SEARCH

SEARCH BY CITATION

Keywords:

  • Alpha-fetoprotein;
  • competing risks;
  • delisting owing to tumor progression;
  • hepatocellular carcinoma;
  • liver transplant candidates;
  • orthotopic liver transplantation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Orthotopic liver transplantation (OLT) is potentially curative for patients with early stage hepatocellular carcinoma (HCC). However, tumor progression before OLT remains a problem. Ninety-three patients were listed for transplantation with HCC or diagnosed with HCC following listing between March, 1997 and September, 2001. Modified TNM Stage was I/II in 82 patients and III in 11 patients. Seventy-one patients (76%) were transplanted with a median waiting time of 3.4 months, and 22 (24%) patients were delisted owing to tumor progression (14), noncompliance (5), and death from liver failure (3). Using a cox model competing risks approach, higher baseline alpha-fetoprotein (AFP) ≥ 100 ng/mL was the only factor independently associated with a higher hazard rate of delisting owing to tumor progression (p = 0.00003), whereas four separate factors were independently associated with a lower hazard rate of transplantation: more recent listing year (1999–2001, p = 0.010), blood type O (p = 0.013), Stage I HCC (p = 0.029), and serum bilirubin < 4 mg/dL (p = 0.032). By logistic regression, AFP ≥ 100 ng/mL was the only factor that significantly influenced the probability of delisting owing to tumor progression (p = 0.001). In conclusion, the initial AFP level may be useful along with tumor stage in defining an urgency score for liver transplant candidates with HCC.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Orthotopic liver transplantation (OLT) is now accepted as potentially curative therapy for select cirrhotic patients with hepatocellular carcinoma (HCC) (1–6). Recent studies have shown that survival after liver transplantation in patients with early stage HCC can exceed 70% at 3 years (1–6) and is comparable to that of patients without HCC (3). However, once the HCC patient is listed and waiting for a transplant, there is a distinct possibility that the patient's disease will progress such that an OLT is no longer a reasonable treatment option. Although the United Network of Organ Sharing (UNOS) (7) is currently using the Milan criteria (2), equivalent to the modified Tumor-Node-Metastasis (TNM) Stage I/II (8), for listing and de-listing of HCC patients, this strategy is based on sparsely available data owing, in part, to the relatively rare frequency of HCC occurrence in the general population (9). In addition, under a recent UNOS policy change, a higher priority status is now assigned to HCC patients according to their modified TNM stage (7,8). The goal was to allow for the likelihood that patients listed for transplant with HCC may generally be less sick at the time of listing but still carry the additional risk of the tumor progressing while on the waiting list. Given the limited available data on the outcomes of HCC patients once listed for OLT (10–12), one important question is whether clinical stage by itself or possibly some other combination of patient characteristics will provide the most reliable prediction of risk of delisting owing to tumor progression. Clearly, the additional priority score assigned to a patient listed with HCC should be based on that patient's probability of being delisted owing to tumor progression.

An analysis of prognostic factors for the probability of being delisted owing to tumor progression is clearly possible. However, one must keep in mind that there are two distinct competing risks here: the patient might be delisted owing to tumor progression or the patient might receive an OLT. Unfavorable tumor-related characteristics may be directly associated with a higher hazard rate of delisting owing to tumor progression. In fact, the recent reports by Yao et  al. (11,12) found such associations for multiple tumor nodules and a larger tumor size among solitary lesions. However, there may be other patient characteristics that are not associated with the hazard rate of delisting but instead are associated with a lower hazard rate of OLT. Patients with such characteristics must wait longer to receive an OLT and are therefore at risk of developing tumor progression for a longer period of time. Logic would suggest that a greater percentage of these patients will be actually delisted owing to tumor progression. In fact, there are reports in the literature suggesting this type of association for a more recent date of listing for transplantation and blood type O (10,13).

A major goal of this study was therefore to distinguish using a competing risks approach (14–17) between factors that are associated with (1) a higher hazard rate of delisting owing to tumor progression, and (2) a lower hazard rate of transplantation. Through such an analysis the reasons why these factors are prognostically important for the probability of being delisted become apparent. As we are aware of only two other groups that have explored risk factors for delisting in this cohort of patients (10,11,12), our purpose was threefold: (1) to offer possible improvements in the criteria used for assigning an HCC-specific urgency score, (2) to compare our results with those of the other studies, and (3) to show by example the importance of using competing risks methodology with this type of data.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Selection of subjects

