Waitlist Priority for Hepatocellular Carcinoma Beyond Milan Criteria: A Potentially Appropriate Decision Without a Structured Approach



Due to the risk of waitlist dropout from tumor progression, liver transplant candidates with hepatocellular carcinoma (HCC) within Milan criteria (MC) receive standardized exception points. An expansion of this process to candidates with HCC beyond MC has been proposed, though it remains controversial. This study sought to better define the utilization of exception points in candidates with HCC beyond MC and the associated outcomes. We reviewed all nonstandardized HCC applications that underwent formal regional review board evaluation between January 1, 2005 and March 2, 2011; 2184 initial HCC exception point applications were submitted. Of these, 41.9% fulfilled MC, 26.6% fulfilled University of California-San Francisco (UCSF) criteria and 17.6% exceeded UCSF criteria. The majority of applications were accepted: 89.8% within UCSF and 71.2% beyond UCSF. There was a significantly (p < 0.001) higher risk of death on the waitlist or within 90 days of waitlist removal for candidates within UCSF (12.4%) or beyond UCSF (13.0%) criteria, compared to candidates with HCC within MC (6.0%). However, posttransplant outcomes were similar. While these results suggest increasing access to candidates with HCC beyond MC, comprehensive documentation of tumor characteristics and of successful downstaging is needed to ensure priority is restricted to those with the highest likelihood of favorable posttransplant outcome.




donor risk index


hepatocellular carcinoma


hepatitis C


inter-quartile range


loco-regional therapy


Milan criteria


Model for End-Stage Liver Disease


Organ Procurement and Transplant Network


radiofrequency ablation


regional review board


sub-hazard ratio


Standard Analysis and Transplant


transarterial chemoembolization


University of California-San Francisco


United Network for Organ Sharing


Since the implementation of the Model for End-Stage Liver Disease (MELD) allocation system, waitlisted candidates with hepatocellular carcinoma (HCC) have been eligible for MELD exception points. This change occurred as HCC patients experienced long wait-times during which tumor progression often precluded transplantation [1]. There has been a sixfold increase in the number of transplant recipients with HCC since 2002, despite research demonstrating worse posttransplantation survival for HCC recipients [2].

HCC MELD exceptions fall into two categories: (1) patients within Milan criteria (MC) who receive standardized points and (2) patients with tumors beyond MC who must receive approval from a regional review board (RRB). Some have argued that standardized HCC exception criteria be expanded nationally to include waitlist candidates within University of California-San Francisco (UCSF) criteria (with or without successful downstaging). Likewise, certain United Network for Organ Sharing (UNOS) regions (notably regions 4 and 5) have developed region-specific policies to expand exception point approval for tumors within UCSF criteria; however, such policies have not been accepted as formal UNOS policy. Based on Markov modeling, it is estimated that a 5-year posttransplantation survival rate of at least 61% would be needed to justify expansion of standardized HCC exceptions [3].

Although only a few select regions have adopted formal policies for allocating exceptions to patients beyond MC, similar approvals occur in other regions on a case-by-case basis. While single-center data have suggested acceptable posttransplant outcomes in HCC patients beyond MC, large-scale national data are limited as no prior studies have reviewed narrative data to ensure accurate characterization of tumor staging [4, 5]. Therefore, questions remain with regard to how and when to give waitlist priority to HCC patients beyond MC. Thus, to gain a better understanding of the patterns of utilization of HCC exceptions for patients beyond MC and their pretransplant and posttransplant outcomes, we reviewed the exception narratives for all patients applying for an HCC exception beyond MC.


Study population

All analyses were based on Organ Procurement and Transplant Network (OPTN)/UNOS data from January 1, 2005 to March 2, 2011. The start date of January 1, 2005 was chosen as the current HCC exception policy was initiated on that date. We included all adult waitlist candidates ≥18 years of age listed for an initial liver transplant on or after January 1, 2005.

We analyzed a data set of all waitlist candidates applying for a MELD exception, linked with a UNOS Standard Analysis and Transplant (STAR) file. The exception data set included: (a) type of application (initial, extension or appeal); (b) reason for application (13 different codes); (c) number of points requested; (d) full narrative submitted by the transplant center; and (e) application outcome. We reviewed the narratives of HCC waitlist candidates applying for nonstandardized HCC exceptions through formal RRB review that were coded as “exc_diag_id” = 3, or “HCC not meeting policy criteria.” HCC waitlist candidates awarded standardized HCC exceptions served as a control group.

