Drop-out rates of patients with hepatocellular cancer listed for liver transplantation: Outcome with chemoembolization



Patients with hepatocellular carcinoma (HCC) are assigned model for end stage liver disease (MELD) scores to provide access to liver transplantation (LT). An equitable policy would equate HCC progression beyond acceptable transplantation criteria with death on the waiting list. However, limited information is available regarding this issue. Thus, our aim was to analyze drop-out rates on the waiting list for patients with HCC. Between January 1994 and August 2001, 54 patients with HCC were listed for LT. Patients underwent chemoembolization prior to LT, and were assessed every three months for disease progression until LT. Two patients were stage T1, 45 patients were stage T2, and 7 patients were stage T3 at time of first chemoembolization. Median time was 211 days (range 28–1099 days) for patients that were eventually transplanted. Eight patients were removed from the list. Cumulative probability of drop out on the waiting list, assessed by Kaplan-Meier analysis, was 15% and 25% at 6 and 12 months, respectively. There were no significant differences in age, gender, initial tumor stage, or serum AFP levels in those who eventually underwent LT vs. those who dropped out. In conclusion, neoadjuvant chemoembolization for patients with HCC has a drop-out rate of 15% over 6 months. (Liver Transpl 2004;10:449–455.)

Hepatocellular carcinoma (HCC) is a common consequence of cirrhosis with a risk of 1–3% per year in cirrhotic patients. Moreover, the incidence of this malignancy has increased worldwide, with an incidence of 1.4/100,000 patients per year in the 1970s to an incidence of 2.4/100,000 patients per year in the 1990s.1, 2 The increasing incidence of HCC has been attributed, in part, to the growing number of patients with hepatitis C. Surgical resection and orthotopic liver transplantation (LT), are the preferred curative therapies for patients with unicentric disease less than 5 cm in size without vascular invasion of major vessels.3–5 Patients with up to three lesions each less than 3 cm are also currently candidates for LT.3 Surgical treatment is dependent on the stage of cirrhosis. In patients with Child -Turcotte- Pugh class A cirrhosis, in the absence of portal hypertension and normal serum bilirubin values, resection can be curative with 3-year survival rates of up to 70%.6 In contrast, LT is often the preferred treatment of choice for cirrhotic patients with portal hypertension, elevated serum bilirubin values, and advanced stage of cirrhosis (Child-Turcotte-Pugh class B and C). Selected patients benefit from LT with 5-year survival rates of 75% and greater.3 Because of the increasing number of patients with HCC and cirrhosis, the success rate of LT for this malignancy, and the limited number of cadaver donor organs available for LT, organ allocation has become strained. Obviously, those HCC and non-HCC patients who may benefit from LT merit timely and equitable access to the procedure. The current allocation issues are what constitute a fair and equitable access in a changing organ allocation environment.

The model for end stage liver disease (MELD) scoring system was implemented as a mechanism to prioritize organ allocation in February of 2002. MELD is a mathematical model based on the serum creatinine, bilirubin, and international normalized ratio for the prothrombin time.7 In part, MELD was implemented because it predicts mortality from end-stage cirrhosis and, therefore, affords prioritization of organs for patients by risk of death.8 The quandary for patients with HCC is that they often have low calculated MELD scores despite the potential fatal nature of their disease. To rectify this dilemma, patients with HCC now receive arbitrarily determined MELD scores thought to be sufficient to provide timely liver transplantation. Although this approach is logical, the score allocated has been based on scarce data. Additional information would be useful to guide these national policy decisions.

For patients with HCC, tumor progression beyond the current criteria for liver transplantation should be equated with death on the waiting list. Indeed, if the tumor progresses beyond transplant criteria, only palliative options are available for most patients and they will ultimately succumb to their malignant liver disease. Thus, MELD score assignments for HCC patients should equate the risk of tumor progression beyond criteria to the risk of death for non-HCC cirrhotic patients over the same time interval. Another confounding variable is the use of tumor directed therapy prior to liver transplantation. Many centers are reluctant to observe patients with HCC on the waiting list without use of neoadjuvant therapy to prevent progression, vascular invasion, and/or to potentially reduce tumor dissemination at the time of transplantation.9 Neoadjuvant therapeutic options include percutaneous ethanol injection therapy, radiofrequency ablation, and chemoembolization. We have systematically used chemoembolization therapy for all patients with HCC prior to LT since 1994.10 Given this systematic, protocol-driven experience, our aims were to assess the drop-out rate for patients with HCC on the waiting list. Overall outcome following LT was also examined. In our patient population, the cumulative drop-out rate by Kaplan-Meier analysis is approximately 15% at 6 months. Five-year outcomes were satisfactory with a survival rate of 77% (confidence intervals of 62–88%).