Between January, 1997 and June, 2001, 2744 patients were referred for evaluation at the University of Miami Liver Transplant Program. Of these, 128 patients carried an original diagnosis of HCC based on: (1) a liver nodule recognized by two different imaging studies [abdominal ultrasonography, contrast-enhanced computed tomography (CT), or contrast-enhanced magnetic resonance imaging], (2) a liver nodule recognized by one imaging study along with sequential elevation of serum alpha-fetoprotein (AFP) level, or (3) histologically proven HCC. Thirty-three of these patients were deemed unsuitable for OLT by the transplant physicians because of a resectable tumor in one, advanced tumors in 13, and other or unknown reasons in 19. After receiving approval by the Institutional Review Board, a retrospective review was conducted, and 13 additional patients were excluded because of an unconfirmed tumor. Thus, 82 listed patients with HCC were identified. In addition, during this time 1177 of the 2744 patients were listed for transplant without HCC. By the end of June 2001, 11 of these 1177 patients were diagnosed with HCC. Therefore, a total of 93 HCC patients were identified as candidates for OLT. A baseline CT with dynamic IV contrast was performed in all of the OLT candidates; a baseline bone scan and chest CT ruled out metastatic disease in each case. Follow up of the 82 patients with HCC who were listed began on the date of listing; follow up of the 11 patients who developed HCC after being listed began on the date of diagnosis with HCC. The date of last follow up for this study was the date of last observed OLT: February 3, 2003.

Data description

The clinical stages of the 93 patients at the time of listing (or diagnosis following listing) were as follows: Stage I (n = 9), Stage II (n = 73), and Stage III (n = 11). Thus, Stage III disease was an acceptable listing criterion at this center during the period of listing for these patients (March, 1997 to September, 2001). Four distinct clinical outcomes following listing (or diagnosis after listing) were observed (see Figure 1): OLT, delisting owing to tumor progression, delisting owing to death without progression from liver failure, and delisting owing to patient choice/noncompliance. Following transplantation, two distinct clinical outcomes were observed: HCC recurrence, and death without HCC recurrence.

image

Figure 1. Clinical outcomes for the 93 patients listed (or diagnosed following listing) with hepatocellular carcinoma (HCC).

Download figure to PowerPoint

The modified TNM classification (8) was used to determine the initial stage of HCC (at listing or diagnosis following listing). The severity of liver cirrhosis was categorized based on the Child-Pugh classification (18). Before 1999, no formal protocols for adjunctive treatment of HCC were instituted. From 1999 onwards, patients with 1–3 tumor nodules (none >5 cm) and a Child-Pugh classification of A or B were eligible to receive microwave coagulation (ablation) therapy (19). If a patient met the eligibility criteria but was unable to receive treatment owing to location of the tumor or a frail clinical status, then a transcatheter arterial chemo-embolization (TACE) was considered if the patient had a hypervascular mass. Delisting owing to tumor progression was determined when the patient had progression to Stage IV disease.

Variables considered for prognostic value included the patient's age and sex, date of listing (or diagnosis of HCC after listing), blood type, modified TNM stage, maximum tumor size, number of nodules, AFP, HBV infection, HCV infection, Child-Pugh classification, serum bilirubin, serum albumin, prothrombin time, prereferral therapy, and postreferral therapy. Pre-referral therapy included hepatectomy, percutaneous ethanol injection, TACE, and local thermal ablation therapy. Post-referral therapy included TACE (n = 8) and ablation therapy (n = 22).

Histopathological evaluation

The explanted liver was examined by an experienced surgical pathologist. The number of nodules, maximal tumor diameter, histology, histologic grade modified from Edmondson and Steiner's classification (20), and presence of microvascular invasion were recorded. Differences between the initial clinical staging and pathologic staging of the explanted liver were observed using the modified TNM classification (8).

Statistical analysis

A multivariable analysis of prognostic factors for two distinct hazard (instantaneous) rates, the hazard rate of delisting owing to tumor progression and the hazard rate of transplantation, was performed using separate stepwise Cox regression analyses. In each analysis, failures owing to causes other than the cause of interest were treated as censored observations, and the score Chi-square test criterion was used. Time was defined as the number of months since listing (or diagnosis of HCC following listing) for OLT. In order to avoid the possibility of obtaining spurious results with a relatively small sample size, only variables with univariable p-values ≤ 0.05 were considered for entry into the Cox models. Subgroup differences for a particular hazard rate were also tested by the log-rank test, and graphical display of these subgroup differences were shown using Kaplan-Meier curves. As it was of primary interest to determine which factors influence the overall probability of being delisted owing to tumor progression, a stepwise logistic regression analysis was performed on this outcome variable. The clinical outcomes following listing were known for each patient (Figure 1); thus, complete data were available.

A separate Cox regression analysis of prognostic factors associated with the hazard rate of death without progression owing to liver failure was not performed, because only three such deaths were observed. One would expect that prognostic factors associated with the hazard rates of delisting owing to tumor progression and death without progression from liver failure would either not be the same or would differ in magnitude: the former would be more affected by tumor-related characteristics and the latter would be more affected by measures of the severity of cirrhosis. Nonetheless, as a check on the reliability of our analysis, the Cox stepwise regression for the hazard rate of delisting was re-run, combining both types as one failure.

