• hepatocellular carcinoma;
  • recurrence;
  • embolization;
  • hepatectomy


  1. Top of page
  2. Abstract


Complete surgical resection is the mainstay of treatment for patients with hepatocellular carcinoma (HCC). Unfortunately, most patients ultimately develop disease recurrence and the median survival from the time of recurrence is <1 year. The purpose of the current study was to review the authors' experience using bland hepatic arterial embolization to treat recurrent HCC after definitive surgical resection.


The authors reviewed their single-center hepatic embolization database from 1995 through 2004 to identify patients who underwent bland hepatic arterial embolization for disease recurrence. Data analyzed included patient demographics, Okuda stage and Child score, imaging findings, and embolization variables. Recurrence-free survival (from surgery to disease recurrence) and survival time (from recurrence to last follow-up) were calculated using the Kaplan-Meier method.


The authors identified 45 patients treated with bland embolization for recurrent HCC after resection. Six patients also underwent ablative therapy after embolization. Of the 45 patients, 42 (93.3%) patients had Okuda Stage 1 disease. The median time to recurrence was 13 months. The median survival after embolization was 46 months, and actuarial survival rates at 1 year, 2 years, and 5 years after recurrence were 86%, 74%, and 47%, respectively, with a median follow-up of 31 months. Patients who developed disease recurrence with a solitary lesion had a significantly improved survival (P = .03) At the time of last follow-up, 3 patients (6.6%) were alive with no evidence of viable disease.


Bland arterial embolization was found to be an effective method of salvage therapy for patients with good liver function with recurrent HCC after prior surgical resection. Patients whose disease recurred with a solitary lesion appear to have a significantly increased survival compared with patients who develop disease recurrence with multiple tumors. A small proportion of patients can be rendered without evidence of viable disease. Cancer 2006. © 2006 American Cancer Society.

Over the past decade, there has been a significant increase in the incidence of hepatocellular carcinoma (HCC) within the U.S. Recent population-based studies have demonstrated that the incidence rate continues to approximate the death rate, suggesting that the vast majority of patients who develop this disease die of it.1–4 Liver transplantation is associated with 5-year survival rates of up to 70%,5 but only a fraction of patients are candidates according to the Milan selection criteria and donor supply is limited. Surgical resection is more broadly applicable than transplantation, and provides patients with an estimated 5-year survival rate of 40%.6, 7 However, recurrence rates are high, with nearly 80% of patients developing recurrence of their disease within 5 years of resection.

Of 164 patients who underwent complete resection of HCC at the study institution between 1991 and 2000, 55% had recurrent disease at a median follow-up of 26 months.6 Of this group, 75 patients (83%) had disease recurrence in the liver only. Of these patients, 49 (65%) were candidates for additional therapy based on comorbidities and tumor volume, including embolization in 27 patients, percutaneous ablation in 8 patients, and reresection in 14 patients. The treatments available to treat such patients include bland particle embolization, chemoembolization, radiofrequency ablation (RFA), ethanol injection, and radiation therapy. Despite the multitude of modalities available, a review of the literature confirms that there is a paucity of data regarding how regional therapies impact patients with HCC who develop disease recurrence after resection.

The Memorial Sloan-Kettering Cancer Center has been a proponent of bland particle embolization in the treatment of both unresectable and recurrent HCC.8 In 1997, we standardized our bland particle embolization technique to include 1) common therapeutic endpoints, 2) diligent follow-up, and 3) combination regional therapies when applicable. Since 1997, we have treated 45 patients with recurrent HCC after resection. This is the patient population included in the current study.

The purpose of the current study was to evaluate the efficacy of bland embolization (in combination with RFA or percutaneous ethanol injection [PEI] in select patients) in the treatment of recurrent HCC after surgical resection.


  1. Top of page
  2. Abstract


Between 1997 and 2004, 806 embolization procedures were performed on 325 patients for HCC. Data were collected and entered prospectively into a secure embolization database within the division of interventional radiology. The database is in compliance with all Health Insurance Portability and Accountability Act (HIPAA) regulations and an Institutional Review Board waiver of authorization was obtained for the current study. Of 325 patients, 45 were treated for recurrent HCC after surgical resection. These patients underwent a total of 142 embolizations. There were 30 men and 15 women, with a median age of 62 years (range, 21-82 years).

