Neoadjuvant therapies for hepatocellular carcinoma before liver transplantation: A critical appraisal

Hepatocellular carcinoma (HCC) is a major health problem, being the fifth most common cancer worldwide.1 The incidence of HCC is increasing in Europe and the United States,2 and it is currently the leading cause of death among patients with cirrhosis.3 The advent of surveillance programs has led to a change in the stage of tumors detected. In more than half of the cases, these tumors will be suitable for potentially curative treatments, such as resection, transplantation, and percutaneous ablation.4, 5 Resection, liver transplantation (LT), and percutaneous ablation achieve high rates of complete response in well-selected candidates4 and prolonged survival when compared with untreated patients (5-year survival of 40-70% vs. <20%).5 Specifically, resection6-15 and LT16-22 compete as the first-line options because they achieve the best outcomes in patients with small HCC (5-year survival of 60-70%; Table 1). Resection has been endorsed as a first-line option by the guidelines of management of HCC from European Association for the Study of the Liver (EASL) and American Association for the Study of Liver Diseases (AASLD) in patients with solitary tumors and well-preserved liver function.4, 23 The main drawback of resection is tumor recurrence, with no adjuvant treatment currently available that has been proven effective. There is no question that LT is the first option for patients with functional liver impairment (Child-Pugh's class B-C patients) and early tumors, or with small multinodular tumors (2-3 nodules ≤3 cm).5 It treats both the neoplastic and the underlying liver disease, and in properly selected candidates, it has a low recurrence rate (<10-20% at 5 years).8, 6-19 From an intention-to-treat perspective, however, survival is decreased because of dropouts from the waiting list due to death or tumor progression.8 Table 1 lists the ultimate effect of transplant survival for HCC analyzed according to this principle. Overall, 5-year survival decreases by 10-20% (from 81-58% to 62-47%) for waiting times of 6-12 months, and dropout rates range 10-30%.24 Living donor LT is an alternative option when a long wait is expected for cadaveric LT. The outcomes after living donor LT in patients with HCC within the Milan criteria are similar than those of cadaveric LT, although more consistent data are still awaited25-27 (Table 1). For unresectable early tumors, percutaneous ablation provides good results, with 5-year survival of 40-50%.4, 5 Recently, radiofrequency ablation (RFA) has shown to provide better local control of the disease compared with ethanol injection.28 The remaining treatments have been assessed in <100 randomized, controlled trials, and only transarterial chemoembolization (TACE) has shown to modestly improve survival in well-selected patients.28, 29 There is no first-line treatment for HCC at advanced stages, a unique situation in oncology. Systemic chemotherapy is associated with 10% partial response rate, no survival benefit, and high incidence of side effects. Several other treatments have not been demonstrated to improve survival (immunotherapy, internal radiation, tamoxifen, antiandrogen agents) or are currently under investigation (tyrosine kinase inhibitors, antiangiogenic and antiproliferative agents, gene therapy).29 HCC, hepatocellular carcinoma; LT, liver transplantation; TACE, transarterial chemoembolization; RFA, radiofrequency ablation; UNOS, United Network for Organ Sharing; AFP, alfa-fetoprotein. The role of LT as treatment of HCC has evolved. The earliest experiences during the 1980s had poor outcomes (recurrence rate of 32-54% and 5-year survival <40%), reflecting the inclusion of advanced-stage tumors.30, 31 These unfavorable experiences were useful in characterizing the factors leading to poor outcome, and they pointed out that patients with small, asymptomatic tumors and incidental tumors had similar outcomes as patients with nonmalignant disease.32 Afterwards, a series of LT selecting these “optimal candidates”—that is, patients with single HCC ≤5 cm or up to 3 nodules ≤3 cm in diameter—reported a 70% 5-year survival with a recurrence rate of <15%.8, 16-18 In a 1996 seminal article, Mazzaferro et al.16 proposed these so-called Milan criteria, which are currently considered to be the gold standard for selection of patients with HCC for LT. These criteria have been validated by several studies, including >1,000 patients (Table 1).8, 17-19 Nowadays, this definition has been adopted by the United Network for Organ Sharing (UNOS) to prioritize patients listed with HCC in the United States. More recently, several studies have suggested that the expansion of Milan criteria does not adversely affect survival.31, 33-37 Expansion of criteria will be further discussed below. Several attempts have been made to prevent tumor progression during waiting time by applying adjuvant therapies, mainly percutaneous ablation and TACE.16, 38-44 These therapies have been tested only in the setting of observational studies that provide limited information because of the heterogeneity of patient and tumor characteristics, variable waiting times, use of different treatment modalities, variable evaluation of response, and lack of consensus about criteria of dropout.16, 38-44 