• Metastasis;
  • multivariate analysis;
  • patient survival rate;
  • prognostic factors;
  • transplant registry;
  • transplantation outcomes;
  • tumor


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

Liver transplantation (LTx) for metastatic endocrine tumors (MET) remains controversial due to the lack of clear selection criteria. From 1989 to 2005, 85 patients underwent LTx for MET. The primary tumor was located in the pancreas or duodenum in 40 cases, digestive tract in 26 and bronchial tree in five. In the remaining 14 cases, primary location was undetermined at the time of LTx. Hepatomegaly (explanted liver ≥120% of estimated standard liver volume) was observed in 53 patients (62%). Extrahepatic resection was performed concomitantly with LTx in 34 patients (40%), including upper abdominal exenteration (UAE) in seven.

Postoperative in-hospital mortality was 14%. Overall 5-year survival was 47%. Independent factors of poor prognosis according to multivariate analysis included UAE (relative risk (RR): 3.72), primary tumor in duodenum or pancreas (RR: 2.94) and hepatomegaly (RR: 2.63). After exclusion of cases involving concomitant UAE, the other two factors were combined into a risk model. Five-year survival rate was 12% for the 23 patients presenting both unfavorable prognostic factors versus 68% for the 55 patients presenting one or neither factor (p < 10−7).

LTx can benefit selected patients with nonresectable MET. Patients presenting duodeno-pancreatic MET in association with hepatomegaly are poor indications for LTx.


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

Endocrine tumors are rare neoplasms that typically have an indolent natural history. The main cause of death is liver metastasis that is often present at the time of diagnosis and remains confined to the liver for protracted periods of time. Some liver metastases are asymptomatic while others produce highly debilitating symptoms due to hormone secretion or pain due to tumor-related hepatomegaly. Since liver metastases are usually multi-focal and diffuse, they are not amenable to partial hepatectomy (1–6). Liver transplantation (LTx) has been proposed as a possible treatment for metastatic endocrine tumors (MET) (7,8). However, there are no clear selection criteria.

In Europe, MET accounts for only 0.4% of LTx (9). Several factors may account for this low utilization rate. The first is that no meaningful data have been published to show that life expectancy after LTx is better than spontaneous survival, i.e. 20% to 30% at 5 years (2,3). Another factor is that numerous treatment alternatives can be proposed to patients with extensive MET including antihormone or interferon therapy, systemic or intraarterial chemotherapy, ischemic therapy and more recently radionuclide therapy. A third limiting factor is that allocation of scarce donor organs to patients with malignancies can only be justified if expected survival is over 50% at 5 years (10).

To assist selection of MET patients who could benefit from LTx, it would be helpful to identify prognostic factors. Ten years ago we compiled a 31-case series including all patients who underwent LTx for MET over a 6-year period (1989 to 1994) in France (11). Data analysis showed that survival after LTx was much better in patients with carcinoid than pancreatic MET (69% at 5 years vs. 8% at 4 years, p < 0.001). This difference was attributed to higher postoperative mortality associated with multi-visceral resection or transplantation as well as to greater severity of tumor recurrence in the pancreatic MET group. We concluded that LTx should be proposed to patients with carcinoid MET from lung and digestive tract but not to patients with pancreatic MET. Several authors took our findings into account in selecting patients for LTx (5,8,12). However, others were unable to confirm the impact of primary tumor location on the outcome of LTx for MET (13–16).

A number of single-center studies attempted to identify other prognostic factors but were inconclusive due to small cohort size as well as to lack of sufficient follow-up data (16–21). Only one recent series has provided useful insight (22). The purpose of this retrospective multi-centric study was to confirm or refute the results of our previous study and to ascertain prognostic factors for long-term survival after LTx for MET based on multivariate analysis in a large patient population.


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

Data collection

Data were compiled retrospectively from the 21 LTx centers in France. For each patient the following data were collected: pretransplant disease and treatment history; transplantation procedures; pathological findings; follow-up data including postoperative course, long-term results, type and delay of recurrence and cause of death. The extent of metastatic liver involvement was evaluated based on CT-scan sections made prior to LTx and expressed as an approximate percentage of liver replacement. Hepatomegaly was arbitrarily defined as enlargement of the explanted liver by 20% or more beyond the patient's normal liver volume determined using the formula of Heinemann et al. (23): liver volume = 1072.8 × body surface area (m2)–345.7.