All patients had undergone complete resection of the primary tumor. Surgical resection consisted of segementectomy in 21 patients (46.6%), hepatic lobectomy in 14 patients (31.1%), and trisegmentectomy in 10 patients (22.2%). Seventeen patients (37.8%) were positive for hepatitis B virus, 4 patients (8.9%) were positive for hepatitis C virus, and 2 patients (4.4%) were positive for both the hepatitis B and hepatitis C viruses. Seventeen patients (37.8%) had pathologic evidence of cirrhosis, 14 cases of which (82.4%) were determined to be Child-Pugh score A.

At the time of initial embolization, 39 patients (86.7%) had recurrent disease to the liver only, and 6 patients (13.3%) had liver-dominant disease with evidence of extrahepatic spread. Tumor recurrence after surgery was documented by either 1) a new hypervascular lesion on cross-sectional imaging in addition to an α-fetoprotein (AFP) level >500 (64.4%), or 2) needle biopsy (35.6%). The average time to disease recurrence after surgery was 13 months. The average size of the target lesion was 3.7 cm (range, 0.8 cm-10.6 cm) and the average number of recurrent lesions was 3 (Table 1).

Table 1. Clinical Characteristics of Patients in the Current Study
Mean age, y (range)62 (21–82) 
Male sex30 patients 
Viral hepatitis23 patients51.1%
Hepatitis B17 patients37.8%
Hepatitis C4 patients8.9%
Hepatitis B/C2 patients4.4%
Cirrhosis17 patients37.8%
Mean size of initial lesion, cm (range)8.6 (2–17.5) 
Type of surgical resection  
 Segmentectomy21 patients46.7%
 Lobectomy14 patients31.1%
 Trisegmentectomy10 patients22.2%
Mean time to recurrence after surgery, mo (range)13 (2–128) 
No. of tumors at recurrence3 
Mean size of largest recurrent lesion, cm (range)3.7 (0.8–10.6) 


Preprocedural evaluation

Before embolization, all patients were evaluated by a multidisciplinary Hepatobiliary Team including hepatobiliary surgeons, gastroenterologists, medical oncologists, and interventional radiologists, and were determined to be appropriate candidates based on a review of lesion morphology, underlying liver function, and comorbid disease. In general, our inclusion criteria for embolization include single tumors measuring <12 cm, multiple or diffuse tumors involving <50% of the liver, the absence of biliary obstruction, and well-compensated (Child-Pugh score A or B) cirrhosis. Hepatic encephalopathy or uncorrectable coagulopathy are considered to be absolute contraindications to embolization at the study institution, and patients with a bilirubin level >2 mg/dL are informed of their increased risk of liver failure. We do not consider portal vein thrombosis to be a contraindication to embolization. In the current study, 2 patients had portal venous occlusion (1 main and 1 left.)

Data collection

Patient demographics, primary tumor characteristics, prior treatment, recurrence, and survival data were recorded. In patients with >1 tumor, the largest tumor was considered to be the “target lesion” and a 2-dimensional size measurement was recorded to assess response. The total number of tumors and the percentage of the liver occupied by tumor were noted as well as the percentage of liver embolized, the number of vessels embolized, and the particles used. Response to therapy was evaluated radiographically approximately 4 weeks after each embolization procedure. Preembolization and postembolization cross-sectional imaging (computed tomography [CT]) or magnetic resonance imaging [MRI]) was evaluated by 1 of 3 interventional radiologists. The size of the treated lesion in 2 dimensions and the percentage of liver occupied by tumor based on operator estimation on the preembolization and postembolization images were recorded. A patient was considered to have responded to therapy if the target tumor was >25% smaller or had an interval development of >50% nonenhancing, low-density appearance on follow-up imaging.