HCC, hepatocellular carcinoma; MELD, model for end stage liver disease; LT, liver transplantation; US, ultrasonography; CT, computed tomography; MRI, magnetic resonance imaging; EASL, European Association for the Study of Liver Disease; AFP, alpha-fetoprotein; INR, international normalized ratio; TNM, tumor-node-metastasis; HCV, hepatitis C virus; HBV, hepatitis B virus; OLT, orthotopic liver transplantation.

Patients and Methods

Diagnostic Criteria for HCC and Preoperative Staging

This study was a retrospective chart review of 54 patients with hepatocellular cancer listed for orthotopic liver transplantation between January 1994 and August 2001. Under the UNOS guidelines during this time period, patients with T1-T2 HCC were registered as status 2B for LT. The tumor staging system utilized by UNOS is provided in Table 1. The diagnosis of HCC was established by imaging criteria using parameters remarkably similar to those recently proposed by the European Association for the Study of Liver Disease (EASL) consensus conference for the nonbiopsy diagnosis of HCC.11 Diagnostic criteria were the presence of a mass lesion in a cirrhotic liver demonstrating contrast enhancement by computed tomography (CT) or magnetic resonance imaging (MRI) and confirmed by neovascularization on angiography. Fine needle aspirate of the mass lesion was performed for indeterminate cases.

Table 1. American Liver Tumor Study Group Modified Tumor-Node-Metastasis (TNM) Staging Classification (1)
Classification Definition
TXNXMX Not assessed
T0N0M0 Not found
T1   1 nodule ≤1.9 cm
T2   1 nodule 2.0-5.0 cm; 2 or 3 nodules, all ≤3.0 cm
T3   1 nodule >5.0 cm; 2 or 3 nodules, at least 1 >3.0 cm
T4a   4 or more nodules, any size
T4b   T2, T3, T4a plus gross intrahepatic portal or hepatic vein involvement as indicated by CT, MRI or ultrasound
N1   Regional (portal hepatis) nodes, involved
M1   Metastatic disease, including extrahepatic portal or hepatic vein involvement
Staging Classification
  1. Abbreviations: T, Tumor; N, Node; M, Metastasis; CT, computed tomography; MRI, magnetic resonance imaging.

Stage I T1  
Stage II T2  
Stage III T3  
Stage IV T4a, T4b  
Stage IVB Any N1, M1  

The initial tumor staging was based on two imaging studies, including ultrasonography (US), (CT), or (MRI). The maximum number of lesions observed on any single study was taken to be the maximum number of lesions. Unlike the recent EASL consensus criteria, a lesion did not need to be observed by more than one study if contrast enhancement was present. All patients had a chest CT scan, a bone scan, and a brain CT scan to exclude metastases.

Preoperative Chemoembolization

Patients underwent elective chemoembolization unless they had poor liver function (clinically evident ascites on diuretic therapy, an albumin of less than 2.5, a bilirubin greater than 5, or an international normalized ratio (INR) for the prothrombin time of greater than 2.0)12; patients with poor liver function underwent chemoembolization immediately prior to LT . Patients with satisfactory hepatic function (n=52) were treated electively with chemoembolization. A second chemoembolization was performed 2 months later for patients awaiting transplantation. Subsequent chemoembolizations were performed for viable tumors as evidenced by persistent contrast enhancement on imaging studies (CT or MRI) and/or a rising or persistently elevated serum alpha-fetoprotein (AFP) concentration. Any patient previously treated with chemoembolization and awaiting LT that was found to have a new hepatic lesion and still met the UNOS criteria for transplantation also received additional therapy with chemoembolization. Two patients with poor liver function received chemoembolization immediately prior to LT.

The embolization solution contained mitomycin-C 10 mg and doxorubicin 50 mg, radiopaque contrast agent Conray 60 (Mallinckrodt Medical Inc., St. Louis, MO), and Ivalon particles (250–355 microns). The mixture of the chemotherapeutic agents and Ivalon particles was suspended in 10 ml of contrast. The dose of doxorubicin was reduced to 20 mg for patients with a bilirubin greater than 3 mg/dl.