Each application narrative was subjected to a primary, detailed review by one of two reviewers (TB and/or BN). Complex patient cases (approximately 10%) underwent a secondary review (DG and MH) and were adjudicated by consensus agreement. All patients with a tumor thrombus on imaging at any point were coded as “outside UCSF criteria.” Determination of whether a lesion was an HCC was based on the information provided in the narrative. Lesions noted to be indeterminate on imaging or were no longer apparent on subsequent scans in the absence of interval loco-regional therapy (LRT) were not counted toward the total tumor burden.

The narratives did not provide sufficient data (Results section) to determine whether LRT was successful or if patients were downstaged. As a result, patients were assigned to an HCC category based on the worst stage of their disease (i.e. a 6-cm tumor that underwent embolization and was reported as being “treated” would be categorized as within UCSF). A small number of patients underwent resection and were categorized based on their preresection tumor size—these patients were excluded in a secondary analysis. Patients were excluded if their application contained no narrative or if they were miscoded (e.g. no HCC).


The primary outcome was the result of the exception application. Applications were classified as approved or denied (which included all possible outcomes besides approved, as any nonapproved exception in practice yielded the same outcome). Secondary outcomes included: (1) pretransplantation waitlist drop-out for death or clinical deterioration, defined as dying on the waitlist based on UNOS coding, or being removed from the waitlist for being “too sick to transplant,” or “other” yet dying within 90 days of de-listing based on Social Security Death Master File death data provided in the UNOS data set; (2) deceased donor transplantation rate, to account only for organs allocated through the match; and (3) posttransplantation mortality, based on UNOS coding. We included “other” patients who died within 90 days of removal in the “died/too sick” category as many of these patients may have been miscoded and in fact removed for advanced HCC [6].

Statistical analysis

We used Fisher's exact test and chi-squared tests for dichotomous variables, Student's t-tests and Wilcoxon rank-sum tests for continuous variables (according to their distributions) and Kruskall–Wallis tests (when comparing more than 2 groups) to compare waitlist candidates across different HCC categories. To compare organ graft quality for transplant recipients, we calculated the donor risk index (DRI) using the formula established by Feng et al [7].

In evaluating pretransplant waitlist mortality, we considered the competing risk of transplantation, as it influences the probability that a waitlist candidate will be removed from the waitlist for death or clinical deterioration [8]. We fit competing risk Cox regression models with waitlist removal for death or clinical deterioration as the outcome, and transplantation as the competing risk [8-10]. All other outcomes were treated as censors. Waitlist candidates on the waitlist at the end of follow-up were censored. The primary covariate of interest was HCC category (within MC, within UCSF and beyond UCSF). Potential covariates included gender, race/ethnicity (as defined by UNOS), age at listing, laboratory MELD score at listing and primary diagnosis. The final models used a stepwise variable-selection process to retain variables with p-values ≤0.1. We used robust standard errors to account for correlation due to patient clustering by UNOS region [11]. Covariates were reported as sub-hazard ratios (SHRs), given the use of competing risk models [8, 10].

We fit Cox regression models to assess posttransplant mortality of transplant recipients within MC, within UCSF or beyond UCSF criteria, based on tumor data in the exception narrative. We adjusted for gender, race/ethnicity, age at transplantation, laboratory MELD score at transplantation, UNOS region, time from applying for exception to transplantation (to account for time of HCC observation, potentially associated with predicting tumor biology), primary diagnosis and use of LRT at the time of initial exception (yes/no). We evaluated center size as a covariate, using encrypted center ID codes in the UNOS data set, based on the number of waitlist candidates at a center after January 1, 2005. Secondary analyses explored the potential association between center-reported complications of portal hypertension (ascites or hepatic encephalopathy) and functional status at transplantation (0–100%) that are available in OPTN/UNOS data. These were considered exploratory analyses given that these data have not been validated.

Institutional Review Board approval was obtained from the University of Pennsylvania. All statistical analyses were performed using Stata 12.0 (College Station, TX).