The indication for LTx was decided at each center. However, for the purpose of this study, we classified indications into three general categories, i.e. ‘hormonal syndrome’ in patients presenting life-threatening or debilitating hormone-related symptoms, ‘tumor bulk’ in patients presenting pain or debility associated with enlargement of the liver and ‘oncological’ in patients with low-grade symptoms.

Pathology reports were reviewed to allow reclassification of tumors as well and poorly differentiated malignant endocrine tumors irrespective of primary tumor location or hormone production according to the new WHO criteria (24). Immunohistochemical markers such as Ki67 (16) and tumor markers such as neuron-specific enolase or chromogranin A were not analyzed because they were available in only a few cases.

Statistical methods

Data were expressed as mean ± S D or as median (range) depending on distribution. Comparisons were made using the Student, Mann-Whitney or chi-square test. Follow-up data was evaluated on June 30, 2006. The main evaluation criterion was overall 5-year survival. For univariate analysis, the correlation between each qualitative covariate and the lifetime variable was determined using the log-rank test. Quantitative variables were analyzed in the same way after first being recoded in two or three strata according to clinical and statistical findings. Pretransplant variables (predictors) with p-value <0.20 in univariate analysis were tested in the multivariate Cox proportional hazards model to select prognostic variables using a step-by-step approach. A p value ≤0.05 was considered as evidence of independent statistical significance. Adjusted relative risk (RR) and 95% confidence intervals (95% CI) were calculated for variables used in the final model. A risk model was created using a two-point scale (0 or 1) for each variable identified by multivariate analysis. The power of the model in predicting long-term survival was determined by analyzing risk groups with Kaplan-Meier curves and the log-rank test. This analysis was performed using SAS software version 9.1 (SAS Institute, Inc., Cary, NC).


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

Population and disease

From February 1989 to December 2005, a total of 85 patients underwent LTx for MET in 17 centers in France (Table 1). This series includes the previous series of 31 patients operated on between 1989 and 1994 (11). There were 39 women and 46 men with a mean age of 45 ± 11 years (range: 18–64) at the time of LTx.

Table 1.  Distribution of cases by center
City, HospitalPeriod 1 1989–1994Period 2 1995–2005Total
Clichy, Beaujon012 12 
Lyon, E. Herriot189
Paris, Cochin819
Marseille, La Conception538
Besançon, Jean Minjoz257
Créteil, Henri Mondor527
Villejuif, Paul Brousse156
Strasbourg, Hautepierre055
Caen, Côte de Nacre314
Paris, St Antoine224
Toulouse, Rangueil044
Lille, Claude Huriez033
Paris, Bichat202
Lyon, Croix Rousse022
Montpellier, St Eloi101
Rennes, Pontchaillou101
Grenoble, Michallon011
Total31 54 85 

The primary tumor was located in the duodenum or pancreas in 41 cases (DP group) including two gastrinomas presumably located in a peripancreatic lymph node (24,25). The primary tumor was located in the bronchial tree in five cases and in the digestive tract (DT) in 29. Tumors in these two locations were included in the same group because they share the same embryological origin (foregut) and often have similar clinical presentation (carcinoid syndrome). In the remaining 10 cases, the primary tumor site was undetermined (UD group) at the time of LTx despite extensive investigation often including one or more diagnostic laparotomies. It should be pointed out that four primary tumors were detected during (n = 1) or after (n = 3) LTx including three small bowel tumors in patients undergoing LTx for presumed hepatocellular carcinoma and one pancreatic tumor that was discovered and treated by Whipple resection 18 months after LTx. To enhance the pertinence of prognostic analysis using data available before LTx, these four tumors were included in the UD group. As a result the number of patients in the DP, DT and UD groups was 40, 31 and 14, respectively. Symptoms of hormonal secretion were observed in 43 cases (51%). Distribution of cases according to primary tumor site and secretion status is presented in Table 2.

