Staging for patients being evaluated for OLT generally includes dual- or triple-phase helical computerized tomography (CT) or magnetic resonance imaging (MRI) of the abdomen and noncontrast CT of the chest to evaluate for metastatic disease.32-34 Valid comparisons between CT and MRI are difficult because imaging is highly dependent on rapidly improving technology and local expertise. Radiological staging is prone to both overestimation and underestimation of tumor size and number. A recent review of 666 explanted livers from patients who underwent transplantation for HCC under the model for end-stage liver disease (MELD) system showed that the stage of at least 30% of all patients was underestimated and did not meet Milan criteria. Additionally, at least 23% of stages were overestimated.35
 
Whole-body scanning with fluorodeoxyglucose positron emission tomography or bone scintigraphy has commonly been used to evaluate for metastatic disease but is no longer required for listing by the United Network for Organ Sharing (UNOS).29 Positron emission tomography scanning has low sensitivity (55%-60%) for HCC but may identify otherwise unrecognized metastatic disease.36,37
 
Biopsies are generally not required for staging provided that a patient meets the European Association for the Study of the Liver (EASL) noninvasive criteria for HCC (2 coincidental imaging techniques with ultrasound, spiral CT, MRI, or angiography showing a focal lesion >2 cm with arterial hypervascularization or 1 imaging technique showing arterial hypervascularization and a serum a-fetoprotein level >400 ng/mL).38 Biopsies, though, can be helpful because they confirm the presence of HCC and can provide additional information about tumor grade and microscopic vascular invasion (see To Biopsy or Not).
 
Follow-up assessments while liver transplantation is awaited are performed to detect evidence of tumor progression beyond preset criteria. Imaging usually consists of every-3-month CT or MRI of the abdomen and chest. New thrombosis of the portal vein or hepatic vein is considered a tumor thrombus, and the patients are removed from the transplant list.33 Some also advocate repeat whole-body imaging every 3-6 months.31
 
Liver transplantation for hepatocellular carcinoma
History
Liver transplantation for HCC before 1995 yielded disappointing results—2-year survivals were 30% or less,39-41 and the recurrence rates were high after transplantation. These results were due to a bias toward performing transplantation for patients with unresectable tumors. Through the 1980s and early 1990s, hepatic resection remained the treatment of choice for patients with early HCC and enough hepatic reserve to tolerate resection.42 Therefore, transplantation was often left to those with unresectable tumors (large, multiple, or both). The disappointing results called into question the value of transplantation for HCC.43,44
 
However, within the total cohort of HCC patients who underwent transplantation, centers also reported on subgroups with early-stage disease that did well.42 Specifically, it was well known that patients who had undergone transplantation and were found to have incidental small HCCs on histological examination of the explanted liver also had acceptable disease-free survival.39,45 Finally, in 1996, Mazzaferro et al34 from Milan, Italy, published their results of a prospective trial of transplantation for early-stage HCC. Enrollment criteria were set at a single tumor ≤5 cm or no more than 3 tumors all ≤3 cm on preoperative imaging studies. The actuarial survival rate at 4 years was 75%, no different from the expected survival for non-HCC cases. Subsequent centers reported similar results with the same or similar criteria26,28,40 (Table 2). These criteria, often referred to now as the Milan criteria, have been widely adopted and are the cornerstone of pretransplantation evaluation for patients with HCC. Indeed, the TNM staging system was modified such that T2 corresponds to the Milan criteria, and UNOS has adopted these changes into their policy on recipient prioritization for liver transplantation.29
 
Although these important changes in patient selection opened the door to liver transplantation for patients with early HCC, the shortage of organs and increased waiting times hindered delivery of this therapy through the late 1990s and early 2000s. In Spain, the 2-year survival of patients listed for transplantation decreased from 84% to 54% after 1996, when the mean waiting time increased from 62 to 162 days.24 Many patients dropped off the list because of tumor progression beyond T2, thus disqualifying them for upgraded priority. In the United States, the cumulative probability of dropout from the waiting list of HCC patients was 25% per year by 2001.46 Before 2002, the prioritization system for all liver transplantation candidates was based on their Child-Pugh score. Under this system, patients with T2 or lower HCC were moved into a higher-priority group (status 2B), but waiting time within the group remained a significant factor. That system put patients with early HCC at a disadvantage because they had a significant additional risk of becoming disqualified because of tumor growth while they waited. Clearly, the prioritization system for HCC patients was in need of modification.
 