It was also of interest to determine whether the Cox stepwise regression results would be affected by our inclusion/exclusion criteria, which clearly differed from the UNOS T1-T2 (Stage I/II) criteria. We therefore re-ran our analysis using the stricter UNOS criteria. Finally, UNOS implemented its MELD policy (7,21) on February 27, 2002. Although all of our patients were listed for OLT before that date, six of the 93 patients were still wait-listed as of that date. The most definitive way to test for the possibility that the magnitudes of the important prognostic factors' effects would differ before and after implementation of the MELD policy is to consider time by covariate interaction effects in the Cox model. However, as there were only six patients followed beyond this date, the stepwise Cox regressions were simply re-run, censoring patients at the date of MELD implementation.

Tests of association among the prognostic variables were performed using Pearson Chi-squared tests, t-tests of mean differences, and t-tests of linear correlation. Categorizations of continuous variables were considered in the prognostic factor analyses along with log transformations of highly skewed variables. Other outcome variables considered were HCC recurrence following OLT and survival.

Finally, we are introducing the use of the correct formula for estimating the probability of failing from a particular cause by time t in the presence of competing risks, known in the biostatistics literature as the cumulative incidence function (14,15,17,22). This includes estimation of the probability of being delisted owing to tumor progression by time t as well as the probability of receiving an OLT by time t. Although use of Kaplan-Meier (KM) curves to show subgroup differences in the hazard rate for a particular cause is appropriate, it has been common fallacy to use 1-KM for estimating the probability of failing from a particular cause by time t; in fact, the 1-KM approach will always provide an overestimate in the presence of competing risks. The cumulative incidence function accounts for the fact that patients must survive from all causes of failure up to time t in order to be at risk of failure from a particular cause at that time. Examples will be provided.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Clinical outcomes

Figure 1 shows that 22 (24%) of 93 OLT candidates were delisted: 14 owing to tumor progression, three owing to death without progression from liver failure, and five owing to patient choice/noncompliance. The stages at progression were IV-A1 in six cases, IV-A2 in seven cases, and IV-B in one case. Two cases were diagnosed with portal vein involvement on the day of attempted transplantation and were immediately delisted. Ten of the delisted patients with tumor progression had initial Stage I/II disease; the remaining four patients had initial Stage III disease. No patients with Stage I/II disease progressed to Stage III disease before delisting; thus, the 10 delisted patients with initial Stage I/II disease met the UNOS criteria for delisting. Two patients with initial Stage I disease were observed to have Stage II progression and were kept on the waiting list.

The estimated probabilities of being delisted owing to tumor progression and receiving an OLT by time t are shown in Figure 2. At 12 months, the cumulative incidence estimates were 11.8% (11/93) and 67.7% (63/93), respectfully. The cumulative incidence estimate of the overall probability of being delisted owing to tumor progression was 15.1%; identical to the observed percentage of 15.1% (14/93) because of complete data. The cumulative incidence estimate of the overall probability of receiving an OLT was 76.3% (71/93). Severe overestimation by the 1-KM estimator is seen in Figure 2, with the overall probabilities of being delisted owing to tumor progression and receiving a transplant being 49.2% and 100%, respectively. The fact that the two 1-KM estimates sum to greater than 100% clearly indicates the inaccuracy of using 1-KM for estimating such probabilities.

image

Figure 2. Cumulative incidence (nonparametric) estimates of the probabilities of being delisted owing to tumor progression and receiving an OLT by time t vs. the 1-Kaplan-Meier (KM) estimates.

Download figure to PowerPoint

There was a greater percentage of earlier times-to-OLT in comparison with the times-to-delisting owing to tumor progression. Specifically, during the first 3 months following listing (or diagnosis following listing) 41% (29/71) of the OLTs had already occurred in comparison with only 14% (2/14) of the delistings owing to tumor progression. Among the 71 patients who received an OLT, the median waiting time was 3.4 months (range 1 day to 29.4 months). Among the 14 patients who were delisted owing to tumor progression, the median time to delisting was 5.5 months (range 10 days to 13.7 months). The three deaths without tumor progression occurred at 12 days, 1.5 and 22.7 months; the five delistings owing to patient choice/noncompliance occurred at 1.1–10.4 months.