Hepatic angiography was performed from a common femoral approach using a 4-French or 5-French angiographic catheter to establish hepatic arterial anatomy, the location of lesion(s), and portal vein patency. All vessels supplying the target tumor were then embolized as selectively as possible, until there was complete stasis in the parent vessel. Most often, embolization was performed through a microcatheter, using polyvinyl alcohol (50 μm; Cook, Bloomington, IN) or trisacryl gelatin microspheres (40–120 μm; Embosphere, Wayland, MA) or a combination of both. The embolic agent was chosen at the discretion of the operator. Occasionally, larger particles were used in combination with the standard size, or alone if an arteriovenous shunt was identified, to minimize the risk of nontarget embolization into the systemic circulation. No chemotherapy or lipiodol was administered concurrently. In our practice, repeat embolization is performed when there is evidence of viable tumor on follow-up imaging. Repeat embolization included both embolization of the same vessels that had previously been embolized or collateral vessels shown to supply the tumor (e.g., the right phrenic artery). Embolization variables are shown in Table 2.

Table 2. Characteristics of Embolizations Performed
  1. PVA: polyvinyl alcohol.

Median no. of embolizations per patient3.5 
Embolization agent used  
 Embospheres62 embolizations43.6%
 PVA61 embolizations43%
 Embospheres and PVA19 embolizations13.4%
Microcatheter112 embolizations78.9%
Embolization type  
 Lobar33 embolizations23.2%
 Selective (segmental)74 embolizations52.1%
 Superselective (subsegmental)35 embolizations24.6%

Ablation was performed the day after embolization (during the same hospital admission) in select patients with <3 lesions measuring <5 cm. The decision to combine 2 techniques and which ablative technique to use was decided on a case-by-case basis by the attending surgeon and interventional radiologist. PEI was performed more often in the earlier part of the study period before RFA was widely available, or when tumor proximity to a large blood vessel or bile duct precluded RFA. PEI was performed under CT guidance using either a 20-gauge or 22-gauge Westcott needle. The volume injected was approximated by the volume of a sphere (4/3)Πr3)) in which the “r” is approximately half the maximal tumor dimension. CT was performed after the injection of each 5 cc to 10 cc to determine that there was good coverage within the lesion. If needed, the needle was repositioned to maximize ethanol distribution within the tumor. RFA was performed using either the RITA or radiotherapeutic devices, and ablations were performed according to the manufacturer's protocol.


Before embolization, all patients had adequate radiographic imaging with either CT or MRI. If patients were referred from an outside institution with imaging, CT or MRI were repeated only if the outside imaging was thought to be inadequate by the multidisciplinary team. The number of tumors, the size of the largest tumor (in 2 dimensions), the percentage of liver occupied by the tumor, and the degree of tumor necrosis were evaluated on preprocedure images. Angiographic images were obtained before embolization to document the degree of tumor vascularity and again after the embolization of tumor vessels to document loss of tumor vascularity and evaluate the presence of collateral vessels. Our practice is to evaluate patients with contrast-enhanced CT or MRI approximately 4 weeks to 6 weeks after treatment. Blood work and follow-up imaging is performed at that time. The size of the largest lesion is again measured and the degree of tumor necrosis is noted.

Postprocedural monitoring

Patients were admitted to the hepatobiliary service for observation. Patients were maintained on maintenance rate intravenous fluids until they were able to eat. Pain was controlled with oral or intravenous medication as needed. Patients are treated with prophylactic cephalexin starting immediately before the procedure and continuing for 24 hours after embolization. Tazobactam sodium/piperacillin sodium (Zosyn; Lederle, Radnor, PA) was used for antibiotic prophylaxis in patients with bilioenteric communication (e.g., bypass or sphincterotomy) who were treated after 2002.9 Postprocedural laboratory values were drawn on postprocedure Days 1 and 2. If a patient developed a fever >38.5°C with associated leukocytosis, blood cultures were drawn and intravenous antibiotics initiated until the final cultures returned. Patients were not discharged receiving oral antibiotics unless the blood cultures indicated a systemic infection. Under these circumstances, patients were given a course of appropriate antibiotics based on the sensitivity of the organism isolated. Once discharged, patients were followed clinically by 1 of 5 hepatobiliary surgeons. Patients were evaluated both clinically and with cross-sectional imaging on a regular basis for evidence of disease recurrence or progression and deterioration of liver function. Indications for repeat embolization included contralateral disease, persistent or new viable tumor demonstrated by contrast-enhanced CT or MRI, ongoing symptoms of mass effect, or rising serum AFP.