The model for end-stage liver disease (MELD)-based prioritization system Through the mid-1990s, a significant change in the overall prioritization system for transplant candidates was being developed in parallel to changes in HCC transplantation criteria. There was a growing concern that waiting time carried an unjustifiable effect on prioritization. Two reports concluded that waiting time did not predict death while patients were listed for transplantation.47,48 In 1998, the Department of Health and Human Services issued the final rule stating that organs should be allocated on the basis of medical urgency, that waiting time should be minimized as a prioritization criterion, and that futile transplantation should be avoided.49
 
The MELD-based prioritization system for liver transplantation replaced the Child-Pugh system in February 2002. Three laboratory values (international normalized ratio, bilirubin, and creatinine) determine a patient's MELD score. The score was initially designed for predicting mortality after transhepatic portal systemic shunt placement50 but subsequently proved capable of predicting 3-month mortality while patients awaited liver transplantation.51 MELD scores are always whole numbers from 6 to 40, and higher numbers imply a higher mortality and listing priority. Waiting time is considered only if 2 patients have identical scores in their blood group. The MELD score calculation is available online through UNOS (available at:
http://www.unos.org/resources/MELDPELDcalculation.asp).
 
The calculated MELD score is unaffected by the presence of HCC. Therefore, the new allocation system initially gave additional points (up to 24 for T1 HCC and 29 for patients with T2 HCC). Extra points were added every 90 days listed to represent a 10% increase in mortality. In the first year of the new system, the rate of liver transplantations for HCC nearly tripled.35 The average waiting time decreased from 2.28 years before the MELD system to 0.69 years under MELD, and >85% of HCC patients waited less than 90 days for transplantation.
 
Clearly the MELD-based system gave a new advantage to patients with HCC, but it apparently went too far. On retrospective analysis, non-HCC patients with MELD scores of 24 to 29 had a higher chance of dying or dropping off the list because they often had more significant hepatic decompensation than HCC patients with score upgrades.35 Also, the increase in transplantations for HCC had an adverse effect on organ allocation.52 Fourteen percent of transplantations performed for HCC had no HCC on explant histology in the first 8 months of MELD.53 This pretransplantation false-positive diagnosis occurred more often for small, single lesions (eg, T1). Moreover, data from the pre-MELD era indicated that patients with T1 lesions had less than a 10% risk of dropout in the first year listed. Conversely, patients with T2 lesions were responsible for much of the poor intention-to-treat outcomes under the old system.31 Because of these data, the assigned MELD scores for patients with HCC were decreased to 20 and 24 for T1 and T2 lesions, respectively, in April 2003. This change decreased the proportion of transplantations performed for HCC from 21% to 14%.53 Before MELD, the rate was 8%. More recently, the score upgrade for T1 lesions (20 points) was eliminated, so that now only patients with T2 lesions may receive a score upgrade (initially 24 points and now 22 points). The effects of these changes are not yet known.
 
The optimal prioritization system for patients with early HCC continues to be a moving target because of many factors, including opinions regarding just allocation of organs, pretransplantation diagnostic and staging accuracy, and evolving adjuvant therapies. Although changes in the MELD-based prioritization of HCC patients will undoubtedly be required, the MELD-based system may adopt such changes more readily than the prior system. The MELD score provides more objective scoring and more levels of stratification (scores 6 through 40) for analysis.
 
