Background. Restrictive staging criteria for liver transplant (LT) patients with HCC in the U.S. have resulted in favorable long-term recurrence-free survival, but these criteria exclude a subgroup of patients who, despite tumor size beyond T2 stage, demonstrate an acceptable outcome. The aim of this study was to assess the waiting list and post-transplant mortality of patients with HCC tumors greater than Milan T2 stage.
Methods. The U.S. OPTN standard transplant dataset was analyzed for patients with a diagnosis of HCC who were listed for liver transplantation between February 2002 and 2008. Those patients with Milan T3 stage tumors were compared to patients with T1 and T2 lesions. Multivariate survival models were developed to investigate independent predictors of death or tumor recurrence post-transplant.
Results. 7,391 patients with HCC were identified. 351 (4.75%) had T3 lesions. Compared to non-T3 patients, total tumor burden was greater and total alpha-fetoprotein (AFP) was higher in the T3 patients. T3 patients also were more likely to receive pretransplant locoregional therapy. There were no significant differences between T3 patients and non-T3 patients in demographic variables or physiologic MELD score at the time of transplant, waiting time, or donor risk index. Waiting list mortality was increased for T3 patients compared to non-T3 and tumor progression while waiting was higher. Independent predictors of waiting list mortality included physiologic MELD score at the time of listing, total tumor burden, and serum AFP. There was significant regional variation in the utilization of exceptions for T3 patients and UNOS regions 4, 9, and 10 performed a higher percentage of their transplants in T3 patients compared to other regions. There was no difference in post transplant survival between T3 and non-T3 patients. Independent predictors of post-transplant mortality included physiologic MELD score at the time of transplant, recipient age, and donor risk index. In patients with T3 tumors, total tumor burden was not an independent predictor of post transplant survival.
Conclusions. Patients who are listed for liver transplantation with Milan stage T3 HCC have higher waiting list mor-tality but have similar post-transplant survival compared to patients with T1 and T2 HCC.
Liver transplantation (LT) is the optimal treatment of hepatocellular carcinoma (HCC) in patients with underlying liver disease as it eliminates both the cancer and the cirrhotic liver. Early experience with LT for cancer yielded disappointing results having a high rate of recurrence and an unacceptably low survival rate. This was thought to result from poor selection of candidates with advanced disease; however, there were some long term survivors.1,2 Mazzaferro, et al., reported a landmark study in 1996 demonstrating an excellent post-transplant outcome which they attributed to their patient selection. As a result, the “Milan criteria” were developed including a single lesion less than or equal to 5 cm diameter, or up to three lesions all with diameter less than 3 cm.3Patients in this group had an excellent 4-year survival rate post-transplant of 75%, similar to patient survival rates with nonma-lignant liver diseases. Following this report, experiences from other institutions showed a 5-year survival rate of 75% or greater for patients fulfilling these criteria.4,5 Currently, the Milan criteria have become widely accepted for selection of LT candidates and they are incorporated into the pre-trans-plant evaluation process for patients with HCC in many transplant centers. Transplant candidates within Milan criteria receive standard MELD exception points in the United States.6
The Milan criteria have been challenged in recent years by many centers as being too conservative. Increasing experience from single center series of LT for HCC have suggested that the selection criteria may be expanded without compromising long-term outcomes.7Furthermore, some patients with advanced HCC can also be aggressively treated with LT following locoregional neoadjuvant therapy.8–10 In this method, down staging of advanced HCCs that initially do not meet the Milan criteria but which fulfill the criteria following treatment and remain stable raise a particular interest. Several groups have demonstrated that tumor down staging in LT recipients has good results.8–16 Various centers have used different treatment options and adopted inconsistent definitions of tumor down staging, typically with a small number of patients. In theory, achieving successful down staging may reflect more favorable tumor biology associated with low recurrence potential. However, not all patients in this category have acceptable post-LT outcomes,15–16 indicating that there may be additional parameters other than tumor dimensions and stability post treatment that affect outcome. Therefore, there is a crucial need for a better understanding of the disease course after transplant for tumors outside standard criteria. To address these issues, the aim of this study was to determine wait list survival and post transplant mortality and factors of predictive of survival with tumors outside standard criteria using a large U.S. national dataset.