Table 2.  Distribution of cases according to primary tumor location and secretion status
Primary tumor siteFunctional tumorsNonfunctional tumorsTotal
  1. This table takes into account the definitive location of the primary tumor.

  2. 1carcinoid 3; 2carcinoid 2; VIPoma1; 3carcinoid 14; 4gastrinoma 13; VIPoma 2; glucagonoma 2; insulinoma 2; carcinoid 1; polymorph 1; 5carcinoid 2.

Bronchial tree 31 2 5
Stomach  3 3
Jejunum 32 3 6
Ileum143 216
Sigmoid colon, rectum  4 4
Duodenum, pancreas2142041
Undetected 25 810

Liver metastases were detected at the same time as the primary tumor in 49 cases, and within a median delay of 36 months (range, 10–84) after the primary tumor in 15 cases. In 21 cases including all 14 in the UD group, liver metastases were diagnosed before detection of the primary tumor. The median percentage of metastatic liver involvement was 50% (range, 5–90%). Percentage of metastatic liver involvement was <20% in 14 cases, between 20 and 40% in 24, between 40 and 60% in 23 and >60% in 24.

Pretransplant treatment history

A total of 68 patients (80%) had surgical treatment prior to LTx. Pretransplant procedures included ablation of the primary tumor in 55 cases (65%), one or more hepatic resection procedures (n = 28) in 23 and surgical exploration of the tumor or metastasis in 12 (Table 3). Primary tumor resection was more frequent in the DT group than in the DP group: (29/31 (94%) vs. 26/40 (65%), p < 0.01).

Table 3.  Pretransplant surgical history
Surgical proceduresN
Small bowel resection18
Distal pancreatectomy with splenectomy16
Whipple procedure10
Pulmonary lobectomy 5
Colectomy, proctectomy, transanal tumorectomy 4
Partial or total gastrectomy 2
Major or minor hepatectomy ± tumorectomy17
Atypical hepatic resection and/or radiofrequency ablation11
Exploratory laparotomy/laparoscopy12

A total of 70 patients (82%) received chemotherapy prior to LTx. The most common protocol was 5 FU-streptozotocin. Transarterial chemo-embolization (TACE) or embolization was used in 35 patients. Only eight patients (9%) had undergone no medical or surgical treatment prior to LTx.

Somatostatin receptor scintigraphy (octreoscan®) was performed before LTx in 49 cases. Octreoscan® was not performed in 25 cases and not specifically mentioned in the patient's record in 11 cases. Octreoscan® was performed in 80% of patients (43/54) after 1995 versus 19% (6/31) before 1995 (p < 0.0001).

Indication for transplantation

The indication for LTx was oncological in 36 cases (42%), tumor bulk in 23 (27%) and hormonal syndrome in 20 (24%). Of the remaining six cases (7%) three involved presumed hepatocellular carcinoma including one case of fibrolamellar variant and one case associated with posthepatitis B cirrhosis. These cases correspond to the three patients in whom small bowel primary tumors were discovered during or after LTx. The remaining three patients presented severe complications of TACE, i.e. ischemic cholangitis in two cases and subacute liver failure in one case. Tumor sites and indications for LTx are summarized in Table 4. Median interval between diagnosis of metastasis and LTx was 30 months (range, 1–128). As shown in Table 4, the interval between diagnosis of metastasis and LTx was comparable according to indication for LTx.

Table 4.  Distribution of cases according to primary tumor site and indication for transplantation
Indication for LTxDP GroupDT GroupUD GroupTotalMedian delay (range)
  1. DP = duodenum/pancreas; DT = digestive tract; UD = undetermined; NR = not recorded.

Oncological19 9 836 29 (34–128)
Hormonal syndrome 911 02029 (6–110)
Tumor bulk12 8 32332 (1–122)
Other 0 3 3 617 (5–60) 
Total4031148530 (1–128)

Transplantation procedures

Seventy-eight patients (92%) received a full-size graft. In the remaining cases, a right split liver graft was used in three cases, a domino liver graft in two and a right liver graft obtained from a living relative in two. Three of the seven patients who underwent upper abdominal exenteration (UAE) received an en bloc composite liver-duodenum-pancreas graft. A total of 34 patients underwent concurrent extrahepatic resection (Table 5).

