein (arrow). B, After balloon occlusion of the middle hepatic vein (arrow), celiac arteriography was repeated and confirmed the disappearance of the shunt. With the balloon inflated, chemoembolization was performed with a mixture of 10 mL of iodized oil and 50 mg of doxorubicin hydrochloride followed by Gelfoam embolization. C, Completion radiograph with the balloon deflated shows successful accumulation of iodized oil in the mass without untoward pulmonary embolization. D, Three months later, follow-up angiogram shows remarkably decreased arteriovenous shunt and the tumor fraction supplied by the hepatic arteries. An overgrowth of the tumor fraction is supplied by omental arteries (arrows).

Embolization of extrahepatic collaterals supplying the tumors is crucial to achieve successful outcomes. When a tumor is adjacent to a hepatic bare area or suspensory ligaments, or it invades into an adjacent organ, selective arteriogram of possible extrahepatic arteries should be obtained. With recent advances in CT imaging, those collateral vessels can be detected in preprocedure imaging studies (Kim et al, 2005). Common extrahepatic collaterals include the inferior phrenic artery, omental artery, internal mammary artery, colic branch of superior mesenteric artery, adrenal artery, intercostal artery, renal capsular artery, and gastric arteries (Chung et al, 2006; Kim et al, 2005, 2007, 2009; Won et al, 2003; Figs. 83.5 and 83.6). When the hepatic artery and extrahepatic collaterals supply the tumor, additional chemoembolization of the extrahepatic collaterals can be tried to increase the therapeutic efficacy of chemoembolization. These extrahepatic collaterals can be used as an access route to the tumors in patients with hepatic artery occlusion.

FIGURE 83.5 Massive hepatocellular carcinoma supplied by multiple extrahepatic collaterals in a 43-year-old man. A, Arterial phase computed tomographic (CT) scan shows a huge hypervascular mass in the right hepatic lobe (arrows). B, Celiac arteriogram shows a well-demarcated hypervascular mass with multinodular confluent appearance. C and D, Selective angiograms of the renal capsular artery (C) and the right inferior phrenic artery (D) also show multifocal tumor stain. Chemoembolization was performed with a mixture of 20 mL of iodized oil and 100 mg of doxorubicin hydrochloride followed by Gelfoam embolization at the right hepatic artery and extrahepatic collaterals. E, Postembolization plain radiograph shows diffuse uptake of iodized oil in the tumor. F, Follow-up CT scan shows a markedly shrunken primary tumor with a thick capsule, persistent iodized oil retention, and well-preserved, nontumorous hepatic parenchyma.

FIGURE 83.6 Hepatocellular carcinoma supplied by the right internal mammary artery as an extrahepatic collateral at its initial presentation in a 34-year-old woman. A, Initial computed tomographic (CT) scan shows a large tumor abutting the anterosuperior aspect of the diaphragm (arrows). B, Celiac arteriogram shows a large hypervascular mass occupying the medial segment of the left hepatic lobe and extending to the adjacent segments (white arrows). A defect in tumor stain appears at the superior aspect of the tumor (black arrows), suggesting the presence of an extrahepatic collateral supply. C, Right internal mammary arteriogram shows tumor stain at the area exactly matched with the defect in tumor stain on the celiac arteriogram. Chemoembolization was performed with a mixture of 15 mL of iodized oil and 50 mg of doxorubicin hydrochloride followed by Gelfoam embolization at the right hepatic artery and the diaphragmatic branch of the right internal mammary artery (arrow). D, Nine-year follow-up CT scans show a markedly shrunken tumor with no evidence of enhancing viable tumor.

Recently, the use of C-arm cone-beam CT has been increasing. It can provide information about a patient’s vascular anatomy, associated liver parenchyma, and the target lesion that constitutes a substantial improvement over conventional digital-subtraction angiography (DSA) and fluoroscopy (Carlson et al, 2001). C-arm cone-beam CT enables more selective catheterizations, which increases the amount of therapeutic agent delivered to the target area and decreases nontumorous liver parenchymal exposure to the agent (Meyer et al, 2007; Wallace et al, 2008; Fig. 83.7) In addition, confident identification of extrahepatic arteries supplying nontarget organs, such as supraduodenal or retroduodenal arteries, right gastric artery, phrenic artery, and falciform artery, is crucial to avoid nontarget embolization (Kim et al, 2009; Liu et al, 2005). When feeding arteries for a tumor are multiple, the portion of the tumor supplied by a catheterized feeding artery can be estimated, and the chemoembolic regimen can be proportioned accordingly. Iodized oil preferentially accumulated within tumors can be imaged without use of additional contrast media; thus completeness of therapy can be assessed immediately after treatment. If the oil has not accumulated in a portion of the tumor, the operator can search for hepatic or extrahepatic collateral tumor supply (Meyer et al, 2007; Wallace et al, 2007).

FIGURE 83.7 Application of cone-beam computed tomography (CT) for chemoembolization of hepatocellular carcinoma in a 49-year-old man. A, Common hepatic arteriography shows a huge hypervascular mass in the right hepatic lobe supplied by multiple feeding vessels. B, Three-dimensional volume-rendered image from cone-beam CT clearly demonstrates tumor feeders, including multiple, small, central feeding vessels (arrows). Real-time interactive anatomic information provided by cone-beam CT helps achieve successful superselective catheterization of tumor feeders and complete chemoembolization.

Intraarterial lidocaine and intravenous narcotic analgesics are given before the procedure to relieve pain during and after (Lee et al, 2001b). After the procedure, hydration is continued with normal saline. Narcotic analgesics, chlorpromazine, and acetaminophen are additionally supplied for 1 to 3 days as needed. Patients can be discharged once they have resumed adequate oral intake, and intravenous analgesics are no longer required.

Laboratory studies of liver function and serum levels of tumor markers should be repeated 2 to 4 weeks after the procedure. Multiphase contrast-enhanced CT or MRI is recommended 4 to 8 weeks after the procedure to evaluate the efficacy of the treatment and to detect local or remote tumor recurrences. If viable tumor is evident on imaging studies, serum levels of tumor markers are elevated, or both, patients return for another session of embolotherapy according to their hepatic functional recovery. No consensus has been reached on the ideal protocol for repeat procedures, but based on a retrospective comparison, survival rates and patient tolerance seemed to be higher when treatment was repeated, but only when progression of HCC was noted radiographically (Ernst et al, 1999).

Therapeutic Efficacy of Chemoembolization