There were 2184 initial MELD exception applications submitted for RRB approval under the code of “HCC not meeting policy criteria” (Table 1) during the study period. Fifty-eight applications were excluded from the analysis, 47 because no narrative was provided and 11 because they had a non-HCC malignancy. Accordingly, a total of 2126 initial applications were analyzed. Among the 2184 MELD exception applications undergoing RRB review, 915 (41.9%) were in fact within MC, 580 (26.6%) within UCSF and 385 (17.6%) beyond UCSF criteria, with the remaining either being: (a) below MC; (b) unknown staging due to missing data; (c) no narrative; or (d) non-HCC malignancy. There was a significant increase in the number of HCC beyond MC exception applications over time (Figure 1A; p < 0.001); however, this trend was not significant (Figure 1B; p = 0.07) when regions 4 and 5, that each have policies on beyond MC exceptions, were excluded.

Table 1. Categorization of exception point applications codes as “HCC exception beyond Milan criteria”
Category*Total applicationsApplications approved, N (%)
  • HCC, hepatocellular carcinoma; UCSF, University of California-San Francisco.
  • *p < 0.001.
  • **Excludes patients with no narrative (N = 47), and patients with non-HCC malignancies (N = 11).
Within Milan criteria915831 (90.8)
Within UCSF criteria580520 (89.8)
Beyond UCSF criteria385274 (71.2)
Unknown staging166118 (71.1)
Always below Milan criteria8037 (46.3)
Overall**21261781 (83.8)
Figure 1.

(A) Number of HCC exception applications for tumors beyond Milan criteria in all regions. (B) Number of HCC exception applications for tumors beyond Milan criteria, excluding regions 4 and 5. HCC, hepatocellular carcinoma; UCSF, University of California-San Francisco.

The patient demographics in the three major groups (within MC, within or beyond UCSF) were similar, except there were more within MC patients with documented hepatitis C (HCV), and more within or beyond UCSF patients with no documented diagnosis and/or an “other” diagnosis (Table 2).

Table 2. Demographic and clinical characteristics of waitlist applicants applying for HCC exceptions through an RRB
 Within MC, N = 915Within UCSF, N = 580Beyond UCSF, N = 385p-Value
  1. AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; IQR, inter-quartile range; MELD, Model for End-Stage Liver Disease; NASH, nonalcoholic steatohepatitis; RRB, regional review board; UCSF, University of California-San Francisco; UNOS, United Network for Organ Sharing.
  2. Other includes waitlist candidates with metabolic disease (i.e. hemochromatosis), vascular (i.e. Budd-Chiari), as well as waitlist candidates whose only diagnosis listed in UNOS coding and/or free-text data were “HCC.”
Male gender717 (78.4)482 (83.1)319 (82.9)0.04
Age at listing, median (IQR)58 (54–62)57 (53–63)57 (53–62)0.70
Race/ethnicity   0.20
White531 (58.0)333 (57.4)222 (57.7) 
Black88 (9.6)54 (9.3)34 (8.8) 
Hispanic150 (16.4)111 (19.1)66 (17.1) 
Asian134 (14.6)66 (11.4)59 (15.3) 
Other12 (1.3)16 (2.8)4 (1.0) 
Primary diagnosis   <0.001
Hepatitis C559 (61.1)302 (52.1)173 (44.9) 
Alcohol68 (7.4)51 (8.8)35 (9.1) 
NASH/cryptogenic45 (4.9)43 (7.4)18 (4.7) 
Hepatitis B102 (11.2)46 (7.9)46 (12.0) 
Cholestatic6 (0.7)6 (1.0)8 (2.1) 
Autoimmune12 (1.3)10 (1.7)10 (2.6) 
Other123 (13.4)122 (21.0)95 (24.7) 
Laboratory MELD score at exception application10 (8–13)10 (8–13)10 (8–13)0.21
AFP at initial application12 (5–42)13 (6–57)12 (5–55)0.11

Among HCC applicants undergoing RRB review who were in fact within MC, approximately 45% had received prior LRT without evidence of residual tumor, 20% had residual HCC post-LRT and 5–7% had post-LRT ablation cavities larger than MC. Other reasons for applying through an RRB included missed deadlines, temporary inactive status during evaluation, clerical error, conflicting imaging reports and incomplete tumor resection.