Transplantation survival
Early case series of patients undergoing liver transplantation for HCC included significant numbers of patients with advanced-stage HCC, resulting in poor outcomes and recurrence rates as high as 29% to 54% and 3-year survival rates of 21% to 47%.39-41 From these early series, important predictors of survival included tumor number, tumor size, and vascular invasion. Using these factors, Mazzaferro et al34 reported 74% 4-year survival for patients who met highly selective criteria (the Milan criteria; see above). These findings were reproduced in multiple series and have been adapted by UNOS as the organ-allocation criteria for HCC in the United States. These survival rates in combination with low recurrence rates are superior to those for any other treatments for HCC.3 An analysis of the entire UNOS database of patients with HCC who underwent transplantation between 1987 and 2001 gives the best picture of survival in typical clinical practice. Steady improvement in 5-year survival was found in 935 patients; the most recent time period's survival was 61.1%.54 The improved survival was attributed to adoption of Milan criteria, but was lower than the 77% survival rate for other indications.
 
All these survival rates apply only to patients who actually underwent liver transplantation. From the patients' point of view, intention-to-treat survival is most important when trying to choose a treatment strategy.24 If waiting time is short and dropout rates are low, OLT is superior to other treatment strategies. If waiting times are longer, dropouts accumulate, and other treatments may produce similar survival rates.
 
In addition to the Milan criteria, other factors that reflect underlying tumor biology also seem to affect post-OLT survival in various series by multivariable analysis. These include, in descending order of apparent effect, macrovascular invasion, microvascular invasion,19,26 the presence of satellite lesions, bilobar nodules, and histological tumor grade.26
 
Adjuvant chemotherapy after transplantation
Chemotherapy for HCC results in less than 25% objective response rates and has no clearly beneficial effect on patient survival.3,55 Theoretically, liver transplantation removes the overwhelming bulk of the tumor burden. Adjuvant chemotherapy needs only to eliminate micrometastases and the tumor cells released during surgery, so it may be more effective than when it is used as a primary treatment. Chemotherapy after transplantation, especially for patients with high-risk features, has been practiced in some centers, but there are no randomized trials or large comparisons between cohorts to assess for beneficial effects.
 
An early series of 25 patients suggested that combination chemotherapy with 5-fluorouracil, doxorubicin, and cisplatin could improve 3-year survival compared with a historical cohort.56 Bassanello et al57 compared 21 post-OLT patients who received 5-fluorouracil and carboplatin with 27 who received no adjuvant chemotherapy and found no differences in overall and recurrence-free survival. Other small early series suggested improved survival with chemotherapy compared with historical series.58-61 These series usually excluded very poor prognostic indicators, such as lymph node and distant metastasis, which were included in the historical controls, thus preventing a valid assessment of the effect of chemotherapy. Some recent series that included patients with more advanced HCC suggested a benefit from doxorubicin, but no comparison groups were identified, and the patients had varying rates of known poor prognostic indicators, such as vascular invasion.58,62 In contrast, most of the studies that showed good long-term survival for patients who met the Milan and expanded criteria did not give adjuvant chemotherapy.26,28,30 Chemotherapy was variably tolerated after transplantation. Early cessation of chemotherapy in 26% to 30% and serious adverse events, including death from pneumonia, have been reported.58,62 Hepatotoxicity and much higher rates of recurrent hepatitis C have been reported.57,63 A recent report from the University of California at Los Angeles suggested that a survival benefit was associated with adjuvant chemotherapy after liver transplantation for HCV-related HCC.64
 
Overall, there is little evidence that chemotherapy has a role in the treatment of HCC after OLT. Despite the general lack of evidence, chemotherapy is often given in the setting of high-risk findings, including vascular invasion and T3B/T4 tumors. Thus, a randomized placebo-controlled trial is warranted in this setting.
 