Prognostic factors for the hazard rate of delisting owing to tumor progression

Results of the stepwise Cox regression analysis appear in Table 1A. Higher log (AFP) (p = 0.00001), AFP ≥100 ng/mL (p = 0.00003), Child-Pugh Class A (p = 0.021), larger tumor size (p = 0.027), and Stage II/III (p = 0.028) were found to be significant univariable predictors of a higher hazard rate of delisting owing to tumor progression. Other factors including number of tumor nodules were not associated with this outcome variable (p > 0.1). As the dichotomy for AFP (<100 vs. ≥100) was highly significant by itself, for simplicity it was considered over log (AFP) in the Cox model. The stepwise Cox regression analysis found that AFP ≥100 ng/mL was the only independent predictor of the hazard rate of delisting owing to tumor progression (p = 0.00003). Maximum tumor size and clinical stage were associated with AFP. Specifically, the percentage of patients with AFP ≥100 was 40% (17/43) and 18% (9/49) among those with size >3 and ≤3 cm (p = 0.024). The percentage of Stage I, II, and III patients with AFP ≥100 was 0% (0/9), 30% (22/73), and 40% (4/10), respectively (p = 0.11). Although nonsignificant, patients with Child-Pugh class A were more likely to have AFP ≥100 (p = 0.21) and size >3 cm (p = 0.09). Finally, it should be noted that in all re-runs of this Cox stepwise regression where (i) the three patients who died without progression from liver failure were included as ‘failures’, (ii) the 11 Stage III patients were excluded while applying the UNOS delisting criteria, and (iii) the six patients who were followed beyond the date of MELD policy implementation were censored on that date, the results were identical to those shown in Table 1.

Table 1.  Results of the stepwise Cox regression analyses: (Panel A) hazard rate of delisting owing to tumor progression; and (Panel B) hazard rate of transplantation
Factors2Patients with the characteristic (%)Panel ADelistingPanel BTransplantation
Univariable p (score test)Selection (√) into Cox model p (β± SE)Univariable p (score test)Selection (√) into Cox model p (β± SE)
  1. 1Continuous variables

  2. 2Effects of HBV infection, HCV infection, serum albumin, and prothrombin time were not significant and for convenience are not shown.

Older age1median 56 years0.210.09
Age ≥60 years33.3% (31/93)0.790.55
Male gender71.0% (66/93)0.760.92
Blood type O37.0% (34/92)0.370.016(√) 0.013 (–0.643±.261)
Log(AFP)1median AFP 18.30.000010.30
AFP≥100 ng/ml28.3% (26/92)0.00003(√).00003(2.321±0.679)0.77
Log(Bilirubin)1median Bil 1.900.870.056
Bilirubin≥4.0 mg/dl12.1% (11/91)0.570.007(√) 0.032 (0.775 ± 0.370)
Larger tumor size1median 3.0 cm0.0270.51
size >3.0 cm47.3% (44/93)0.140.74
≥2 tumor nodules29.0% (27/93)0.130.86
Stage I9.7% (9/93)0.0280.022(√) 0.029 (–0.948 ± 0.443)
Stage III11.8% (11/93)0.140.15
Child-Pugh class A32.3% (30/93)0.0210.25
Child-Pugh class C18.3% (17/93)0.470.22
Microwave Ablation23.7% (22/93)0.710.72
Listing year1median Oct., 19990.580.008
Listing year (99-01)66.7% (62/93)0.710.002(√) 0.010 (–0.667 ± 0.263)

The Kaplan-Meier curves in Figure 3A display the significantly higher hazard rate of delisting owing to tumor progression among patients with baseline AFP ≥100 ng/mL. Figure 3B shows a similarly dramatic prognostic influence of AFP among the 82 patients with Stage I/II disease (p = 0.00002); again, the UNOS criteria for delisting was met for these patients. It should also be noted that among Stage I/II patients, the log-rank test comparing the number of tumor nodules (1 vs. ≥2) was not significant (p = 0.12); nor was the comparison of size ≤3 vs. >3 cm among patients with a solitary lesion (p = 0.25).

image

Figure 3. (A) Kaplan-Meier curves for delisting owing to tumor progression for patients with alpha-fetoprotein (AFP) levels ≥100 ng/mL vs. <100 ng/m. (B) Kaplan-Meier curves for delisting owing to tumor progression among initial Stage I/II patients with AFP levels ≥100 ng/mL vs. <100 ng/m.

Download figure to PowerPoint

Prognostic factors for the hazard rate of transplantation

Results of the stepwise Cox regression analysis appear in Table 1B. Listing year 1999–2001 (p = 0.002), bilirubin <4  mg/dL (p = 0.007), blood type O (p = 0.016), and Stage I (p = 0.022) were found to be significant univariable predictors of a lower hazard rate of transplantation. The other factors were not associated with this outcome variable (p > 0.05). The stepwise Cox regression analysis found the same four characteristics to be independently associated with a lower hazard rate of transplantation: listing year 1999–2001 (p = 0.010), blood type O (p = 0.013), Stage I (p = 0.029), and bilirubin <4 mg/dL (p = 0.032). These results remained unchanged if the Cox stepwise regression was re-run (i) excluding the 11 Stage III patients while applying the UNOS delisting criteria, and (ii) censoring the six patients who were followed beyond the date of MELD policy implementation.