The overall application approval rate was 83.8% (1781/2126; Table 1), excluding the 58 cases for which no narrative was provided or the patient had a non-HCC malignancy. There were significant differences (p < 0.001) in the proportion of applications approved within UCSF (89.8%) versus beyond UCSF (71.2%). Among all beyond MC exceptions, the approval rate increased over time (p < 0.001), ranging from approximately 75% from 2005 to 2007 to over 83% from 2008 to 2010, driven largely by increase in approvals of HCC candidates within UCSF criteria in regions 4 and 5.

Narrative versus numeric data

The numeric data provided in the UNOS HCC STAR file incorrectly classified 405 (18.6%) patients, with 205 (9.4%) being under-staged as within MC, when based on the narrative they were beyond MC and 200 (9.2%) over-staged (numeric data staged patient as beyond MC, when in fact the patient was within MC). The numeric data under- or over-staged tumor burden because tumors were incorrectly sized as a result of LRT or because lesions were miscalled as an HCC (i.e. indeterminate or not seen on a second scan despite no treatment, but still reported as having a tumor size and being HCC).

Regional variability in applications beyond MC

Nearly one-half of within UCSF (49.9%, N = 289) and beyond UCSF (46.2%, N = 178) applicants were from regions 4 and 5. The application approval rate for within UCSF applicants was similar (Table S1), with >85% of applicants being approved (excluding region 1, with only nine applicants). There was significant variability (p = 0.003) in the beyond UCSF approval rate, ranging from 42.3% in region 7 to 90.9% in region 8 (Table S1).

Center variability in applications beyond MC

Of the 965 waitlist candidates within or beyond UCSF criteria, 94.4% (911/965) had encrypted center data. While these waitlist candidates were waitlisted at 1 of 97 centers, 55.5% (506/911) were listed at 1 of 19 transplant centers—only 33.6% (19 109/56 870) of new listings after January 1, 2005 were from one of these centers. There was no correlation (R = 0.13) between center size and the proportion of waitlist candidates applying for beyond MC exceptions.

Tumor characteristics

Among waitlist candidates within or beyond UCSF criteria, 98.2% (948/965) had numeric data on the total number of tumors (i.e. some only included descriptive data such as “several” or “multiple”), 96.2% (928/965) had data on the diameter of the largest tumor and 95.9% (925/965) had sufficient data to calculate total tumor diameter.

The median number of tumors in both groups was two (inter-quartile range [IQR] of one to three for within UCSF, and two to four for beyond UCSF), with larger tumor diameters in the beyond UCSF cohort. Within each category, the diameter of the largest tumor was greater among waitlist candidates whose initial exception application was denied, while the total tumor diameter was similar (Tables 3a and 3b).

Table 3a. Tumor criteria for exception point applicants within UCSF criteria, N = 580
CategoryLargest tumor diameter in cm, median (IQR)Total tumor diameter in cm, median (IQR)Transplanted, N (%)Died/too sick, N (%)
  1. IQR, inter-quartile range; UCSF, University of California-San Francisco.
Application accepted, N = 5214 (3.3–5.3)5.5 (5.1–6.1)338 (64.9)58 (11.1)
Application denied, N = 595.0 (3.8–5.6)5.5 (5.1–6.0)31 (52.5)14 (23.7)
Table 3b. Tumor criteria for exception point applicants beyond UCSF criteria, N = 385
CategoryLargest tumor diameter in cm, median (IQR)Total tumor diameter in cm, median (IQR)Transplanted, N (%)Died/too sick, N (%)
  1. IQR, inter-quartile range; UCSF, University of California-San Francisco.
Application accepted, N = 2744.9 (3.8–6.5)7.3 (6.4–8.5)159 (58.0)31 (11.3)
Application denied, N = 1115.7 (4.4–7.2)7.8 (6.6–8.8)34 (30.6)19 (17.1)

LRT in waitlist candidates beyond MC

Of waitlist candidates, 84.9% (819/965) applying for an exception for “HCC beyond MC” received at least one form of LRT. In those within UCSF criteria, 79.3% (460/580) received LRT, compared with 93.2% (359/385) of beyond UCSF candidates. Of within UCSF candidates receiving LRT, the most common forms were transarterial chemoembolization (TACE) (64.1%; N = 295), radiofrequency ablation (RFA) (13.0%; N = 60) and TACE + RFA (9.1%; N = 42). Among beyond UCSF candidates, the most common were: TACE (60.7%; N = 218), combination therapy that was not TACE + RFA (i.e. TACE + radioembolization or TACE + external-beam radiation) (12.8%; N = 46); TACE + RFA (9.5%; N = 34); and RFA alone (7.5%; N = 27). Within each category, the proportion receiving at least one form of LRT was similar based on: (1) whether the exception application was approved; (2) whether the patient was transplanted; and (3) whether the patient died posttransplantation.