Management before transplantation
Once a patient is listed for transplantation, there is an ongoing risk of tumor progression to the point at which transplantation is no longer possible. The time from listing to transplantation seems to be a major predictor of dropout from the transplant list because of tumor progression. For example, as the waiting time increased from 62 to 162 days, dropout rates increased from 0% to 25% in a cohort of 87 patients who did not receive preoperative treatment to slow tumor progression.24 In another series, dropout rates increased by more than 3 times for tumor size >3 cm or 3 nodules, whereas tumor ablation or chemoembolization reduced dropout by 2.5 times.65
 
Commonly used options for slowing tumor progression while patients are on the waiting list include local ablation with alcohol or radiofrequency ablation and chemoembolization. None of these strategies has been assessed against no treatment or each other for the prevention of tumor progression during the waiting period. In the nontransplantation setting, radiofrequency ablation is becoming favored over alcohol ablation because of better tumor destruction and the need for fewer treatment sessions.3 A recent case series reported that radiofrequency ablation was performed with low morbidity (8%) and no mortality in 50 patients, with 80% meeting the Milan criteria.66 With a mean waiting time of 9.5 months before OLT, there were no dropouts from the waiting list, and there was an 83% 3-year patient survival and a 4% posttransplantation recurrence rate. Another series of 23 patients treated with radiofrequency ablation as a bridge to transplantation resulted in successful transplantation in 65% after a mean waiting time of 7.9 months, but 22% developed progressive disease and died before transplantation. The survival of those who received transplants was good, at 85%, but intention-to-treat survival is much lower given the dropout rate.
 
Chemoembolization has been shown in a meta-analysis of randomized trials55 to improve survival when it is used as a primary treatment of HCC. Its use to prevent tumor progression for patients awaiting liver transplantation has been assessed only in case series. Complications arising from pre-OLT chemoembolization that lead to waiting list dropout or increased post-OLT morbidity are rare.33,67,68 Graziadei et al67 described 48 patients who met Milan criteria and received chemoembolization. They reported no waiting list dropouts and 5-year survival rates of 94% despite a mean waiting time of 178 days. Another series of 54 patients receiving pre-OLT chemoembolization resulted in a 15% dropout rate due to tumor progression at 6 months and 25% at 12 months.33 The 5-year survival after OLT was 74%, but this decreased to 61% if assessed by intention-to-treat analysis. Majno et al69 compared 54 patients who received pre-OLT chemoembolization with 57 who did not. They found no improved survival in patients who received chemoembolization, but responders to treatment did fare better than nonresponders, and there was a trend toward improved survival compared with the untreated group. When case series with and without adjuvant chemoembolization are compared, there seems to be a lower rate of waiting list dropouts with adjuvant therapy, especially for waiting times of 6 months and beyond.
 
Given that local ablation and chemoembolization are low risk, they should be considered as a means to prevent tumor progression while patients are on the waiting list, especially if the local waiting time is expected to exceed 3 to 6 months, if the tumor size exceeds 3 cm, or if multiple nodules exist. In the setting of transplantation for patients beyond Milan criteria, preoperative treatment should be strongly considered and may serve to identify patients with aggressive tumor biology who will have poor outcomes with transplantation.30,31
 
Several areas of management of HCC in the context of liver transplantation have given rise to persisting controversy, and these are discussed in more detail below.
 
Living donor liver transplantation for hepatocellular carcinoma The use of adult-to-adult living donor liver transplantation (LDLT) has increased in the United States and Europe in response to the donor organ shortage.70 Donors are typically young, healthy individuals with a significant relationship with the recipient (eg, family member or friend). The right lobe or 60% of the donor liver is taken, and both this donated portion and the 40% remaining will grow to >85% of the original volume, usually within months. Although controversies and questions surround this fledgling field, LDLT fills an important niche in liver transplantation. Patients with circumstances that are associated with higher mortality or that incur significant morbidity but that are not reflected in the prioritization system may be well served by the LDLT's ability to bypass the organ-shortage bottleneck.
 