The Kaplan-Meier curves in Figure 4 demonstrate the prognostic power of the Cox model results. Patients with bilirubin ≥4 mg/dL, who at this level show clinical signs of jaundice, appeared to be given the highest priority for transplantation. Conversely, Stage I patients (with bilirubin <4  mg/dL) on average had the longest waiting times for transplant. These two groups are shown separately in Figure  4. Among the remaining patients, the Kaplan-Meier curves show a clear separation in the hazard rate of transplantation according to listing year (1997–98 is favorable) and blood type (non-O is favorable). The estimated median times-to-OLT for Groups 1–5 in Figure 4 were 1.2, 1.7, 3.5, 7.6, and 13.2 months.

image

Figure 4. Kaplan-Meier curves for transplantation by bilirubin, tumor stage, listing year and blood type (log-rank test comparison of the five groups yielded p = 0.00001). Group 1: bilirubin ≥4 mg/dL (n = 11, two censored); group 2: bilirubin <4 mg/dL, Stage II or III, blood type not O, listing year 1997–98 (n = 15, none censored); group 3: bilirubin <4 mg/dL, stage II or III, either blood type O or listing year 1999–2001 (n = 39, 12 censored); group 4: bilirubin <4 mg/dL, stage II or III, both blood type O and listing year 1999–2001 (n = 18, six censored); and group 5: bilirubin <4 mg/dL, stage I (n = 8, one censored).

Download figure to PowerPoint

Prognostic factors for the overall probability of delisting owing to tumor progression

Stepwise logistic regression found that AFP ≥100 ng/mL was the only independent predictor of the probability of being delisted owing to tumor progression (p = 0.0011). Logistic model coefficients (± SE) for the intercept and AFP effect were –2.501 (± 0.465) and 1.865 (± 0.622), respectively, and the observed percentages of patients delisted owing to tumor progression were 7.6% (5/66) for AFP <100 and 34.6% (9/26) for AFP ≥100.

Although the observed percentages of patients delisted owing to tumor progression were higher for listing year 1999–2001 [19.4% (12/62) vs. 6.5% (2/31) for 1997–98, p = 0.10], blood type O [17.7% (6/34) vs. 13.8% (8/58) for non-O, p = 0.62], and bilirubin <4.0 mg/dL [16.3% (13/80) vs. 9.1% (1/11) for ≥4.0 mg/dL, p = 0.54], none of these differences was statistically significant. Thus, although these characteristics (including Stage I disease) were associated with significantly lower hazard rates of transplantation (i.e., longer waiting times), their influence on the overall probability of being delisted owing to tumor progression were smaller than that of AFP.

Among Stage I/II patients, the observed percentages who were delisted owing to tumor progression were 5.0% (3/60) for AFP <100, and 31.8% (7/22) for AFP ≥100 ng/mL (p = 0.0010). In comparison, the difference by stage was less: 0% (0/9) for Stage I, and 13.7% (10/73) for Stage II (p = 0.24). As stated earlier, none of the nine Stage I patients had AFP ≥100, whereas 30% (22/73) of the Stage II patients had AFP ≥100 (p = 0.054).

Pathologic review of the explanted tumors

Diagnoses of the 71 explanted tumors were as follows: hepatocellular carcinoma (n = 67), cholangiocarcinoma (n = 2), combined hepatocellular cholangiocarcinoma (n = 1), and regenerative nodule (n = 1). Histologic grade of HCC was well-differentiated in 25, moderately differentiated in 34, poorly differentiated in six, fibrolamellar in one and not graded in one. Microvascular invasion was seen in 17 (24%) patients. Gross involvement into intrahepatic portal vein (Stage IV-A2) and extrahepatic involvement (Stage IV-B) were each seen in two cases. A comparison of the modified TNM stage initially and at explant is shown in Table 2. The initial and pathologic stages were identical in 46 (65%) patients, whereas the pathologic stage was lower in five (7%) cases and higher in 20 (28%) cases. It should be noted that a higher pathologic stage at explant was less likely among patients who received (vs. did not receive) postlisting ablation therapy: 6% (1/18) vs. 36% (19/53), p = 0.014.

Table 2.  Comparison of the modified tumor-node-metastasis (TNM) stage at initial listing (or diagnosis following listing) with the modified TNM stage of the explant
Initial stageStage 0Stage IStage IIStage IIIStage IV-A1Stage IV-A2 and IV-B
  1. Stage I: single tumor <2.0 cm.