We will review the evaluation of response and safety of the available treatments and their effect in relevant outcomes: dropout, disease recurrence, and survival. There are no large studies specifically designed to evaluate the response of the adjuvant treatments for HCC. Moreover, the data accumulated so far is heterogeneous because response has been analyzed through different strategies (histology and by imaging techniques). It has been assumed that the results obtained after HCC treatment in patients listed for LT may be similar to those obtained in the nontransplantation scenario. This approach has several pitfalls. First, the main aim of treating patients in the waiting list is to delay progression and to prevent dropout, whereas the aim of primary treatments is to cure the disease. This explains why the number of sessions and treatments are substantially different between both scenarios. Second, most of the studies reporting treatment response have shown a mismatch between histological examination and imaging techniques in terms of the rate of necrosis, as happens with the conventional HCC staging before transplantation. The response assessment can be based on pathologic examination. Sometimes, the imaging assessment of response was performed long before the transplant procedure. This gap favors the possibility that minor nests of malignant cells not detectable by conventional imaging techniques may have enough time to become detectable by pathology. In a prospective study including 50 patients treated by RFA, Mazzaferro et al.43 found 70% of complete necrosis by computed tomographic scan and 55% by histology. Pompili et al.45 reported complete response in 75% of 40 HCC nodules treated by RFA, percutaneous ethanol injection, or a combination of both; only 50% of these nodules had total necrosis at the pathologic examination. In another retrospective study, radiographic local tumor control was obtained in 85% of the nodules treated by RFA, but the explanted livers showed complete necrosis in only 65%.44 Other studies showed the same trend.46, 47 The size of the nodule is the most relevant predictor of tumor necrosis. Tumors <30 mm in size achieved 53-78% complete necrosis, vs. 14-39% in tumors >30 mm.43-45 Data regarding which of the adjuvant treatments has a better profile in terms of histological response are scarce, and there are no controlled studies comparing different treatments in patients listed for LT. All the information comes from case series, case-control studies, and cohort studies; the results show that RFA achieves the highest rates of complete necrosis (12-55%)43-45 compared with TACE (22-29%).8, 42, 47, 48 A combination of treatments appears promising in this setting, with complete response of 33%5 up to 66%.2, 45, 47 In summary, there is evidence suggesting that imaging techniques usually overestimate the degree of complete necrosis of HCC in patients listed for LT. Complete necrosis is best achieved with percutaneous ablation in tumors <30 mm in diameter. The safety profile of adjuvant treatments of HCC is consistent with the data reported for these treatments in other scenarios. The rate of major complications related to percutaneous ablation is reported to be of 3.9-8% the cases,43-45 and complications include acute peritonitis, biliary strictures, and liver failure. Needle-track seeding has been described as a complication of radiofrequency in high-risk patients (high levels of alfa-fetoprotein [AFP], location of the tumor in the surface),49 but it has not been extensively reported in the pretransplantation setting.47 Conversely, treatment-related complications after TACE account for 5% of the cases (including liver decompensation),50 leading to treatment-related death due to liver failure in 2.5% of cases.34 It has to be taken into account that the accepted treatment-related death rate for chemoembolization in patients with intermediate HCC is approximately 4%.51 Therefore, from the data available so far, percutaneous ablation and chemoembolization show similar complication rates that are acceptable for a neoadjuvant therapy. Historical data suggest that the median tumor doubling time in HCC is 3-6 months.52 Therefore, it is intuitive that longer waiting times will relate to higher dropout rates. In fact, in the early 1990s, the dropout rate was marginal in almost all transplant programs, when the waiting times were <6 months.53 We reported the seminal article pointing out the importance of the intention-to-treat analysis in the assessment of outcomes for HCC and LT.8 After its publication, other groups reported similar results, pinpointing an actuarial probability of dropout at 1 year of 15-30%. Overall, this type of assessment has led to an actuarial decrease in overall 5-year survival of 10-20% (from 58-81% to 47-62%; Table 1).24 Since the implementation of the Model for End-Stage Liver Disease score allocation system in February 2002, the waiting time of patients with HCC in United States has greatly diminished. However, waiting times exceeding 1 year are still the rule in some regions,54-56 favoring high dropout rates (20-30%) and modifying the outcomes after transplantation.8, 57-59 Following the concept that the benefit of LT for HCC depends on the time spent on the waiting list, UNOS gives priority to patients with HCC that meet the Milan criteria