Before the MELD-based allocation system, LDLT was considered a promising route for patients with early HCC and an appropriate living donor. Modeling studies suggested that acceptable life expectancy and cost-effectiveness could be reached if waiting times exceeded 7 months for the patient with early HCC.71 Waiting times for patients with HCC in many centers exceeded 7 months by 2001. Results from China, where low deceased donation rates and long waiting times for HCC patients persist, show that LDLT has a clear benefit over waiting for a deceased donor organ on an intention-to-treat analysis.72
 
However, the MELD-based system significantly decreased the waiting time for patients with HCC in the United States (see The Model for End-Stage Liver Disease-Based Prioritization System), thus decreasing the demand for LDLT in this population. In addition, a widely publicized donor death that occurred in early 2002 may have contributed to a dampening of enthusiasm for LDLT.73 Indeed, the number of adult-to-adult LDLTs performed in the United States declined in 2002 after increasing steadily before that. Should recent adjustments in the MELD allocation scheme for early HCC significantly increase waiting list mortality, then the demand for LDLT may increase again.
 
LDLT may still have a role for patients with HCC beyond stage T2. Such patients do not receive additional MELD points and are often excluded from liver transplantation. However, the transplant center at Mt Sinai Medical Center in New York has shown a 48% posttransplantation recurrence-free survival at 5 years for patients with tumors >5 cm.62 Some argue that this is an acceptable cancer outcome that would justify LDLT for stage III and IVa tumors.74 Arguably, LDLT could be offered if the donor and recipient are informed, willing, and appropriate candidates. Such patient and donor wishes must be weighed against the small but significant mortality (1 in 300) and morbidity risk for young, healthy donors. The engrafted right lobe may initially fail (eg, primary nonfunction). These cases are usually relisted for a deceased donor organ, thus tapping this limited resource. Moreover, third-party payers and the community at large may balk at supporting the cost of LDLT for more advanced HCC. At this time, there is no consensus on LDLT use in stage T2 or higher HCC cases. Decisions are made on a center-by-center and case-by-case basis. Data from a multicenter, prospective, National Institutes of Health-funded study aimed at determining the outcomes of LDLT donors and recipients75 may help guide the decision to perform LDLT for more advanced HCC in the future.
 
Expanded criteria
Proponents of expanding the current criteria are driven by the increasing number of HCC patients in need of treatment, the competition for scarce organs, and the observation that some patients with tumor burdens exceeding the Milan criteria do have long, disease-free survival after transplantation. Some advocate modest expansion of existing criteria (ie, size and number of tumors), whereas others suggest incorporating other parameters (eg, pretransplantation tumor histology), thus de-emphasizing tumor size and number.
 
There are data to suggest that the Milan criteria may be too conservative. Yao et al30 suggest a mild expansion of the tumor size limits on the basis of retrospective data at the University of California, San Francisco. They found that patients who had undergone transplantation with single tumors up to 6.5 cm or no more than 3 tumors with maximum sum of diameters up to 8 cm and no tumor larger than 4.5 cm had acceptable disease-free survival, similar to that of patients who met Milan criteria (Table 2). Their data were based on explant histology sizing and not on pretransplantation imaging. Nevertheless, the criteria maintained good discriminatory function when applied retrospectively to the subset with adequate pretransplantation imaging data. A follow-up study on a larger number of patients confirmed an acceptable 5-year disease-free survival of 88.5%, compared with 93.8% for those who met Milan criteria.31 Although the data on this larger cohort of HCC patients were gathered prospectively, the patients still underwent transplantation under Milan criteria. Patients who met the mildly expanded criteria were again defined retrospectively and according to explant histology. Other groups have also published retrospective data suggesting that a loosening of the Milan size criteria may not adversely affect outcomes. Herrero et al76 found no difference in posttransplantation survival between those with and without HCC when transplantation was offered to those with single lesions ≤6 cm or up to 3 tumors, none >5 cm. Recurrence-free survival was an acceptable 70% at 3 years.
 