Complications were classified according to the Society of Interventional Radiology (SIR) Standards of Practice into “minor,” which require no therapy or nominal therapy and “major,” which require treatment, have permanent adverse outcomes, or death. In addition, we reported complications from interventional procedures using the Common Terminology Criteria for Adverse Events (version 3)10 as follows. Grade I complications require no intervention, but include asymptomatic vascular complications. Grade II complications include those requiring bedside medical management and would include bacteremia requiring intravenous antibiotics and cardiovascular arrhythmias requiring additional medications. Grade III complications are those that require an intervention; these would include vascular complications requiring stent placement or cardiovascular compromise requiring a monitored setting or extended therapy. Grade IV complications result in some form of chronic disability, and a Grade V complication is a periprocedural death.

Statistical Analyses

Statistical analyses were performed using SPSS statistical software (SPSS Inc., Chicago, IL). Statistical significance was defined as P <.05. Survival time was defined as the time from the date of first embolization to the date of death or last follow-up and the survival curve was estimated using the Kaplan-Meier survival method. Kaplan-Meier estimates of survival time in various groups were compared using the log-rank test (discrete variables) or Cox proportional hazards model (continuous variables). A significance level of 15% in the univariate setting was used as the criterion for including a variable in the multivariate modeling procedure. Stepwise regression was used to determine variables incorporated into the multivariate Cox regression model. The associations between covariates were assessed using the Fisher exact test.


  1. Top of page
  2. Abstract


The median number of embolization procedures per patient was 3.2 (range, 1–14 procedures). The median follow-up for all 45 patients was 31 months. The median follow-up for surviving patients was 34 months. The overall survival (OS) from the time of initial embolization was 86% at 1 year, 74% at 2 years, and 47% at 5 years, with an overall median survival of 46 months (Fig. 1). At the time of last follow-up, 18 patients (40%) had died 0.4 months to 45.9 months after the diagnosis of their recurrence (including 1 patient who had a fatal myocardial infarction 12 days after embolization) and 7.4 months to 100.4 months after surgical resection. Twenty-two patients (48.9%) were alive with radiographic or serologic evidence of disease 11 months to 88.3 months after recurrence. Three patients (6.6%) were alive with no evidence of disease 22 months, 30 months, and 43 months, respectively, after their recurrence. Two patients (4.4%) were lost to follow-up 22 months and 32 months, after their last embolization procedure.

thumbnail image

Figure 1. Kaplan-Meier graph showing the overall survival of patients undergoing salvage embolization for recurrent liver tumors after definitive surgical resection. HCC: hepatocellular carcinoma; DOD: dead of disease.

Download figure to PowerPoint

On univariate analysis, significant risk factors for poor survival included more than 1 recurrent tumor, lack of radiographic response to embolization, extrahepatic disease, and the presence of satellite tumors at the time of initial surgery. There was no survival difference noted between patients who had cirrhosis and those who did not (P = .85). Similarly, there was no difference noted in survival between patients with hepatitis and those without (P = .73) (Table 3.)

Table 3. Univariate Survival Analysis
Variable No.No. Dead1-Year Survival, % (95% CI)3-Year Survival, % (95% CI)P Value*
  • 95% CI: 95% confidence interval.

  • *

    The P values for discrete variables were computed using the log-rank test. The P values for continuous variables were computed using a Cox regression model.

GenderMale301486 (74–99)59 (39–78).31
 Female15486 (67–100)78 (56–100) 
Age at first visit, y 4518  .68
SurgeryLobectomy/trisegmentectomy241379 (63–95)56 (35–77).09
 Segmentectomy21595 (85–100)74 (51–97) 
Satellite tumors at the time of initial surgeryPresent15973 (51–96)43 (16–70).05
Absent29993 (83–100)76 (58–93) 
Size of the initial tumor, cm 4418  .20
No. of recurrent tumors112210092 (76–100).01
 2–418689 (74–100)75 (53–97) 
 ≥5/diffuse151070 (45–95)23 (0–50) 
Size of the lesion prior to embolization< 5 cm361288 (78–99)74 (58–90).10
≥ 5 cm9678 (51–100)30 (0–62) 
Percentage of liver embolized< 5020595 (85–100)80 (59–100).03
 ≥ 5013869 (44–94)32 (4–61) 
CirrhosisPresent18683 (65–100)77 (57–97).85
 Absent271288 (76–100)56 (35–77) 
HepatitisPresent22786 (71–100)67 (45–90).73
 Absent231187 (72–100)60 (39–82) 