  2. Stage II: single tumor ≤5 cm, or two or three tumor nodules all ≤3 cm.

  3. Stage III: single tumor >5 cm, or two or three nodules with at least one >3 cm.

  4. Stage IV-A1: four or more nodules, any size.

  5. Stage IV-A2: gross intrahepatic portal/hepatic vein involvement, regardless of tumor size or number of nodules.

  6. Stage IV-B: any regional nodes involved or metastatic disease, regardless of tumor size or number of nodules (8).

Stage I (n = 8)14 3000
Stage II (n = 58)0241654
Stage III (n = 5)00 2120

HCC recurrence following OLT

Hepatocellular carcinoma recurred in five cases at 3–10 months following OLT. Recurrence sites were graft, bone and brain, bone and lung, graft, and skin and graft. All five patients with recurrence died of their disease. None of these five cases received adjuvant therapy during the waiting period, and microvascular invasion or a high AFP at explant was seen in each case.

Survival

Forty-one deaths were observed: 14 pretransplant and 27 post-transplant. Causes of death among pretransplant patients were HCC progression in 11 and liver failure without progression in three. Causes of death following OLT were tumor recurrence in five, infection in 12, and other/unrelated to OLT in 10. Kaplan-Meier survival curves following delisting owing to tumor progression (n = 14) and following HCC recurrence after OLT (n = 5) yielded similarly poor outcomes: 80% probabilities of death were estimated to occur by 6 months (figure not shown). The Kaplan-Meier 1- and 3-year survival probabilities following transplantation were 73% ± 6% and 62% ± 6% (figure not shown). As there were only five deaths owing to HCC recurrence, their impact on survival following OLT was relatively small.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In recent years there have been a number of reports summarizing clinical outcomes and identifying prognostic factors among HCC patients following an OLT (1–6,23–26). To date, however, we are aware of only two other groups that have focused on the identification of risk factors for delisting among HCC patients listed for OLT (10,11,12). In the Llovet et al. study (10), an increased AFP concentration and a more recent listing year (1996–97 vs. 1989–95) were associated with a significantly higher probability of delisting owing to tumor progression. Patients with a more recent listing year had significantly longer waiting times for transplant in that study (10). The results of our competing risks analysis support these findings in that an elevated baseline AFP level (≥100 ng/mL) was associated with a significantly higher hazard rate of delisting owing to tumor progression, with a consequential detrimental effect on the overall probability of being delisted owing to tumor progression. However, our results appear to contradict those of Yao et al. (11,12) in that we did not find a strong impact of tumor size or number of tumor nodules on the hazard rate of delisting.

Our result for AFP as a prognosticator of tumor progression risk is consistent with the findings of other studies. For instance, AFP is currently used in the diagnosis of HCC; its sensitivity ranges from 33%-65% (27,28). Higher AFP levels have also been associated with larger (29) and poorly differentiated tumors (30), and a correlation between biological aggressiveness of HCC and an elevated AFP has been suggested (31). Lastly, higher AFP has been associated with significantly poorer survival among HCC patients receiving surgical resection (32).

Our associations of a more recent listing year (1999–2001 vs. 1997–98) and blood type O with significantly lower hazard rates of transplantation are consistent with the findings of others (10,13). Our Cox regression model for the hazard rate of transplantation also included the effects of early stage (I vs. II-III) and lower bilirubin (<4 vs. ≥4 mg/dL). Although each of these four characteristics was associated with longer waiting times for transplant, none of them had a significant impact on increasing the probability of being delisted owing to tumor progression. Similar magnitudes of differences would be significant if observed in a large enough cohort of patients. Second, it is likely that a larger impact of these characteristics would exist if the waiting times had been longer. In our group of 71 patients who received an OLT, the median waiting times among the 26 and 45 patients listed during 1997–98 and 1999–2001 were only 2.1 and 3.8 months, respectively. Other HCC studies have reported longer waiting times (10–12).

Under the current UNOS policy (7), HCC patients receive an additional urgency score for transplantation, and the additional amount is determined by the patient's modified TNM stage (I or II). If we consider a simple combination using AFP level with stage, then our data show the following percentages who are delisted owing to tumor progression: 0% (0/9) for Stage I; 6% (3/51) for Stage II and AFP <100 ng/mL; 32% (7/22) for Stage II and AFP ≥100 ng/mL; and 40% (4/10) for Stage III. Our data certainly support two recent conclusions by Yao et al. (12): (1) Stage I and possibly low-risk Stage II patients may not require an additional urgency score, and (2) high-risk Stage II patients clearly require the additional urgency score. Because the sample sizes in studies of HCC patients listed for OLT are relatively small, good power for detecting significant and independent effects of multiple prognosticators may not exist. Thus, decisions regarding further refinement of the HCC-specific urgency score may first require a pooled analysis of data from multiple institutions.