Marsh et al27 at the University of Pittsburgh advocate criteria that include pretransplantation histological information. Their system has no preset upper limit of size or number of tumors; rather, it relies on the presence or absence of vascular invasion (microvascular and macrovascular), lymph node status, and obvious metastatic disease to define criteria for transplantation candidacy. Retrospective analysis of cases from that center shows an acceptable recurrence rate after transplantation (9.1%) when tumors up to stage IIIB are considered transplantable. Stage IIIB patients have microvascular invasion, bilobar distribution, and a largest tumor diameter of >2 cm. Staging was again based on retrospective explant histology and not on pretransplantation imaging. The Pittsburgh criteria carry the disadvantage of requiring pretransplantation histology. The same group has suggested a neuronal net decision tree that includes clinical parameters and genetic analysis of HCC tissue to predict posttransplantation recurrence. Such a shift to genetic analysis of tissue is based on the argument that size criteria are simply surrogate markers for more aggressive tumor behavior (eg, metastasis and vascular invasion) dictated by genetic changes.77
 
Despite the calls for expanding or changing the current Milan criteria, others advise caution. The present system clearly maintains acceptable posttransplantation outcomes during a period of increasing organ demand.78 The value of cadaveric livers and the need for good stewardship of this resource remain high. Pretransplantation imaging understages HCC in 20%-30% of cases.30 Some have argued that the present Milan criteria are conservative enough to allow for this understaging and that even mild expansion of the criteria would lead to unacceptable transplantation of even more advanced disease.32 Moreover, the incidence of HCC is on the rise. Screening may lead to earlier diagnosis and more transplantation candidates. Bruix et al32 argue that soon the waiting list may have so many HCC patients that exclusion, rather than inclusion, criteria will be necessary. In this case, listing priority for HCC would no longer prevent long waiting times and list dropouts. Transplant centers may be pushed to exclude those with the more likely chance of recurrence rather than trying to include those with an acceptable chance of recurrence.
 
Definitive data on expanded recipient criteria would require a shift in present UNOS policy, and some argue that the time is right for such a trial.79 Others are hopeful that data from LDLT for more advanced HCC will lend valuable insight. However, the LDLT experience is limited by small numbers of cases, and the ethics of shifting such advanced HCC cases to living donors have been questioned.19 For now, the Milan criteria remain in use. Although minor adjustment in the size criteria may occur, large changes are unlikely without significant improvement in pretransplantation imaging accuracy or a shift toward the use of histology and genetic analysis of HCC tissue.
 
Downstaging
The term downstaging is used to describe the practice of treating a person with a tumor beyond a preset size limit for liver transplantation with adjuvant therapy, usually chemoembolization, in the hope of making the tumor shrink and therefore enabling the patient to become a transplantation candidate. It is not certain whether the adjuvant treatment directly modifies the biological behavior of the tumor or whether it just helps select out which patients have less aggressive tumors.
 
Two large case series have addressed downstaging.62,69 The Mt Sinai group enrolled 80 patients with tumors >5 cm into a multimodality adjuvant therapy program in which patients received chemoembolization followed by OLT (if they did not have progressive disease) and postoperative chemotherapy.62 Forty-six percent dropped out, predominantly because of progressive disease, and the remainder underwent transplantation.62 Five-year overall and tumor-free survival were 44% and 48%, respectively, for those who reached OLT. Recurrent disease was associated with tumor size >7 cm and vascular invasion.62 Intention-to-treat survival was poor in this study, but given the advanced stage of the patients, it is perhaps as good as can be expected. The <50% 5-year survival after OLT is probably not acceptable given the current organ donor shortage, but perhaps selecting patients with tumors <7 cm and no vascular invasion could produce acceptable results.
 
Majno et al69 reported the results of treating 54 patients, at least half with tumors beyond Milan criteria, who received chemoembolization followed by transplantation and sometimes chemotherapy compared with 57 patients who were similar but did not receive chemoembolization. Downstaging, defined as a 50% reduction in the product of the perpendicular diameters of the largest tumor nodule, occurred in 51% of the chemoembolization patients. Overall there was no difference in survival between the 2 groups, but patients with tumors >3 cm who were successfully downstaged had a better tumor-free survival than those who did not respond and trended toward better tumor-free survival than those who did not receive chemoembolization. It is difficult to effectively assess whether downstaging had any real effect on outcomes, because the outcomes were not stratified by the Milan and expanded criteria.
 