in tumors ≥2 cm in diameter. The validity of this priority policy will require an extensive analysis. At present, there are no uniform criteria to exclude patients from the waiting list on the basis of tumor progression. The Barcelona group excluded patients with either evidence of macrovascular invasion or extrahepatic spread or with uncontrolled symptomatic tumors.8 Other groups shared this exclusion criteria.42, 43 In United States, progression beyond the Milan criteria represents de facto exclusion because priority is required in order to obtain a cadaver organ.46, 50, 57, 60 Different factors have been evaluated as predictors of dropout. Baseline AFP levels >200 ng/mL, tumors >3 cm, and tumor kinetics (50% increased in size within 6 months) were identified as the main predictors of dropout.61 In other studies, an increased AFP level (cutoff 100-400 ng/mL), serum des-γ-carboxy prothrombin level >100 mAU/mL, tumor size >3 cm, and multicentricity were also identified as predictors of dropout.46, 58, 60 These factors might provide the basis of a rational priority policy. Treatment of HCC to reduce dropouts has become an appealing strategy in most transplant centers (Fig. 1). Other strategies to reduce the dropout rate include expanding the pool of donors through public health campaigns or by the use of marginal donors, and also through living donor LT. From initial studies reporting dropout rates, an actuarial probability of 15-30% at 1 year was established.8, 58 There are no randomized studies assessing the effect of neoadjuvant therapies regarding this end point. Among the case series and cohort studies reported, some investigations suggest that treatment decreases the dropout rate. This is the case of the study of Mazzaferro et al.,43 in which no dropouts were reported in their series of 50 patients whose disease met the Milan criteria (mean size 27.5 mm; 41 single lesions) treated with a single session of RFA. The heterogeneity of the data is reflected by the fact that some studies reported no dropouts in patients whose disease met the Milan criteria, who were treated by TACE, and who had short waiting times (178 days),42 whereas others reported a cumulative probability of dropout of 15% at 6 months and 25% at 12 months with the same treatment but longer waiting times (211 days).57 Strategies to prevent dropout of patients with HCC on the waiting list for transplantation. Adapted with permission from Llovet et al.24 It is estimated that dropout rates will increase along with the expansion of selection criteria. In a prospective study, Roayaie et al.34 reported a dropout rate of 46% (29% due to tumor progression) in a series of 80 patients with HCC beyond the Milan criteria who were listed for LT and who were treated with preoperative TACE and systemic doxorubicin. Other retrospective studies reported dropout rates of 3-20% in patients treated with TACE50 or multimodality palliative therapy,46 with no clear evidence of benefit in outcome measures.46 Mathematical modeling has also been used to simulate the dropout from the waiting list, but the translation of this model to clinical practice is questionable.62 In summary, only few heterogeneous, uncontrolled studies suggested a slight decrease in the dropout rate in the cohort of patients whose disease met the Milan criteria who were mainly treated with RFA. For larger tumors, the effect of treatments is even more uncertain. Because treatments on the waiting list have been applied in an uncontrolled fashion, their effects on survival after LT are difficult to assess (Table 2). In one of the most informative articles, Majno et al.40 reported the results of applying neoadjuvant chemoembolization followed by adjuvant chemotherapy. Although no differences were observed between treated patients (chemoembolization, n = 54) and untreated patients (n = 57), with 5-year survival of 55 vs. 62%, respectively, the authors reported that response to treatment provided an advantage in disease-free survival. Since that report, other studies published a 5-year survival rates ranging from 44 to >90%. This wide range reflects the heterogeneous nature of the tumors treated and transplanted in these studies. When analyzing patients with tumors within the Milan criteria, most series have reported predictably good results, with 5-year survival >70% with TACE and 83-88% survival at 3 years with RFA, TACE, or combined treatment.47, 42, 62 It is difficult to determine how adjuvant treatments contributed to these outcomes because similar results have been reported in patients undergoing LT for early tumors without adjuvant therapy.18 This has been exemplified in 2 case-control studies including index treated cases and matched controls from the United States and France. In the French study (n = 200 patients), the 5-year survival was of 59% from cases and controls,48 whereas in the U.S. study (n = 64 patients), the 3-year survival was similar, approximately 90%.47 Regarding outcomes in patients with tumors beyond the Milan criteria, a study conducted in Mount Sinai reported a 5-year survival of 44% with an overall 5-year intention-to-treat survival of 25%.34 Graziadei et al.42 found similar results. These studies confirm that transplanting patients whose disease is beyond the Milan criteria will result in far lower survival rates than in patients whose disease