Table 5.  Surgery in addition to LTx
Surgical proceduresN
Major procedures 
 Upper abdominal exenteration 7
 Whipple procedure 6
 Total gastrectomy 1
Minor procedures 
 Diaphragm resection 8
 Extended lymph node dissection 7
 Small bowel resection 2
 Peritonectomy 2
 Distal pancreatectomy 1

Pathological findings

The mean weight of the explanted liver was 2922 ± 2025 g (range, 1147–13 000). Mean weight was significantly higher (p < 0.0001) in patients undergoing LTx for tumor bulk (5142 ± 2437 g) than for oncological (1974 ± 719 g) or hormonal indications (2272 ± 1007 g).

According to our criteria, hepatomegaly was present in 53/85 cases (62%). There was a significant difference in the estimated extent of liver involvement in function of whether or not hepatomegaly was present (median, 60% vs. 20%, p < 0.0001). However, some values overlapped between groups with percentage of involvement ranging from 5% to 75% among patients without hepatomegaly and from 25% to 90% among patients with hepatomegaly (Figure 1).


Figure 1. Liver involvement according to the liver size.

Download figure to PowerPoint

Accurate information on the number of metastasis was not available in all cases. Based on information available for 63 specimens, lymph node status was positive in 35 cases (55%). Tumors were classified as well differentiated in 74 cases (87%) and poorly differentiated in 11 (13%). Resection margins were classified as R0 in 74 cases (87%), R1 in 7 and R2 in four.


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

Postoperative data and risk factors

Twelve patients (14%) died during the postoperative period (interval, 2 to 157 days) due mainly to surgical complications such as hemorrhage, pancreatitis and sepsis. Three-month mortality was also 14%. Risk factors for postoperative death included DP primary tumor (24% vs. 4.5% for other locations, p < 0.001), use of UAE (57% vs. 10% without UAE, p < 0.001) and requirement for early re-LTx (57% vs. 10%, p < 0.001). Reoperation was required for complications in 30 patients (35%) including 7 re-LTx. The median duration of hospitalization was 26 days (range, 2–180).

Long-term follow-up

No patient was lost from follow-up. Mean duration of follow-up was 46 ± 47 months (range, 0–202). In 68 cases (80%) LTx had been performed 5 or more years before the end of follow-up. The cause of death after discharge was tumor recurrence in all but five patients who died due to delayed surgical complications with no sign of recurrence. Overall actuarial survival was 72%, 67%, 59%, 51% and 47% at 1 to 5 years respectively (with 24 patients still at risk). Median survival was 56 months (range, 0–202). Disease-free survival was 56%, 45%, 37%, 22% and 20% at 1 to 5 years, respectively (Figure 2). Survival after diagnosis of liver metastases was 92%, 88%, 78%, 73% and 69% at 1 to 5 years, respectively.


Figure 2. Overall and disease-free survival.

Download figure to PowerPoint

Prognostic factors

Univariate analysis (Table 6) showed that the following factors were indicators of good prognosis: 1995–2005 treatment period, primary tumor site in the digestive tract or undetermined, resection of primary tumor prior to LTx, serotonin secretion, assessment using Octreoscan®, R0 resection and well differentiated tumor. The indicators of poor prognosis were primary tumor site in the duodeno-pancreas (Figure 3), gastrin secretion, hepatomegaly and UAE. It is noteworthy that age over 50 years and percentage of liver involvement was not discriminating factors. For percentage of liver involvement the difference was almost significant between patients divided into two groups using a cut-off of 40%.