On the basis of these series, downstaging remains a theoretical construct. There is not sufficient evidence that pre-OLT adjuvant therapy alters the natural behavior of HCC. It is more likely that adjuvant therapy allows discovery of biologically aggressive tumors.
 
Rescue transplantation
Because of the limited supply of donor organs and the resulting delay in transplantation, some have suggested that patients with preserved hepatic function who meet Milan criteria should undergo hepatic resection as the primary treatment of HCC, with liver transplantation as a backup. Patients who are discovered to have high-risk features, such as microscopic vascular invasion and undetected nodules, upon pathologic examination of the resected specimen could receive salvage transplantation. Alternatively, those who develop recurrence of tumor within the limits of Milan criteria could receive rescue transplantation. These strategies could spare some patients the need for liver transplantation and would reduce use of the limited donor resource. However, there is considerable controversy regarding whether this strategy is effective for individual patients and whether it ultimately could be applied to a significant number of patients. A direct evaluation is impossible because no randomized trial has been or likely will be performed to compare direct transplantation with resection and rescue. Furthermore, series used to support one strategy over the other are hard to compare because of differing waiting times for transplantation and limited stratification of results by tumor stage.
 
Early reports suggested similar survival between patients treated with resection and transplantation but included many patients in both groups who did not meet Milan criteria and included some in the resection group without cirrhosis.80-82 Clearly, highly selected patients undergoing resection can have equivalent 5-year survivals with a transplantation strategy at the expense of a much higher recurrence rate.23,83 A Markov-based decision model suggested that similar life expectancy could be achieved with resection with possible salvage transplantation vs primary OLT, with a lower use of scarce donor organs.84
 
The safety of OLT after prior surgical resection of HCC is good in some series but is associated with increased perioperative mortality in others.85,86 An Asian case series with predominantly hepatitis B-related HCC reported that 79% of their selected resection patients had recurrences eligible for rescue OLT but that only 4% went on to OLT.87 A Western series of HCV-associated HCC found that most recurrences after resection were outside of the Milan criteria.88 Other Western series confirmed less than 10% utilization rates of rescue transplantation in clinical practice.24,86
 
Planned salvage transplantation for resection patients with microscopic vascular invasion or multiple nodules was reported by the Barcelona group.89 Of the 8 patients eligible for salvage transplantation, 5 underwent transplantation, and only 1 developed recurrence. Two developed recurrence while waiting for OLT, and 2 refused OLT and later died from recurrent HCC. Thus, especially in the West, salvage and rescue transplantation are unlikely to produce survivals equivalent to those with liver transplantation by intention to treat unless wait list times are long. It is not clear how this type of salvage transplantation would fit into the current MELD scoring system in the United States.
 
To biopsy or not
A definitive diagnosis of HCC usually requires histological examination of tumor tissue. The tumor may be biopsied by one of several means. Simple percutaneous biopsy may be performed, particularly in advanced HCC, when irregularities may be palpated on the surface of an enlarged liver. More often, a needle biopsy must be performed under ultrasound or CT guidance. Occasionally, laparotomy may be necessary to accurately identify and biopsy a mass within the liver. Similarly, laparoscopy may be used, although its value is limited because HCC is so often found deep within a cirrhotic liver, so that it is not visible from the surface, in contrast to metastatic liver cancer, which is readily seen at laparoscopy. The risk of complications, particularly bleeding, after needle biopsy of HCC is greater than when other nonmalignant conditions are biopsied, presumably because of the vascular nature of the tumor.90 There are cases reported of HCC spread along the track of a needle biopsy, although the frequency of this complication91 is probably approximately 1%.
 