On multivariate analysis, the number of recurrent tumors (P = .003), the presence of satellite tumors at the time of initial surgery (hazards ratio of 3.5; 95% confidence interval [95% CI], 1.2-10 [P = .02]), and a largest lesion measuring >5 cm prior to embolization (hazards ratio of 2.9; 95% CI, 0.9-8.8 [P = .07]) were found to be independent predictors of overall worse survival. Patients with ≥5 tumors at the time of disease recurrence or with diffuse tumors were found to have a hazard of death that was 14 times (95% CI, 2.6-74.5) that of those with 1 tumor whereas patients with 2 to 4 tumors were found to have a hazard risk of death that was 3.5 times (95% CI, 0.6-19.9) that of patients with 1 tumor (Table 4; Fig. 2).

Table 4. Multivariate Survival Analysis Performed on Variables at a Significance Level of 15% by Univariate Analysis
Variable No.Hazard Ratio (95% CI)P Value
  1. 95% CI: 95% confidence interval.

No. of recurrent tumors1121.0.003
 2–4183.5 (0.6–19.9) 
 ≥5/diffuse1514 (2.6–74.5) 
Satellite tumors at the time of initial surgeryAbsent291.0.02
Present153.5 (1.2–10) 
Size of the lesion prior to embolization<5 cm361.0.07
≥5 cm92.9 (0.9–8.8) 
thumbnail image

Figure 2. Survival from the time of embolization according to (A) presence of cirrhosis, (B) the number of recurrent tumors, and (C) the size of the largest recurrent tumor before embolization was performed.

Download figure to PowerPoint


Thirty-three patients (73%) had a demonstrable imaging response to embolization on cross-sectional imaging. This included either an interval development of a >50%, low-density, nonenhancing tumor in 29 patients and a >25% decrease in the target tumor size in 13 patients. The median survival from the time of embolization for patients with a demonstrable radiographic response was significantly longer (47.3 months vs. 35.5 months) than for patients without such a response (P = .002).

Length of Stay

For the 45 patients undergoing 142 embolizations, the median length of hospital stay was 3 days (range, 2–25 days). The extended days in the hospital for some patients was related to procedural complications.

Morbidity and Mortality

Information regarding procedure-related complications was obtained after all embolizations. There were 23 complications (14%) in 15 patients after a total of 164 embolizations. All complications were categorized as “major” based on the SIR Standards of Practice criteria. One patient had a periprocedural myocardial infarction and died (SIR category of “major”; Grade V). There were 5 Grade III complications (2 patients with duodenitis, 1 patient with pancreatitis, and 2 patients had external iliac artery dissections treated with metallic stents.) Seventeen complications were Grade II and included 13 episodes of prolonged fever (lasting >48 hours) in 4 patients (1 of whom had documented bacteremia), and a periprocedure atrial arrhythmia requiring medication in 4 patients.


  1. Top of page
  2. Abstract

HCC most often occurs in the setting of cirrhosis due to viral or alcoholic cirrhosis.11 Hepatic resection and transplantation are traditionally considered the only curative treatment options for these patients.6 Untreated, the prognosis of patients with HCC is dismal (reportedly 8.3 months in patients with Okuda Stage 1 disease, 2 months in those with Okuda Stage 2 disease, and 0.7 months in patients with Okuda Stage 3 disease).3 Because the underlying cirrhosis persists as a “field defect,” these patients are predisposed to developing additional foci of HCC. Although repeat hepatectomy has been reported to provide survival similar to the initial hepatectomy, it is possible in only a minority of patients because of limited hepatic reserve.6, 12, 13

Parenchyma-sparing local therapies are often the only feasible options to treat hepatic recurrences in this patient population. To our knowledge, there is a paucity of data in the literature regarding the true impact of regional therapies on the treatment of recurrent HCC and to our knowledge, the current series is the only study to date to report the impact of bland particle embolization in this patient population.