Waitlist outcomes

Waitlist candidates within (12.4%, 72/580) and beyond UCSF criteria (13.0%, 50/385) were significantly (p < 0.001) more likely to die on the waitlist or within 90 days of waitlist removal, compared to within MC candidates applying through an RRB (6.0%, 55/915) or through the standard HCC exception process (8.0%, 561/7051). Only 3 (3.8%) waitlist candidates with tumors always below MC died pretransplant, while 45 (56.3%) were transplanted, with similar results in those with unknown staging (data not shown).

In unadjusted competing risk models, patients within UCSF (SHR: 2.14, 95% CI: 1.51–3.03) and beyond UCSF criteria (SHR: 2.46, 95% CI: 1.66–3.66) had a significantly increased sub-hazard of waitlist mortality compared with waitlist candidates within MC (Table 4). In competing risk curves, the waitlist mortality for candidates within and beyond UCSF criteria was significantly worse than either cohort of candidates within MC (Figure 2), and persisted, albeit attenuated, when restricted only to exception point applicants with approved applications (data not shown).

Table 4. Competing risk models evaluating waitlist mortality of patients within versus beyond Milan criteria
VariableUnivariable SHRMultivariable SHR1p-Value
  • AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; MELD, Model for End-Stage Liver Disease; SHR, sub-hazard ratio; UCSF, University of California-San Francisco.
  • 1Final multivariable model did not include primary diagnosis, race/ethnicity or male gender as these were nonsignificant (p > 0.3) in univariable models and were not confounders (did not change SHR by 10%). Final multivariable model accounted for patient clustering within regions. Primary diagnosis (hepatitis C, hepatitis B, alcohol, nonalcoholic steatophepatitis/cryptogenic, autoimmune, cholestatic and “other”).
  • 2Sub-hazard for every 10-year increase in age at listing.
  • 3Sub-hazard for every increase in 1 MELD point.
  • 4Sub-hazard for every increase in 50 µg/mL.
Age at listing21.29 (1.06–1.57)1.39 (1.12–1.72)0.003
Listing laboratory MELD score31.12 (1.08–1.15)1.13 (1.09–1.17)<0.001
HCC criteria
Within Milan11 
Within UCSF2.14 (1.51–3.03)2.20 (1.51–3.20)<0.001
Beyond UCSF2.46 (1.66–3.66)2.44 (1.67–3.56)<0.001
AFP41.01 (1.00–1.01)1.01 (1.00–1.01)<0.001
Figure 2.

Competing risk model of waitlist mortality. RRB, regional review board; UCSF, University of California-San Francisco.

In multivariable competing risk models (Table 4), having a tumor within UCSF (SHR: 2.20, 95% CI: 1.51–3.20) or beyond UCSF (SHR: 2.44, 95% CI: 1.67–3.56) was significantly associated (p < 0.001) with an increased hazard of waitlist mortality, as were listing laboratory MELD score, increased age at listing and increased alpha-fetoprotein (AFP) at initial exception application (Table 4).

Among HCC waitlist candidates within or beyond UCSF criteria, 562 (58.2%) were transplanted, with a significantly higher proportion among those within UCSF criteria (63.6%, 369/580 vs. 50.1%, 193/385; p < 0.001). The median DRI was not significantly different (p = 0.82) for transplant recipients within MC (1.42, IQR: 1.16–1.74), within UCSF (1.43, IQR: 1.16–1.76) or beyond UCSF criteria (1.44, IQR: 1.17–1.73).

Posttransplant outcomes

Unadjusted posttransplant survival was significantly different between the four HCC cohorts evaluated (log-rank test p = 0.002), and persisted when restricted to the three cohorts applying through an RRB (log-rank test p = 0.01; Figure 3). However, posttransplant patient survival for patients within versus beyond UCSF criteria was not significantly different (log-rank test, p = 0.44), and there were no differences when stratified by HCV status (log-rank test, p = 0.08) or when beyond MC patients with prior tumor resection were excluded (data not shown). Transplant recipients below MC and with unknown staging had a similar risk of posttransplant mortality (data not shown).