MethodsThe United States Organ Procurement and Transplantation Network database was analyzed for patients listed for liver transplantation with hepato-cellular carcinoma between February 2002 and January 2009. This analysis includes all patients with an HCC reported to United Network for Organ Sharing (UNOS) and eventually listed for liver transplantation. Not all patients included in this study necessarily received a MELD exception for HCC. For instance, some patients might have been listed with a primary indication of hepatitis C and had a MELD score high enough to enable liver transplantation without MELD exception points for HCC. In the U.S., UNOS uses a modified version of the Milan criteria3to establish transplantation priority for patients with HCC.17Patients were considered to be stage T3 if they had one lesion greater than 5 cm, or 2-3 nodules, at least one greater than 3 cm in size. Pre-transplant tumor staging is based on radiographic tumor appearance, not explant findings. The decision to assign transplant priority to patients with T3 lesions is decided at the regional level in the U.S. and is determined by a review board of transplant physicians and surgeons.
In this analysis, patients with T3 lesions were compared to patients with non-T3 lesions. Non-hepa-tocellular carcinoma liver tumors were excluded (cholangiocarcinoma, hepatoblastoma, metastatic carcinoma). Patients were also categorized by total number of lesions and by cumulative total tumor size. Locoregional interventions performed on patients with HCC are also routinely reported to UNOS as part of the listing process in patients with HCC. For this analysis, the types of locoregional intervention performed were categorized into the following:
- •
None
- •
Transarterial chemoembolization (TACE) or radioactive bead embolization.
- •
Radiofrequency ablation (RFA) or cryoablation.
Tumor size estimates were based on measurements prior to locoregional therapies.
Patient demographics and other patient characteristics were compared between groups. Univariate and multivariate waiting list and post-transplant survival models were developed to investigate independent predictors of waiting list or post-transplant death. Waiting list removals for “death”, “condition deteriorated” or “too sick to transplant” were considered waiting list deaths. Reason for post-transplant mortality was generally not available in the dataset, although recipient status is routinely verified through the U.S. Social Security National Death Registry. Categorical variables were analyzed using the chi-square or Fisher’s exact test where appropriate. Continuous variables were analyzed using analysis of variance, independent sample t-test or Wilcoxon sign rank test as appropriate. Univa-riate survival modeling was performed using the Kaplan-Meier technique and the log-rank test with censoring at death, transplant, or last follow-up. Multivariate adjusted survival models were constructed using the Cox proportional hazards technique with censoring at death or last follow-up. Candidate variables were included in the multivariate survival models if they were statistically significant to a level of 0.20 in the univariate analysis or have been shown in other published models or clinical experience to be significant predictors of survival. No data imputation was performed. All statistical testing was two sided with the level of statistical significance for type one error set at p less than or equal to 0.05. All dataset manipulation and statistical analysis was performed using SAS (Cary, NC, ver-sion 9.1.3). No institutional review board approval was needed for the use of this deidentified dataset.
Results7,391 patients with HCC were identified. 351 (4.75%) had T3 lesions. As expected, patients were predominantly male and Caucasian. There were no differences between patients with T3 lesions and those with non-T3 lesions in mean age (56.7 in T3 patients vs. 56.0 years in non-T3 patients, p = 0.228), gender (81.5% male vs. 77.1%, p = 0.057), race (65.8% Caucasian vs. 61.6%, p = 0.113), MELD score at transplant (14.0 vs. 13.7, p = 0.503), HCV infection (21.4% vs. 25.8%, p = 0.063), or time spent on the transplant waiting list (199 days vs. 195, p = 0.720). Mean donor age was marginally higher in the T3 patients (44.8 years vs. 41.8, p = 0.007) but the overall donor quality as assessed by the donor risk index (DRI)18was similar between groups (1.85 vs. 1.81, p = 0.104). See table 1 for details.
Patient population and tumor characteristics. Continuous variables are expressed as means and 95% confidence inter¬vals. Categorical variables are expressed as number and percent.