Transarterial chemoembolization (TACE) has been reported to provide 3-year and 5-year survival rates of 24% to 38% and 0% to 21%, respectively, in the postresection patient population,14 and similar survival has been reported after PEI and RFA.12 As evidenced by a recent metaanalysis of TACE for the treatment of HCC, the addition of chemotherapy in addition to embolization was associated with an increase in morbidity, including persistently elevated liver enzymes.15 At the Memorial Sloan-Kettering Cancer Center, it is our practice to treat patients with intrahepatic recurrence of HCC in the setting of well-compensated cirrhosis using transcatheter arterial embolization, also know as bland arterial embolization, followed by local ablation with PEI or RFA when there are <3 lesions measuring <5 cm in size.

In our database of > 800 embolization procedures, the morbidity associated with bland particle embolization is <10% (unpublished data). However, the current study included only patients who had undergone prior liver resection, some with bilioenteric bypass. Before November 2002, when Geschwind et al.9 published their series demonstrating prophylaxis with tazobactam sodium/piperacillin sodium to have a lower incidence of hepatic abscess and infection after chemoembolization in patients with bilioenteric reconstruction, we used cephalosporin prophylaxis for all patients. In the current study, approximately 54% of complications (13 of 24 complications) were prolonged fever or bacteremia that occurred in 4 patients, 3 of whom had undergone prior bilioenteric reconstruction. Three of 164 embolizations (1.8%) were complicated by duodenitis or pancreatitis. We hypothesize that the surgical compromise of normal collateral circulation to these organs in combination with complicated postoperative anatomy account for the relatively high occurrence of these complications.

In our early experience with bland arterial embolization for HCC, we reported 1-year and 2-year actual survival rates of 50% and 33%, respectively, with a median survival of 11.7 months.8 Similar to many of the TACE series published at the time, 35% of patients had poorly compensated liver disease, either Okuda Stage 2 or Stage 3 disease, and this subset of patients had a significantly poorer survival compared with patients with Okuda Stage 1 disease, in whom the 1-year and 2-year survival rates were 60% and 45%, respectively. Based on this experience, we have revised our criteria for embolization to include patients with Okuda Stage 1 and Stage 2 cirrhosis, and in our current cohort, 93.3% of patients had Okuda Stage 1 disease at the time of first embolization. In addition, we currently are at a different point in the “learning curve” of the technique of bland arterial embolization. All 5 interventional radiologists performing this procedure to treat HCC at the Memorial Sloan-Kettering Cancer Center use a consistent technique, thereby minimizing interoperator variability.

Specifically, we aim to embolize tumor(s) as selectively as possible using small (40–120 μm of trisacryl gelatin microsphere or 50 μm of polyvinyl alcohol) particles until stasis in the target vessel is achieved. In the case of diffuse tumors, lobar or segmental embolization is performed, and if hepatic vein shunting is identified angiographically, larger particles are used to minimize the risk of nontarget pulmonary embolization.16 In practice, the type of embolization performed is dependant on several factors, including the volume and distribution of tumors in the liver, as well as technical considerations (i.e., the ability to place a catheter selectively or subselectively.) Greater than 75% of the 142 embolizations performed in the current study were either selective (segmental) or superselective (subsegmental.) However, the effect of the type of embolization as an independent predictor of outcome is impossible to determine on the basis of these data because patients who underwent lobar embolization were more likely to have higher burden of disease.

We chose to study this group of patients because it is common in our practice to treat patients who have undergone prior curative resection for HCC, and the surgical literature describing survival after resection for HCC often fails to describe the impact of embolotherapy on overall survival. In fact, a recent surgical series from our institution reported 164 patients who underwent liver resection for HCC.6 At a median follow-up of 26 months, 90 patients (55%) developed tumor recurrence. Of these, 75 patients (83%) had liver-only recurrence, and 34 of the 75 patients (45%) underwent either bland embolization or PEI. Unfortunately, as in many surgical series reporting HCC survival, the data comparing the survival of patients who underwent salvage treatment by regional therapies with those who did not were not included in the analysis.