In addition to our reporting the clinical findings, we want to emphasize that the use of the cumulative incidence function (14,15,17,22) to properly estimate the probability of failing from a particular cause by time t in the presence of competing risks must now become accepted practice in the organ transplantation literature. In the Results section we presented examples showing the severe bias that can occur in using the 1-KM estimator. Another example is shown in Figure 3A for the 26 patients with AFP ≥100 ng/mL; clearly, 1-KM estimates the probability of being delisted owing to tumor progression at 100%. Yet, the observed percentage in this subgroup was only 35% (9/26), and in fact 62% (16/26) received a transplant!

With the recent reports showing excellent survival following OLT in patients with early stage HCC (1–6), a proposal was made to expand the eligibility criteria to include patients with larger single or multiple nodules (25). As candidate selection is based on clinical stage, which can underestimate the true stage (2,4,25), a major concern is that by being more inclusive the recurrence rate following OLT will increase. Among our five Stage III patients who received an OLT, two patients who met Yao's expanded criteria have been free of recurrence at 46 and 67 months post-transplant while one patient not meeting the criteria recurred and died. Thus, our data, although sparse, support this expansion idea.

Our results were based on a retrospective review of 93 HCC patients who were listed or diagnosed following listing for OLT at our Center between March, 1997 and September, 2001. It is always possible that no matter how careful a retrospective review is performed that some type of bias may have existed unknowingly. In addition, as this study was performed at a single-center organ procurement organization (OPO) for liver transplants, it is possible that these patients were prioritized in a different fashion than OPOs with more than one center. Given the retrospective nature of this data, we did not have the waiting time-to-transplant information available on listed non-HCC patients. Thus, no such comparison of listed HCC and non-HCC patients could be made.