Variable | T3 Patients (n = 351) | Non-T3 Patients (n = 7,040) | P-value |
---|---|---|---|
Recipient age (years) | 56.7 (55.7-57.7) | 56.0 (55.9-56.3) | 0.228 |
Recipient male | 286 (81.5) | 5430 (77.1) | 0.057 |
Recipient Caucasian | 231 (65.8) | 4337 (61.6) | 0.113 |
MELD score at transplant | 14.0 (13.3-14.7) | 13.7 (13.6-13.9) | 0.503 |
Hepatitis C infection | 75 (21.4) | 1816 (25.8) | 0.063 |
Waiting time (days) | 199 (172-226) | 195 (189-201) | 0.720 |
Donor age (years) | 44.8 (42.7-47.0) | 41.8 (41.3-42.5) | 0.007 |
Donor risk index* | 1.85 (1.80-1.91) | 1.81 (1.80-1.82) | 0.104 |
Total tumor burden (cm) | 6.91 (6.78-7.05) | 3.10 (3.07-3.13) | <0.0001 |
Number of lesions | 2.23 (2.17-2.30) | 1.36 (1.35-1.38) | <0.0001 |
Serum alpha-fetoprotein (ng/mL) | 864 (588-1140) | 340 (278-401) | 0.0003 |
Pre-op chemoembolization | 159 (45.3) | 1699 (24.1) | <0.0001 |
Pre-op percutaneous ablation | 73 (20.8) | 914 (13.0) | <0.0001 |
No pre-op intervention | 137 (39.0) | 4568 (64.9) | <0.0001 |
* As defined by Feng, et al. See text.
Because of the definition of the groups, there were significant differences in the characteristics of the tumors between groups. The total tumor burden (6.91 cm in T3 patients vs. 3.10 in non-T3 patients, p <0.0001), mean number of lesions (2.23 vs. 1.36, p <0.0001), serum AFP (864 ng/mL vs. 340, p = 0.0003), and percentage of patients undergoing no pre-transplant tumor locoregional intervention (39.0% vs. 64.9%, p <0.0001) were all significantly different between T3 patients and non-T3 patients (Table 1). Amongst the UNOS regions, there was significant variation in listing practices for candidates with T3 lesions. Regions 9 (New York, n = 79, 22.5%), 5 (California, Nevada, Utah, Arizona, and New Mexico, n = 76, 21.7%), and 4 (Texas and Oklahoma, n = 72, 20.5%) listed more than 60% of all candidates with T3 tumors. T3 candidates accounted for 12.2% of all regional candidates with HCC within Region 4, 7.2% in Region 9, and 7.1% in Region 10 (Michigan, Indiana, and Ohio). These regions were significantly different compared to all other regions (p <0.0001).
Waiting list mortality was increased for T3 patients compared to non-T3 (18.5 vs. 11.3%, p <0.0001). Figure 1 shows the unadjusted Kaplan-Meier waiting list survival for the population. Ninety-day survival for candidates with T3 tumors was 94.2% compared to 96.7% for non-T3 candidates (p = 0.0006). This survival disadvantage continued at one year with 71.3% for T3 candidates and 74.6% for non-T3 patients (p = 0.0006). In the multivaria-te survival model, independent predictors for increased waiting list mortality included MELD score at listing (HR 1.126, 95% CI 1.108-1.144, p <0.00001), total tumor burden (HR 1.113 per cm of tumor, 1.069-1.158, p <0.0001), serum AFP (HR 1.001 per unit change in AFP, 1.000-1.002, p <0.0001). See table 2 for details.
Independent predictors of adjusted waiting list and adjusted post-transplant mortality.
Variable | Hazard Ratio | Interval | 95% Confidence p-value |
---|---|---|---|
Waiting List Mortality | |||
MELD score at listing | 1.126 | 1.108 - 1.144 | <0.0001 |
Any pre-op tumor intervention | 0.904 | 0.774 - 1.056 | 0.203 |
Total tumor burden (cm) | 1.113 | 1.069 - 1.158 | <0.0001 |
Serum alpha-fetoprotein (ng/mL) | 1.001 | 1.000 - 1.002 | <0.0001 |
Post-Transplant Mortality | |||
Recipient age (years) | 1.012 | 1.004 - 1.019 | 0.002 |
Hepatitis C infection | 1.040 | 0.924 - 1.171 | 0.512 |
MELD score at transplant | 1.024 | 1.016 - 1.033 | <0.0001 |
Any pre-op tumor intervention | 0.981 | 0.873 - 1.103 | 0.751 |
Total tumor burden | 1.022 | 0.983 - 1.063 | 0.277 |
Donor risk index* | 1.473 | 1.301 - 1.668 | <0.0001 |
* As defined by Feng, et al. See text.