Table 6.  Univariate analysis of prognostic factors
Prognostic variablesStratification (n)5-year survivalMedian survivalp-Value
Study period1989–1994 (31)36%19 months0.0392
1995–2005 (54)52%>60 months 
Patient age<50 years (54)46%49 months0.8838
≥50 years (31)49%40 months 
Primary tumor siteDuodenum/Pancreas (40)27%18 months0.0008
Digestive tract (31)69%>60 months 
Undetermined (14)59%>60 months 
Duodenum/Pancreas (40)27%18 months0.0002
Others (45)65%>60 months 
Liver involvement<20% (14)62%>60 months0.2391
20–40% (24)60%>60 months 
40–60% (23)35%28 months 
60–90% (24)40%39 months 
≤ 40% (38)60%>60 months0.0538
>40% (47)37%35 months 
HepatomegalyYes (53)34%30 months0.0037
No (32)72%>60 months 
Hormonal syndromeYes (43)45%42 months0.2493
No (42)50%58 months 
Serotonin secretionYes (22)67%>60 months0.0062
No (63)40%38 months 
Gastrin secretionYes (13)23%21 months0.0392
No (72)51%>60 months 
Primary tumor resection before LTxYes (55)54%>60 months0.0423
No (30)32%26 months 
Hepatic resection before LTxYes (23)52%>60 months0.1629
No (62)46%56 months 
Chemotherapy before LTxYes (70)47%46 months0.3050
No (15)46%58 months 
TACE before LTxYes (35)62%>60 months0.1060
No (50)38%38 months 
Octreoscan® assessmentYes (49)58%>60 months0.0074
No (25)36%12 months 
Indication for LTxTumor bulk (23)30%24 months0.0512
Hormonal (20)58%>60 months 
Oncological (36)45%56 months 
Other (6)83%>60 months 
Delay from diagnosis of metastases to LTx<36 months (48)39%41 months0.0841
≥36 months (37)60%>60 months 
Upper abdominal exenterationYes (7)0% (3 year)1.5 months<0.0001   
No (78)52%>60 months 
Resection marginsR0 (74)53%>60 months0.0002
R1-R2 (11)9%10 months 
HistologyWell-differentiated (74)49%57 months0.0488
Poorly differentiated (11)27%10 months 

Figure 3. Overall survival according to the primary tumor site.

Download figure to PowerPoint

Only prognostic factors available before LTx were used in multivariate analysis since they are the only ones useful for patient selection. Accordingly primary tumor site was taken into consideration insofar as it was determined before LTx. Although hepatomegaly was determined postoperatively in this study, it was analyzed because it can now be readily determined before LTX in clinical practice. Conversely study period and resection margins (R0 vs. R1/R2) were not taken into account. Octreoscan® assessment was not taken into account since it has become a routine procedure. In order of decreasing RR, the prognostic factors identified by multivariate analysis were UAE, DP primary tumor site and hepatomegaly (Table 7). When UAE was discounted, DP primary tumor (RR: 3.06, p = 0.002) and hepatomegaly (RR 2.98, p = 0.01) displayed the same prognostic value.

Table 7.  Multivariate analysis of prognostic factors
Prognostic variableRelative risk95% CIp-Value
  1. DP = duodenum/pancreas.

Upper abdominal exenteration3.721.54–8.950.0034
Primary tumor site in DP2.941.49–5.790.0018

A prognostic score was devised after exclusion of the seven cases involving UAE. The remaining two variables were scored as follows: 0 point for DT or UD primary site and 1 point for DP primary site; and 0 point for no hepatomegaly and 1 point for hepatomegaly. Using this scoring system retrospectively, patients were classified into three groups, i.e. 0 (20 cases), 1 (35 cases) or 2 points (23 cases) with 5-year survival rate of 76%, 66% and 12% respectively (Figure 4). There was not a significant difference in survival between patients with 0 and 1 point. Nor was there a difference in survival among patients with 1 point according to whether the primary tumor was located in the DT with hepatomegaly or in the DP without hepatomegaly. However, there was a highly significant difference (p < 10−7) between the survival of the 55 patients with 0 or 1 point (68% at 5 years) and the 23 patients with 2 points, i.e. DP primary tumor associated with hepatomegaly (12% at 5 years). Only two of 23 two-point patients survived more than 5 years.


Figure 4. Actuarial survival according to the risk model score.