Monitoring Patients for Disease Progression

All patients underwent a CT scan of the abdomen and chest every three months to assess for tumor progression. Patients were excluded for LT if there was tumor progression beyond UNOS criteria (solitary tumor > 5cms, > 3 lesions, or 3 lesions with one of the lesions exceeding 3 cms in maximum diameter, or tumor thrombosis of the portal vein, its major tributaries, or a hepatic vein). All major venous thromboses were equated with tumor thrombus and patients delisted; attempts were not made to biopsy the thrombus and confirm or disprove tumor thrombosis.

Surgical Procedure

At the time of LT abdominal exploration was performed to be certain that there was no evidence of extrahepatic metastasis. This evaluation included biopsies of the hilar lymph nodes only if they appeared suspicious for metastatic disease. A standard LT was performed as previously described.10 There were 7 patients who had caval sparing hepatectomy (piggyback procedure) and 39 patients who had a donor caval interposition graft with venovenous bypass.

Immunosuppression Therapy and Patient Follow-Up

Between 1994 and 1996, patients received tacrolimus (therapeutic blood levels of 5 to 15ng/ml) along with prednisone and azathioprine. In 1997, we discontinued the use of azathioprine in patients with hepatitis C. In 1999, our standard immunosuppressive regimen was changed to tacrolimus, prednisone, and mycophenolate mofetil. Prednisone and mycophenolate mofetil were discontinued 4 months after liver transplantation.All patients were evaluated 4 months and annually after LT. A CT scan of the chest and abdomen and serum AFP determination were performed at these intervals. No patients were lost to follow-up.

Pathologic Examination of the Explanted Liver

The explanted livers were examined in the fresh state. The entire organ was sliced into parallel 1-cm thick sections. All nodules with different color or texture from the background, all nodules greater than 1 cm, four random liver sections and hilar tissue including lymph nodes were examined microscopically by a hepatopathologist using H&E stains on paraffin-embedded tissue. Additional cytochemical and immunohistochemical stains were used as indicated.

Statistical Analysis

The Kaplan-Meier method was used to calculate the overall cumulative probability of drop out on the waiting list and patient survival after LT. Cox proportional hazards regression was used to evaluate the association between drop-out rate on the waiting list and age, gender, AFP value > 100ng/ml, number of lesions, and understaging.


Patient Population and Tumor Stage

The characteristics of the 54 patients are summarized in Table 2. The most common etiologies of cirrhosis were HCV, alcohol, a combination of HCV and alcohol, followed by hepatitis B, cryptogenic cirrhosis, and hemochromatosis. Initial HCC staging is summarized in Table 3. Forty patients (73%) had solitary lesions and 14 (27%) had multicentric disease. Seventeen patients (32%) had normal serum AFP concentrations of less than 11ng/dl, 18 patients (33%) had values between 15 to 100 ng/dl, and 19 (35%) had values greater than 100 ng/ml; a serum AFP value that is felt to be necessary to confirm the diagnosis of HCC in the presence of a mass lesion in a cirrhotic liver.13 Using the current UNOS staging classification for HCC, 2 patients were stage T1, 45 patients were stage T2, and 7 patients were stage T3. The T3 patients were listed prior to UNOSs institution of consensus national criteria for status 2B registration; during this time our criteria for registration was 3 lesions < 5 cm.

Table 2. Baseline Clinical and Demographic Characteristics
Patients N = 54OLT (46)HCC Progression (6)Medical Delisting (2)
  1. Abbreviations: HCC, hepatocellular carcinoma; OLT, orthotopic liver transplantation; HCV, hepatitis C virus; HBV, hepatitis B virus; αFP, alpha-fetoprotein; BMI, body mass index.

Gender (M:F)31 : 154 : 21 : 1
BMI28.5 + 4.630.9 + 2.930.8 + 4.2
Median Age (years, range)55.4 (39.2–68.2)58.2 (38.9–66.3)59.1 (54.4–63.7)
Blood Group   
Child–Turcotte–Pugh Score   
Etiology of Liver Disease   
 HCV + Alcohol720
Cryptogenic cirrhosis212
αFP (ng/ml) at time of listing for OLT   
Table 3. Tumor Characteristics
Tumor SizeOLT (46)HCC Progression (6)Medical Delisting (2)
  1. Abbreviations: OLT, orthotopic liver transplantation; HCC, hepatocellular carcinoma.