Despite the differences in waiting list mortality, there was no difference in post transplant survival between T3 and non-T3 patients (p = 0.926). Figure 2 shows the unadjusted survival curves for the T3 patients vs. the non-T3 patients. Independent pre-dictors of post-transplant mortality (Table 2) included physiologic MELD score at the time of transplant (HR 1.024, 1.016-1.033, p <0.0001), recipient age (HR 1.012 per year, 1.004-1.019, p = 0.002), and donor risk index (HR 1.473, 1.301-1.668, p <0.0001. Total pretransplant tumor burden was not an independent predictor of post transplant survival (p = 0.277).
DiscussionWith recent advances in therapeutic options and surveillance for HCC, more patients with large HCC have come to be evaluated for transplantation. Data suggest that there is a select group of patients with advanced stage HCC that do well with liver transplantation. There is also evidence that some tumors which are initially outside of standard transplant criteria may be successfully down staged pre-opera-tively and have similar outcomes.7–13,19,20 The challenge is to determine if expanded criteria and down staging offers a similar long-term survival to those candidates with smaller tumors in this era of organ shortage. In light of the success in tumor down staging demonstrated by UCSF,7,9,10,13 many centers have pushed forward with this type of protocol asking for MELD exception points on candidates outside standard accepted Milan criteria who show stability after locoregional down staging.9–13 Currently, review boards consider these situations on a case by case basis. Those tumors that can be down staged into Milan T2 criteria and have documented stability for three months are typically given the same exception points that standard T2 tumors receive. Tumors outside the standard criteria may also receive exception points or, in some cases, be transplanted based upon native MELD score.
In this analysis, patients with T3 lesions were compared to patients with non-T3 lesions and found to have similar outcomes after transplantation. It is somewhat challenging to determine which patients underwent a down staging treatment prior to transplant or not due to the nature of this database analysis. However patients who survived to transplant did as well regardless of tumor stage at listing. Whether or not pre-transplant locoregional therapy is beneficial is still controversial. Some have argued that pre-operative locoregional down staging have not demonstrated a significant effect on survival which is dramatically demonstrated in those waiting greater than 6 months for trans-plantation.21Recently Northup, et al., presented data demonstrating an increased list removal rate with locoregional therapy in patients with T1 lesions compared to T2 lesions.22There was not a significant survival advantage to locoregional therapy in patients with T2 lesions post transplant. Moreover, at a national consensus conference on liver allocation for patients with HCC, a work group reviewed locoregional therapy and concluded that devices and image guidance techniques are constantly improving and the true complete ablation rate will likely improve with advancing technologies.23Despite this, locoregional therapy should currently be viewed as a bridge to rather than a replacement for transplantation.
Locoregional interventions performed on patients with HCC are also routinely reported to UNOS as part of the listing process in patients with HCC. Non-T3 patients were more likely to not have pre-transplant locoregional therapy as compared to T3 patients (64.9% vs. 39.0%). Furthermore, it was interesting to discover that only 65% of patients with T3 tumors had pre-transplant locoregional intervention. Despite the favorable tumor biology expected in successfully down staged patients, previously published data have yielded inconsistent results in relation to post-transplant outcomes.10–16 This highlights the need for improved understanding of the long-term clinical course of down staged HCC. Identifying prognostic factors would establish the basis for which we elect LT for selected down staged patients.