Figure 3.

Posttransplant patient survival of HCC transplant recipients approved through RRB review. HCC, hepatocellular carcinoma; RRB, regional review board; UCSF, University of California-San Francisco.

Unadjusted 3- and 5-year patient survival was higher in transplant recipients initially within or beyond UCSF criteria (Table 5). In regions with a sufficient sample size to estimate unadjusted 3-year posttransplant survival of HCC transplant recipients beyond MC (UNOS regions 4, 5, 7, 8, 9 and 10), the 3-year posttransplant patient survival was >70% in all regions, and >85% in three regions.

Table 5. 1-, 3- and 5-year posttransplant survival among HCC transplant recipients after January 1, 2005
HCC category1-Year survival, % (95% CI)3-Year survival, % (95% CI)5-Year survival, % (95% CI)
  1. HCC, hepatocellular carcinoma; RRB, regional review board; UCSF, University of California-San Francisco.
Standard Milan exceptions90.6 (89.8–91.3)77.4 (76.1–78.6)67.1 (65.0–69.1)
Within Milan RRB exceptions87.8 (84.5–90.8)72.1 (65.2–77.8)68.9 (59.5–76.6)
Within UCSF criteria91.5 (87.6–94.3)82.4 (76.2–87.1)75.5 (64.5–83.5)
Beyond UCSF criteria91.8 (86.2–95.2)85.1 (77.2–90.5)70.9 (36.9–88.8)

In multivariable models, HCC recipients within UCSF (HR: 0.64, 95% CI: 0.44–0.93, p = 0.02) and beyond UCSF criteria (HR: 0.56, 95% CI: 0.34–0.92, p = 0.02) had a significantly decreased risk of posttransplant mortality when compared with MC recipients (Table 6). There were no significant differences when stratified by low-, medium- and high-MELD regions (log-rank test, p = 0.82), and no association between UNOS region and posttransplant mortality (p = 0.24).

Table 6. Cox regression models evaluating posttransplant mortality of HCC transplant recipients receiving exception points through an RRB
VariableUnivariable HRMultivariable HR1p-Value
  • AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; HR, hazard ratio; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; RRB, regional review board; UCSF, University of California-San Francisco.
  • 1Multivariable model did not include male gender, primary diagnosis, race/ethnicity, time from exception point application to transplant, AFP or use of pretransplant loco-regional therapy as these variables were not significant (p > 0.2) in univariable models. Final multivariable model accounted for patient clustering within regions.
  • 2Hazard ratio represents increased hazard for every 10-year increase in age at OLT.
Age at time of OLT21.45 (1.14–1.86)1.49 (1.16–1.92)0.002
Final laboratory MELD score1.03 (1.00–1.05)1.03 (1.00–1.05)0.02
HCC criteria
Within Milan11 
Within UCSF0.64 (0.44–0.92)0.64 (0.44–0.93)0.02
Beyond UCSF0.56 (0.34–0.91)0.56 (0.34–0.92)0.02

There was no association or interaction between center volume and posttransplant survival in transplant recipients within or beyond UCSF criteria (p > 0.1 for center size variable and interaction term of center size and HCC category). Among transplant recipients within or beyond UCSF criteria, largest tumor size and total tumor diameter were not associated with an increased hazard of posttransplant mortality (p > 0.05 for both variables tested either as continuous or categorical variables). Although the pretransplant waiting time from initial exception approval to transplantation was significantly shorter for within or beyond UCSF waitlist candidates compared to within MC candidates applying through an IRB, waiting time was not significantly associated with posttransplant survival (data not shown).

In secondary analyses, there was no association between complications of portal hypertension (ascites and/or encephalopathy) and posttransplant survival. In multivariable models restricted to the 92% of transplant recipients with available functional status data, worse functional status at transplantation was associated with significantly increased posttransplant mortality (functional status of 80–100% as reference): functional status 50–70%: HR: 1.51, 95% CI: 1.09–2.08, p = 0.012; functional status 10–40%: HR: 2.80, 95% CI: 1.50–5.22, p = 0.001. However, the hazard ratios for HCC within and beyond UCSF were unchanged in multivariable models that included functional status.