Unicentric (n = 40)3541
 ≤3 cms2530
 >3 cms and ≤5 cms901
 >5 cms110
Multicentric (n = 14)1121

Chemoembolization and Drop Out on the UNOS Waiting List

Fifty four patients with HCC underwent an average of three (range 1–4) chemoembolization procedures prior to LT. None of the patients had decompensation of their underlying liver disease related to the chemoembolization procedure itself. One patient developed a liver abscess requiring percutaneous drainage and antibiotic therapy. The median time on the waiting list was 211 days (range 28–1099 days) for the 46 patients that eventually underwent transplantation. Eight patients dropped off the waiting list. Two patients had medical complications with one patient developing an acute myocardial infarction 8 months after registration and the second patient died of a subdural hematoma 4 months after registration. The remaining 6 patients had disease progression on the waiting list. One patient developed pulmonary metastases 2 months after registration, two developed vascular invasion with portal vein thrombosis after 2 months and 4 months, one developed adrenal metastases after 2.5 months, and two had intrahepatic disease progression after 2 months and 4 months. There were no significant association between tumor characteristics (size p=0.27, multicentricity p=0.74), and time until removal from waiting list although the numbers are limited. There was also no significant association between age (p=0.88) or gender (0.59) and time until removal from waiting list. A serum AFP concentration of > 100ng/ml has been suggested to identify patients at high risk for disease progression.14 However, in our study five of the patients who dropped out had serum AFP values > 100 ng/ml where as only three had values > 100ng/ml. These data were not statistically significant (p=0.56). The Kaplan-Meier analysis for drop out on the waiting list with 95% confidence intervals is shown in Figure 1. The cumulative probability of drop out was approximately 15% at 6 months, and 25% at 12 months.

Figure 1.

Cumulative incidence of drop out from the UNOS waiting list.

Accuracy of Preoperative Staging

The relationship between preoperative staging by imaging and the postoperative staging by pathologic examination of the explanted liver was assessed. Because chemoembolization may alter the size of the lesions, only the number of lesions was analyzed. All patients had an ultrasound, CT-scan of the abdomen, and an angiogram prior to transplantation. The maximum number of lesions on any study was taken as the number of lesions present. Understaging (more lesions in the explanted liver than observed by preoperative staging) was observed in 10 explanted livers (21.7%). In nine out of ten patients, the missed lesions were all less than 2 cm; the other patient had diffuse nodular replacement of the right lobe of his liver with hepatocellular cancer. Although approximately 20% of patients were over staged in this analysis, this may reflect the efficacy of chemoembolization to down stage the tumor burden prior to LT. Thus, with current imaging modalities, small lesions are frequently missed.

Efficacy of Chemoembolization

The explanted livers were assessed for the efficacy of chemoembolization. Percent necrosis of the dominant lesion was analyzed as well as the presence of viable tumor. Complete necrosis of all lesions was observed in 7 patients (15%); all others had viable tumor in the explanted liver. However, greater than 90% necrosis was observed in 27 dominant lesions (59%) while 90–50% necrosis was identified in 12 (26%) of the dominant lesions. Chemoembolization was more effective in unicentric lesions than in multicentric disease. Indeed, 74% of unicentric lesions (26 out of 35) had greater than 80% necrosis of the identified cancer whereas only 54% of multicentric lesions (6 out of 11) displayed this degree of necrosis. Interestingly, chemoembolization was more effective in multicentric lesions > than 3 cm (all lesions had > 80% necrosis) than in lesions < 3 cm (only 2/3 of the lesions had > 80% necrosis). The ability to better chemoembolize larger lesions may relate to better identification of the tumor during the angiographic procedure. Only one of the lesions had microscopic vascular invasion and none had gross vascular invasion of a major hepatic or portal vein. Embolization, as has been identified in other studies, seldom eliminates the cancer from the cirrhotic liver.15 However, the presence of microscopic vascular invasion was surprisingly low and the ability of embolization to prevent and obliterate vascular invasion deserves further study in this clinical context.