Finally, there were 22 patients in our study who received ablation therapy. Although the percentage of patients delisted owing to tumor progression was not significantly lower in this group (14%, 3/22), the 18 patients who received an OLT were significantly less likely to have had a higher pathologic stage at explant, and at last follow up no recurrences were observed. Clearly, a determination of the exact role(s) of adjuvant treatment such as TACE or ablation therapy (2,32,33) to reduce the risk of delisting and/or HCC recurrence following OLT will require a sufficiently large sample size. The most appropriate uses of living-donor liver transplantation, split-liver transplantation, and high-risk donors will also need to be determined in patients listed with HCC (34,35).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The authors wish to acknowledge Ms. Moira Morgan for her excellent assistance in the preparation of this manuscript, and are saddened by her recent death.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Romani F, Belli LS, Rondinara GF et al.. The role of transplantation in small hepatocellular carcinoma complicating cirrhosis of the liver. J Am Coll Surg 1994; 178: 379384.
  • 2
    Mazzaferro V, Regalia E, Doci R et al.. Liver transplantation for the treatment of small hepatocellular carcinoma in patients with cirrhosis. N Engl J Med 1996; 334: 693699.
  • 3
    Figueras J, Jaurrieta E, Valls C et al.. Survival after liver transplantation in cirrhotic patients with and without hepatocellular carcinoma: a comparative study. Hepatology 1997; 25: 14851489.
  • 4
    Llovet JM, Bruix J, Fuster J et al.. Liver transplantation for small hepatocellular carcinoma: the tumor-node-metastasis classification does not have prognostic power. Hepatology 1998; 27: 15721577.
  • 5
    Tamura S, Kato T, Berho M et al.. Impact of histological grade of hepatocellular carcinoma on the outcome of liver transplantation. Arch Surg 2001; 136: 2530.
  • 6
    Hemming AW, Cattral MS, Reed AI, Van der Werf WJ, Greig PD, Howard RJ. Liver transplantation for hepatocellular carcinoma. Ann Surg 2001; 233: 652659.
  • 7
    United Network for Organ Sharing. Available at http://www.unos.org, accessed on August 8, 2003.
  • 8
    American Liver Tumor Study Group. A randomized prospective multi-institutional trial of orthotopic liver transplantation or partial hepatic resection with or without adjuvant chemotherapy for hepatocellular carcinoma. Investigators booklet and Protocol, 1998.
  • 9
    El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999; 340: 745750.
  • 10
    Llovet JM, Fuster J, Bruix J. Intention-to-treat analysis of surgical treatment for early hepatocellular carcinoma: resection versus transplantation. Hepatology 1999; 30: 14341440.
  • 11
    Yao FY, Bass NM, Nikolai B et al.. Liver transplantation for hepatocellular carcinoma: analysis of survival according to the intention-to-treat principle and dropout from the waiting list. Liver Transpl 2002; 8: 873883.
  • 12
    Yao FY, Bass NM, Nikolai B et al.. A follow-up analysis of the pattern and predictors of dropout from the waiting list for liver transplantation in patients with hepatocellular carcinoma: Implications for the current organ allocation policy. Liver Transpl 2003; 9: 684692.
  • 13
    Everhart JE, Lombardero M, Detre KM et al.. Increased waiting time for liver transplantation results in higher mortality. Transplantation 1997; 64: 13001306.
  • 14
    Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data, 1st edn. New York : John Wiley and Sons, 1980.
  • 15
    Lawless JF. Statistical Models and Methods for Lifetime Data, 1st edn. New York : John Wiley and Sons, 1982.
  • 16
    Gaynor J, Chapman D, Little C et al.. A cause-specific hazard rate analysis of prognostic factors among 199 adults with acute lymphoblastic leukemia: The Memorial Hospital Experience since 1969. J Clin Oncol 1988; 6: 10141030.
  • 17
    Gaynor JJ, Feuer EJ, Tan CC et al.. On the use of cause-specific failure and conditional failure probabilities: Examples from clinical oncology data. J Am Stat Assoc 1993; 88: 400409.
  • 18
    Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, William R. Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973; 60: 646649.
  • 19
    Seki T, Wakabayashi M, Nakagawa T et al.. Ultrasonically guided percutaneous microwave coagulation therapy for small hepatocellular carcinoma. Cancer 1994; 74: 817825.
  • 20
    Edmondson HA, Steiner PE. Primary carcinoma of the liver. a study of 100 cases among 48,900 necropsies. Cancer 1954; 7: 462503.
  • 21
    Kamath PS, Wiesner RH, McDiarmid et al.. MELD, PELD Application of survival models to liver allocation. Liver Transpl 2001; 7: 567580.
  • 22
    Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of failure probabilities in the presence of competing risks: New representations of old estimators. Stat Med 1999; 18: 695706.
  • 23
    Marsh JW, Dvorchik I, Bonham CA, Iwatsuki S. Is the pathologic TNM staging system for patients with hepatoma predictive of outcome? Cancer 2000; 88: 538543.
  • 24
    Jonas S, Bechstein WO, Steinmuller T et al.. Vascular invasion and histopathologic grading determine outcome after liver transplantation for hepatocellular carcinoma in cirrhosis. Hepatology 2001; 33: 10801086.
  • 25
    Yao FY, Ferrell L, Bass NM et al.. Liver transplantation for hepatocellular carcinoma: expansion of the tumor size limits does not adversely impact survival. Hepatology 2001; 33: 13941403.
  • 26
    Molmenti EP, Klintmalm GB. Liver transplantation in association with hepatocellular carcinoma: an update of the international tumor registry. Liver Transpl 2002; 8: 736748.
  • 27
    Collier J, Sherman M. Screening for hepatocellular carcinoma. Hepatology 1998; 27: 273278.
  • 28
    Taketa K. Alpha-fetoprotein: reevaluation in hepatology. Hepatology 1990; 12: 14201432.
  • 29
    Nomura F, Ohnishi K, Tanabe Y. Clinical features and prognosis of hepatocellular carcinoma with reference to serum alpha-fetoprotein levels. Cancer 1989; 64: 17001707.
  • 30
    Fujioka M, Nakashima Y, Nakashima O, Kojiro M. Immunohistologic study on the expressions of α-fetoprotein and protein induced by vitamin K absence or antagonist II in surgically resected small hepatocellular carcinoma. Hepatology 2001; 34: 11281134.
  • 31
    Stuart KE, Anand AJ, Jenkins RL. Hepatocellular carcinoma in the United States. prognostic features, treatment outcome, and survival. Cancer 1996; 77: 22172222.
  • 32
    Roayaie S, Frischer JS, Emre SH et al.. Long-term results with multimodal adjuvant therapy and liver transplantation for the treatment of hepatocellular carcinoma larger than 5 centimeters. Ann Surg 2002; 235: 533539.
  • 33
    Yamashiki N, Kato T, Bejarano P et al.. Histopathological changes after microwave coagulation therapy for patients with hepatocellular carcinoma: review of 15 explanted livers. Am J Gastroenterol 2003; 98: 20522059.
    Direct Link:
  • 34
    Cheng SJ, Pratt DS, Freeman RB, Jr Kaplan MM, Wong JB. Living-donor versus cadaveric liver transplantation for non-resectable small hepatocellular carcinoma and compensated cirrhosis: a decision analysis. Transplantation 2001; 72: 861868.
  • 35
    Sarasin FP, Majno PE, Llovet JM, Bruix J, Mentha G, Hadengue A. Living donor liver transplantation for early hepatocellular carcinoma: a life-expectancy and cost-effectiveness perspective. Hepatology 2001; 33: 10731079.