In our recent series of patients with solitary lesions measuring up to 7 cm in size who were treated with either surgical resection or bland particle embolization plus ablation, 19 of 33 patients in the surgical group (57.6%) developed a disease recurrence with a median follow-up of 22 months. Eight patients were candidates for “salvage embolization” and we observed a mean survival in this group of 26 months (median survival not met) from the time of embolization, with an 87% 2 year survival. The 11 patients who did not meet criteria for embolization for their disease recurrence (5 due to extrahepatic disease and 6 due to inadequate liver reserve), had a median survival of 4.9 months from the time of embolization, with no 2-year survivors reported.20 Although this series of patients was highly selected, including only those with solitary tumors measuring <7 cm, the impact of embolization on these patients who developed disease recurrence after surgical resection is similar to that noted in the current series.

There have been several studies published to date evaluating the effectiveness of regional therapies using response as an endpoint. However, improving survival is the ultimate goal of local therapy in the majority of cases, so we chose survival as our primary endpoint for the current study. The only available data regarding embolization for recurrent HCC involves using chemoembolization. We can extrapolate from the chemoembolization experience14, 15, 17–19 to put our patient survival in perspective. However, one must remember that the majority of the randomized controlled trials excluded patients with portal venous obstruction or multifocal disease. In the current series, 2 patients (4.4%) had portal venous obstruction and 10 patients (22%) had >5 tumors or multifocal HCC.

In 2002, a 3-arm trial published from Barcelona compared chemoembolization (gelatin sponge with doxorubicin), bland embolization (using a gelatin sponge alone), and conservative therapy.17 This study was stopped when sequential analysis demonstrated a survival benefit for chemoembolization compared with conservative management. In 112 patients (40 randomized to chemoembolization, 37 randomized to bland embolization, and 35 randomized to conservative management), the survival rates at 1 year and 2 years were 82% and 63%, respectively, for the chemoembolization group; 75% and 50%, respectively, for the bland embolization group; and 63% and 27%, respectively, for the conservative therapy arm. Similar to the results of the metaanalysis performed by Camma et al.,15 the Barcelona study failed to demonstrate a significant survival advantage of chemoembolization over bland embolization. Although the current study involves a different patient population (patients with inoperable, recurrent HCC) from the Barcelona study (initial presentation of unresectable HCC), the overall survival compares quite favorably, with 1-year and 2-year survival rates from the time of the first embolization of 86% and 74%, respectively.

In the current series, the 1-year, 2-year, and 5-year actuarial survival rates from the time of first embolization after surgical resection of HCC were 86%, 74%, and 47%, respectively. On multivariate analysis, only the number of lesions and the size of the largest lesion were found to have a significant impact on prognosis (Table 4) (Fig. 2). It should be noted that our selection criteria were based on the overall clinical status of the patient rather than specific tumor characteristics. Patients with a large disease burden or bilobar disease will require multiple embolizations over the course of their treatment based on interval imaging findings. Despite the wider range of tumors treated in the current study, the results are similar to those of Choi et al.,12 whose recent series reported 1-year, 2-year, and 3-year survival rates of 82%, 72%, and 54%, respectively, in treating patients with recurrent HCC measuring <4 cm in size using RFA alone. Unfortunately, the impact of percutaneous ablative techniques on survival in patients with HCC has been limited to selected patients with solitary tumors or low-volume disease. Therefore, although percutaneous ablative therapies alone may be suitable for patients with small tumors, bland embolization has the potential to increase the number of patients who are eligible for treatment while providing a similar survival benefit.

It has been hypothesized that arterial embolization for HCC may have the counterproductive effect of inducing angiogenesis and neovascularization by causing a locally hypoxic environment.14 We are not aware of any clinical data in humans that substantiate this theory and believe that by using bland arterial embolization we are able to achieve mechanically what antiangiogenic drugs aim to accomplish pharmacologically—namely, to devascularize tumor and effect tumor cell death.

Bland embolization is an effective therapeutic modality in the treatment of recurrent HCC after surgical resection. These type of survival data are currently missing from the majority of surgical series in which overall survival includes patients who undergo salvage therapy with regional techniques. In the current series, patients who were eligible for salvage embolization were found to survive significantly longer than the median survival of 11 months we have previously quoted to patients who develop a recurrence after surgical resection.6 Diligent follow-up after the resection of HCC and the aggressive treatment of disease recurrence with bland embolization may allow a minority of patients to be rendered free of disease and the majority of patients to experience a prolonged survival benefit.


  1. Top of page
  2. Abstract