Outcome on an Intent-To-Treat Basis and Following Liver Transplantation

On an intent-to-treat basis survival was 61% at 5 years (Fig. 2). This analysis takes into account survival following registration on the UNOS list for transplantation and is most meaningful to the patient. The intent-to-treat analysis accounts for death on the waiting list, following drop out from the waiting list, and after liver transplantation. Of the patients who underwent liver transplantation the 5-year survival was 74%. Thus, the intent–to-treat analysis drops survival by 13%. Following transplantation, 12 patients died: one from hepatic artery thrombosis, five due to recurrent hepatitis C, one due to head and neck cancer, and five due to recurrent HCC. There were no significant differences in age, gender, tumor characteristics (size, multicentricity), and serum AFP levels in those who developed tumor recurrence vs. those who did not (data not shown).

Figure 2.

Probability of survival on an intent-to-treat basis following initial registration with UNOS for liver transplantation.


The principal results of our study relate to organ allocation for patients with hepatocellular cancer. Our results demonstrate that: 1) in patients with preoperative chemoembolization, the cumulative probability of drop out due to death and/or disease progression is 15% at 6 months; and 2) in patients undergoing LT the overall 5-year survival is 75% comparable to survival after transplantation without HCC. We will discuss these observations further in the following paragraphs.

Llovet and associates6 reported a 25% drop-out rate due to tumor progression in the first 6 months of waiting for liver transplantation. Of note, these patients did not receive any neoadjuvant therapies. Furthermore, strict criteria were used to stage the tumor; to be counted as a malignant lesion, each tumor needed to be visualized on two imaging studies. For example, a patient with two lesions on MRI and ultrasound, but four lesions on CT would be considered to have two lesions. This approach results in liberal listing criteria, and perhaps a high drop-out rate. Yao and associates16 have also reported on the drop-out rate for patients with HCC awaiting liver transplantation. In their series, the cumulative probability for drop out at 6 months was also approximately 25%, although patients were not censored for transplantation. This group recently updated their study censoring the date for transplantation; in the revised analysis, the drop-out rate was <10% at six months.17 In our current study the drop-out rate was approximately 15% at 6 months. The differences between our observations and the previous studies may reflect differences in staging and listing as noted above, and or the protocolized use of chemoembolization prior to liver transplantation. Indeed, we staged patients based on the maximum number of lesions on any imaging study, and employed elective chemoembolization prior to LT. Although the differences between the current and published results are not drastically different, they do translate into different MELD scores and as such are of pragmatic value. Our results also highlight the importance of multiple centers reporting drop-out rates to help shape policy decisions.

We employed systematic arterial chemoembolization in our patients waiting transplantation, and this is the first report of drop-out rates with a neoadjuvant therapy. The rationale for preoperative chemoembolization is to prevent tumor metastases, progression, and vascular invasion while on the waiting list for LT. The efficacy of chemoembolization in treating HCC has been confirmed in two recent randomized controlled trials and a meta-analysis.18–20 However, the use of preoperative neoadjuvant tumor directed therapy has never been examined in a randomized controlled trial, and, therefore, its impact on drop-out rates cannot be directly ascertained. Our data are somewhat contradictory to the study of Graziadei et al who found no drop out from the waiting list when patients received neoadjuvant chemoembolization.21 However, the monitoring protocol and the criteria for drop out are not specified in the study, and the tumor recurrence rate was 30% as compared to a recurrence rate of 10% in this study. The discrepancy may relate to the thoroughness of repeat staging studies and criteria for drop out between the two studies. Whether prioritization and MELD score assignments for patients with HCC waiting for LT should attempt to provide LT within 6 months, or take into account neoadjuvant therapies and permit longer waiting times will require additional and thoughtful analysis.

Our overall 5-year survival for patients undergoing LT with HCC was 75%. These results are comparable to published results from other centers and overall UNOS survival following LT.22 Of interest was that 5 of the 12 deaths following LT were due to recurrent HCV infection. These data highlight the problems of recurrent HCV following LT, and support the contention that patients with HCV plus HCC do worse as a cohort.23

In summary, this study reports drop-out rates for HCC patients waiting for LT undergoing elective and serial chemoembolization prior to LT. The cumulative probability of drop out for patients in this study with HCC due to death and or tumor progression was 15% at 6 months. This information may contribute to a growing data base to help make policy decisions regarding organ allocation for patients with HCC. This study in itself is insufficient to base policy decisions because of the limited number of patients and relatively short waiting times. Policy decisions can only be made when numerous experiences are assessed and collated.