Post–March 2, 2010 applications

Between March 2, 2010 and March 2, 2011, there were 523 exception point applications submitted with the code of “HCC beyond Milan criteria,” with 124 (23.7%) categorized as within UCSF and 77 (14.7%) beyond UCSF criteria. Applications within UCSF criteria were significantly more likely to be approved (96.0%, 119/124) than those beyond UCSF (76.6%, 59/77; p < 0.001), which persisted with the exclusion of UNOS regions 4 and 5 (93.7%, 59/63 vs. 75.8%, 25/33; p = 0.01). The report of the national consensus conference on HCC set forth recommendations for what should be required of an exception application in HCC tumors exceeding standard criteria. Although not formal policy, these recommendations defined criteria for downstaging (Table S2; [[12]]), yet only 8.4% (15/178) of approved applications contained the recommended documentation, with similar results when excluding regions 4 and 5. However, 43% of approved exceptions since March 2, 2010 contained documentation of original and residual tumor size, 41% only documented one of these two and 16% provided neither. There was significant regional variability in the amount of documentation in these approved exceptions (p = 0.024).


This is the first study to provide a detailed examination using patient-level narrative data from all waitlisted candidates applying for exception points for HCC beyond MC in the United States. This work demonstrates that the submission of exception point applications outside of the standard HCC exception process not only is commonplace, but more importantly includes a significant proportion of candidates with HCC beyond MC. Although the data do not allow for a determination of successful downstaging, there has been a progressive increase over time in the application for exception points for tumors beyond MC, with the overwhelming majority being approved. Compared with waitlist candidates with HCC within MC, candidates beyond MC have increased pretransplant waitlist dropout but similar posttransplant survival, suggesting that transplant centers and RRBs are appropriately selecting waitlist candidates for beyond MC exceptions. However, these results must be taken in context of the increasing use of these exceptions, and the potential impact on waitlisted candidates without HCC. Although proposals for an expansion of the MC have thus far been controversial for a number of reasons, the results of this study highlight the potential for expanding HCC exceptions without sacrificing posttransplant outcomes.

Our study was the first to use data within HCC exception narratives to precisely stage tumors for candidates waitlisted nationwide. Prior studies evaluating outcomes in HCC patients beyond MC were either based on single-center data [13] or used numeric data provided in a UNOS STAR file [3]. As we have shown, the numeric data provided by UNOS is insufficient on its own to appropriately categorize HCC tumor stage due to both under- and over-staging of HCC. Thus, we believe our data is the most robust US data on outcomes of HCC patients beyond MC to date.

Since the implementation of MELD-based allocation, there have been several policy adjustments on the national level to maximize access to transplantation for waitlist candidates with HCC within MC, without unduly decreasing access among non-HCC patients. Despite the lack of a formal national policy on HCC exceptions beyond MC, the national approval rate for exceptions for HCC within UCSF criteria was >85% in all regions, with high approval rates for HCC beyond UCSF as well. Although the posttransplant outcomes in these transplant recipients are satisfactory, the lack of standardization for approval is concerning, especially given the high waitlist mortality for non-HCC waitlist candidates. Furthermore, despite the publication of national guidelines for granting of exception points for downstaged tumors in 2010, nearly all narratives were missing the necessary data to accurately determine successful downstaging.

The differences in pretransplant waitlist dropout based on tumor stage are not surprising, given the increased risk of tumor progression in patients with a larger tumor burden. It may also be that patients with more advanced or larger tumors require more LRT, which can be associated with increased morbidity and mortality related to hepatic decompensation. Although counter-intuitive, the comparable posttransplant outcomes in transplant recipients within and beyond MC are consistent with prior studies. These results are likely due to selection bias, as transplant centers only applied for HCC exceptions beyond MC for those with more favorable HCCs based on a number of factors such as the rate of tumor progression or response to LRT, which are not easily discernible from an analysis of the UNOS database. In addition, the pretransplant waiting times and increased pretransplant dropout may have also selected for transplantation of those with more favorable tumor biology. Third, the “beyond MC” patients had a lower prevalence of HCV. This could be a selection bias introduced by transplant centers aiming at minimizing the risk of adverse posttransplant outcome in patients with a larger tumor burden. We did adjust for HCV status in our multivariate models, which approached but did not reach statistical significance. Despite this, there is the potential for the reduced overall mortality to be driven by a lower risk of recurrent HCV-related mortality in the beyond MC groups, despite adjustments in our models.