With needle biopsy, it is possible to obtain either a core of liver tissue or a cellular aspirate. A core biopsy specimen is necessary for histological evaluation. The diagnosis of HCC may also be made on evaluation of cytological specimens obtained by aspiration through a fine needle. Because of concerns about the risk of spreading HCC through needle biopsy, there has been some reluctance to require histological confirmation of HCC before offering treatment to patients, particularly in the case of potentially curative therapies such as resection or transplantation. For clinical purposes, a typical radiological appearance with or without increased serum a-fetoprotein may be considered sufficient to allow the patient to be listed for liver transplantation. Similarly, for research purposes, a recent EASL conference listed standard criteria for diagnosis of HCC that do not involve biopsy.38 Radiological criteria include 2 imaging techniques that both show a focal lesion greater than 2 cm in diameter with features of arterial hypervascularization or a single radiological study with these features combined with a serum a-fetoprotein level of greater than 400 ng/mL.
 
False-positive identification of HCC does occur with radiological methods, and biopsy carries the risk of tumor seeding along the needle track. Thus, Schotman et al92 described 15 cases in the literature of seeding of HCC after needle biopsy, whereas Frilling and Broelsch93 summarized the literature and reported recurrence rates of 2%-5% after needle biopsy. A study among 60 patients undergoing resection for HCC without prior biopsy indicated that in only 2 of 65 tumors was the radiological diagnosis of HCC incorrect.94 These authors do point out that the radiological false-positive rate for smaller tumors was quite high (2 of 10; 20%).
 
Because current management guidelines for HCC do not require biopsy to prove the diagnosis, we have evaluated our experience of patients with liver disease and hepatic lesions suggestive of HCC who underwent both fine-needle aspiration and core biopsy at Saint Louis University. We correlated the biopsy results with various clinical, biochemical, and radiological features.
 
HCC was diagnosed by biopsy in 74 of 118 cases (63%). The use of both fine-needle aspiration and core biopsy increased the diagnostic yield of the procedure from 54% for fine-needle aspiration and 55% for core biopsy individually to 63%. On mean follow-up of 27.5 months, an additional 10 patients were found to have HCC, whereas 26 had no further evidence of HCC. Patients with positive biopsy results had significantly higher serum a-fetoprotein levels (median, 57 μg/mL vs 12 μg/mL; P = .014) than those with negative biopsies, although the groups were otherwise comparable with regard to other tests of liver function, tumor diameter, and Child-Pugh class.
 
We compared the diagnosis of HCC from a combination of these biopsies and follow-up with standard noninvasive diagnostic criteria advocated by the EASL. Of 62 patients who met the noninvasive EASL criteria for HCC, only 50 were found to have HCC on biopsy or follow-up. Thus, 8 (13%) would have been falsely diagnosed with HCC, because follow-up data on the remaining 4 patients were incomplete. No patient developed evidence of tumor spread along the needle track after biopsy.
 
We believe, therefore, that image-guided biopsy of lesions clinically suggestive of HCC should be routinely performed to allow adequate treatment planning, because the risks of biopsy seem small and the potential benefits seem significant. Obtaining material for both cytological and histological examination at biopsy maximizes the diagnostic yield. Additionally, in the future, biopsies may be used for molecular profiling of tumors, which may be better predictors of tumor biology.
 
Summary and conclusions
In summary, a remarkable transformation has occurred in the management of patients with HCC over the last decade. It is now apparent that liver transplantation offers prolonged disease-free and overall survival for carefully selected patients with small HCCs and that there has been substantial growth in transplantation for this indication. Unfortunately, many patients present with advanced HCC that is not transplantable, or they may not be transplantation candidates because of other comorbidities. Transplantation is also not widely available in many countries where HCC is common. Screening for HCC in individuals at risk has been shown to identify HCC at an earlier stage, and this may make transplantation an even more important treatment option.
 
Considerable controversy still remains about some aspects of transplantation for HCC—controversies that must be settled convincingly, given the scarcity of donor organs available for transplantation. Clearly, further research is needed to settle these controversies, and controlled clinical trials may be required in some instances.