is within the Milan criteria, regardless of antitumoral response. In the series of Roayaie et al.,34 however, a 5-year survival rate >50% was achieved in patients with tumors 5-7 cm in diameter without vascular invasion. These data highlight the need for more refined tools to select patients beyond the Milan criteria—for instance, molecular signatures. Recurrence rates are more related to tumor stage than to adjuvant therapies. It is well known that recurrence rates are low when applying the Milan criteria,8, 43, 50 whereas they are high with a wide selection of candidates, as has been reported by several authors (>30-50% recurrence rate).34, 42, 43 We have recently reviewed the topics of expansion of criteria and downstaging to Milan criteria elsewhere.63, 64 Despite the obvious donor shortage, an increasing number of studies propose that expansion of the conventional criteria or downstaging to the Milan criteria may not adversely affect patient survival.27, 33, 35, 37 It is true that a subgroup of patients whose disease is beyond the Milan criteria may have outcomes similar to those whose disease was within the Milan criteria. However, we are still unable to identify this subgroup of patients by radiologic methods. Two studies from Europe proposed a modest expansion of the criteria on the basis of either tumor size (single ≤6 cm or up to 3 tumors ≤5 cm)33 or exclusion of patients with poor degree of differentiation, regardless of tumor size.37 The major drawback of these studies is that only few patients whose disease was beyond the Milan criteria were included (n = 12 and 13, respectively). This limitation may explain why this variable was not identified as a prognostic predictor in the multivariate analysis. It is crucial to define the effect of expansion criteria on dropout rates as a result of selecting more advanced tumors. A retrospective study based on posttransplantation pathology of 70 patients with HCC from the University of California–San Francisco (UCSF) undergoing LT proposed a new tumor classification (1 tumor ≤ 6.5 cm or up to 3 tumors ≤ 4.5 and total tumor diameter ≤8 cm).27 The overall survival of the patients meeting the new criteria (n = 60) was 90% at 1 year and 75% at 5 years—far better than those outside the criteria. Again, the proportion of patients meeting the expanded criteria was small, approximately 5% of all enlisted.22, 35 The major concerns of this proposal are the lack of specific data on overall survival and dropout rate on the waiting list for the patients whose disease is outside the current criteria but fulfills the expanded criteria.27 Downstaging refers to a change in tumor stage as a result of treatment so that disease will reach the Milan criteria, as assessed by imaging techniques. In the seminal study from Majno et al.,40 TACE induced downstaging of HCC >3 cm and resulted in good survival rates after LT, but the benefit was not confirmed in other investigations that studied TACE or RFA.41, 43, 44 The UCSF group reported a cohort of 30 patients heterogeneously treated so that disease would reach the Milan criteria.65 Half the patients with downstaged tumors were effectively transplanted, although we agree with the editorial accompanying in the sources of bias detected, and in the urgent need of controlled investigations. Only few prospective studies have assessed the effect of treatments on the dropout rate and survival after LT,42, 43, 66 with contradictory results; one study even suggests that neoadjuvant treatments might result in worse survival.41 With lengthening waiting times, treatment of HCC while the patient is on the waiting list has become an appealing option at most transplant centers. However, at present, there is no evidence of any clear effect in preventing dropout or HCC recurrence after LT, or any survival benefit for these procedures, and thus, randomized, controlled trials are needed to assess whether neoadjuvant therapies might decrease the dropout rate or prevent recurrences after transplantation.40 Ideally, these studies should compare active treatments vs. conservative management. In that regard, cost-effectiveness analysis performed by the Markov model suggested a survival advantage of 5 months for adjuvant therapies that used resection or percutaneous treatment compared with conservative treatments.67, 68 However, because of the nature of the randomization and the widespread use of neoadjuvant therapies in the transplant community, it is unlikely that sufficient number of centers and patients could be engaged to conduct such a trial. Alternatively, the 2 most widely used treatments (chemoembolization and percutaneous ablation) can be compared. Considering dropout rate as a primary end point, a 1-year dropout rate for percutaneous ablation of 10-15% might be estimated, compared with 25-30% in the chemoembolization arm. It is highly recommended to include patients whose disease falls within the Milan criteria and who have well-preserved liver function. Stratification before randomization is crucial and should be based on Child-Pugh class, baseline size, and AFP levels. Because of the characteristics of the study and sample size required (>200 patients), a multinational effort is required. Until then, no reliable data will help the scientific community unravel the efficacy and cost-effectiveness of adjuvant therapies.