We attempted to identify patient or tumor characteristics associated with increased posttransplant mortality, including AFP, largest tumor size and total tumor volume, but these were not found to be significant. This may in part be explained by the relatively small number of posttransplant deaths in our study. In a recent publication that demonstrated similar outcomes for candidates with HCC beyond MC, pretransplant AFP was a significant predictor of posttransplant HCC recurrence [3]. While our study demonstrates the challenges associated with implementing new downstaging guidelines, including the serum AFP in the exception point submission, might be a simple way to better understand transplant suitability.

This study has several limitations. First, an important general limitation is the lack of reporting of posttransplantation HCC recurrence. This is a key factor in determining graft survival in patients beyond MC, and is therefore essential to establish whether an expansion of the MC is justified. Second, given the lack of standardization in documentation, the listing center may choose to include as much or as little information regarding radiology reports as deemed necessary. In many situations, determining lesion size and number is complex: for example, residual tumor sizes post-LRT are often unknown, and smaller lesions may be indeterminate. Third, the timeline of candidates' HCC prior to application was also poorly documented. This impacted our ability to differentiate waitlist and posttransplantation outcomes in patients with slower- compared to faster-growing tumors. Finally, unlike previous downstaging studies, the available data did not allow us to verify if patients with tumors beyond MC were successfully downstaged. Last, we did not have pathologic data (tumor biopsy or explant reports) to evaluate the radiology to pathology correlation in tumor extent and determine if tumor grade or the presence of microvascular invasion explained the differences in posttransplant outcomes. Future work utilizing all explant pathology reports sent to OPTN/UNOS will explore this further.

There is a need for more evidence-based guidance on how and when to prioritize transplant candidates with HCC beyond MC. In developing a standardized process with the potential to impact organ allocation nationally, several aspects must be taken into consideration. In 2010, an international consensus conference recommended that “patients with a worse prognosis may be considered for liver transplantation outside the MC if the dynamics on the waiting list allow it without undue prejudice to other recipients with a better prognosis” (14, [14]. Thus, an expansion of the MC that takes into consideration regional differences in the proportion of candidates with HCC and overall waitlist mortality could avoid jeopardizing access to transplantation for non-HCC candidates. Second, any expansion of standardized HCC exception policies should include a mandatory surveillance period to observe candidates for disease stability to identify tumors with more favorable tumor biology [12]. Finally, any proposal for the expansion of the MC should include more stringent documentation requirements, both for standardized and for nonstandardized applications, thereby ensuring the authenticity of the requests, along with a more robust reporting system to evaluate posttransplant tumor recurrence. We believe the ultimate authority to regulate and monitor the data that are submitted rests with OPTN; however, implementing policies to require increased documentation should not overburden the system. One such mechanism would be an automated system that both ensures more detailed data submission, and also prevents submission of exceptions without full data documentation. At the very minimum, the submission process to the RRB should require: (1) tumor size for each cross-sectional scan (computed tomography or magnetic resonance imaging) to better understand tumor growth rate; (2) radiographic features suggestive of HCC; (3) response to LRT based on modified Response Evaluation Criteria in Solid Tumors criteria, with sizes of ablation cavities and residual enhancement areas; and (4) dates of all loco-regional treatments to allow for a calculation of the observation period. There should still be a mechanism to submit a brief supplemental narrative, and OPTN/UNOS should develop more comprehensive guidelines for the evaluation of nonstandardized HCC applications.

In conclusion, this study demonstrates that it is not only appropriate to consider transplantation for patients with HCC beyond MC, but may also be suitable to prioritize certain candidates because of their higher risk of waitlist mortality. The full impact of awarding these exceptions is unknown, and there remain unanswered questions as to how an expansion of HCC exceptions can be accomplished without placing non-HCC patients at risk for increased dropout and mortality. Our work emphasizes the need for further research to better understand the potential consequences of this proposal to both HCC and non-HCC candidates in the United States. In the interim, we recommend that transplant centers nationwide be held to higher documentation standards to optimize the patient selection process in a consistent manner.


Dr. Goldberg has received research support from Bayer Healthcare.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.