Regional variation is difficult to account for within this database analysis. Regions that tend to transplant at higher MELD scores also had a higher incidence of T3 tumors. Again, it could be postulated that these cases were transplanted on standard MELD score vs. exception. Regions 9 (New York, n = 79, 22.5%), 5 (California, Nevada, Utah, Arizona, and New Mexico, n = 76, 21.7%), and 4 (Texas and Oklahoma, n = 72, 20.5%) listed more than 60% of all candidates with T3 tumors. T3 candidates accounted for 12.2% of all regional candidates with HCC within Region 4, 7.2% in Region 9, and 7.1% in Region 10 (Michigan, Indiana, and Ohio). These regions were significantly different that all other regions (p <0.0001). The regional variation in transplant practice in the U.S. is of great debate as supply and demand are not equally distributed throughout the country. The current allocation system is being evaluated for alterations to allow for more equitable organ distribution more equitable based on population and geography.
Waiting list mortality or removal from the waiting list was expected to be higher in the T3 patients. Waiting list removals for “death”, “condition deteriorated” or “too sick to transplant” were considered as waiting list deaths in the analysis. Waiting list mortality was increased for T3 patients compared to non-T3. This waiting list death and removal accounted for a significant decrease in overall survival at 90 days and one year from the time of listing. However, those patients that survived to transplant had no survival difference between T3 and non-T3 patients. In the mul-tivariate wait list survival model, independent predictors for increased probability of wait list removal included MELD score at listing, total tumor burden and serum AFP. These results suggest sicker patients, and those with larger and more aggressive tumors, have a higher rate of waiting list removal. AFP continues to be a significant predictor of poor outcome in tumor patients.
There are inherent weaknesses of large data set analyses such as incomplete reporting to UNOS, unclear cause of death in the dataset and incomplete reporting of recurrent tumors post-transplant, relatively short follow-up period, and no justifiable accounting for tumor patients that are never listed. There is also a lack of granularity to show details about explant pathology or the total denominator of HCC patients undergoing transplant and these weakness in the OPTN database have been discussed elsewhere.24Finally, mortality is a poor endpo-int for many disease processes as the true cause of mortality is multifactorial in most cases. Other factors such as local standard of care, regional surgical variation, and unnamed factors are likely to contribute greatly to overall survival. Despite these obvious weaknesses, this study offers the largest sample size of well defined T3 HCC patients with extended follow-up published to date.
The survival model post transplant was strikingly different from the pre-transplant model. It revealed that the MELD score at the time of transplant was an independent predictor of post-transplant mortality as was recipient age and donor risk index. These findings confirm previously published single center studies. Interestingly, total pretransplant tumor burden was not an independent predictor of post transplant survival. This suggests that patients who have tumors outside of the UNOS (Milan) T2 criteria who successfully navigate the transplant process have similar survival to candidates with lower tumor burden pre-transplant. Our results support the assertion that transplant centers and the various regional review board methods allow select patients outside standard criteria to be transplanted with good results. Whether these patients are transplanted on their MELD score, through regional review board exception points, or tumor down staging protocols, these data demonstrated that there is a role for expanding the listing criteria with demonstrated good outcomes. Unfortunately defining the expanded criteria is difficult for T3 and greater lesions with the present data. Further work is needed to develop improved and more specific pre-transplant predictive models for aggressive tumors such as serum markers, tumor genetic profiles, and improved radiographic analysis in order to establish if there a role for pre-and post-transplant chemoprophylaxis for high risk tumors given the advent of new chemotherapeutic agents. The current data demonstrates further the need for a down staging waiting period in patients who have tumors outside standard criteria in order to identify aggressive tumors until better methods of assessing tumor activity are available.
ConclusionAnalysis of the US MELD-era data shows that patients who are listed for liver transplantation with Milan stage T3 HCC have increased waiting list mortality but have similar post-transplant survival rates compared to patients with stages T1 and T2 lesions and provides further evidence that criteria for transplantation for HCC should be liberalized, probably through the use of down staging waiting periods and improved detection methods of aggressive tumors.
Abbreviations:- •
AFP: Alpha fetoprotein.
- •
HCC: Hepatocellular carcinoma.
- •
HCV: Hepatitis C virus.
- •
OPTN: Organ Procurement and Transplantation Network.
- •
RFA: Radiofrequency ablation.
- •
TACE: Transarterial chemoembolization.
- •
UNOS: United Network for Organ Sharing.
This work was supported in part by Health Resources and Services Administration contract 234-2005-370011C. The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.