Download figure to PowerPoint


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

This series, which is the largest in the literature, shows that LTx for MET leads to an overall survival rate of 47%. Although this rate seems disappointing, it must be recalled that patients indicated for LTx represent a ‘worst-case’ group because LTx is generally used only after all other alternatives have been exhausted (5,15). In our series only 9% of patients were untreated prior to LTx. Five-year survival after diagnosis of metastasis was 69%, i.e. far better than the 20–30% reported in past series of untreated MET (2,3). Comparison of the two study periods showed that results of LTx improved significantly over time with survival over 50% at 5 years for the period from 1995 to 2005. Two possible explanations for this improvement are better pretransplant assessment with more widespread use of octreoscan® and discontinued use of multi-visceral resection. The seven patients that underwent UAE were treated in 1989 and 1990 shortly after the description of the procedure by Starzl et al. (26). Despite the small number of cases, use of UAE, which we already reported as deleterious (11), was identified as an independent factor of poor prognosis. This finding is in agreement with Lehnert's multivariate analysis that included UAE and Whipple procedures (14). Based on these findings it seems advisable to avoid UAE and, if necessary, to use a resection technique that preserves part of the stomach and/or pancreas.

Our multivariate analysis identified two additional independent but synergetic predictors of poor prognosis, i.e. DP primary tumor site and hepatomegaly. DP primary tumor site remained a predictor of poor prognosis even after exclusion of UAE that had a strongly negative impact on this group (4,27). These findings also confirm those of our previous study (11). In this study, patients in the UD group had the same prognosis as patients in the DT group (Figure 3). However, this finding must be tempered due to a possible distribution bias since the three 5-year survivors in the UD group are also the ones whose primary tumor was discovered and treated after LTx. In the remaining 11 patients, follow-up was less than 5 years (range, 7–56 months). Failure to locate the primary tumor is common in patients with MET, i.e. 16% in our series, as in others (1). In a 19-case series the primary tumor was located 25 and 46 months after LTx in two cases and remained undetermined in another (16). A possible explanation for the difficulty in determining primary location is that the liver tumor itself is the primary (28,29). However, this seems unlikely not only because primary hepatic endocrine tumors are extremely rare (29,30) but also because the multi-nodular appearance of the lesions in our series is inconsistent with the usual appearance of primary hepatic endocrine tumors that are typically solitary and centrally located (28). Primary carcinoid tumors of the liver are possible indications for LTx (28). However, the use of LTx for MET from undetermined primary is contraindicated by some (16,21) and accepted by others (10,18,31)

The second more novel predictor identified by our multivariate analysis was hepatomegaly arbitrarily defined as enlargement of 20% or more beyond standard liver volume calculated according to the formula of Heinemann et al. (23) Another formula described for western patients (32) gave similar results. The 20% enlargement cut-off is compatible with clinically detectable hepatomegaly while being much lower than the volume observed in patients indicated for LTx due to tumor bulk (mean weight, >5 kg). Although this definition was based on the weight of the posttransplant surgical specimen, measurement can be made preoperatively using CT-scan volumetry. In clinical practice, assessment of hepatomegaly is a more accurate parameter than percentage of hepatic involvement that is difficult to determine in case of scattered, poorly visible multiple lesions. In this regard no previous MET series has described the method of calculating hepatic involvement (1,12,17,18,22). In our series assessment of hepatomegaly was shown to be a more discriminating prognostic factor than percentage of liver involvement regardless of the cut-off used for stratification. Hepatomegaly was well correlated with estimated liver involvement although replacement up to 75% was observed in patients without hepatomegaly (Figure 1).

The only previous multivariate analysis on LTx for MET was by Lehnert (14) who studied 103 cases compiled from the literature. Findings identified two predictors of poor prognosis, i.e. age >50 years and concomitant multi-visceral resection (UAE or Whipple procedure). Although it is frequently cited, this study was not a meta-analysis and contained two major biases, i.e. analysis of selected cases and heterogeneous follow-up. It is interesting to note that our series did not confirm the age-related risk.

Rosenau et al. (16) investigated the impact of tumor biology on survival. Expression of immunohistochemical Ki67 (cell proliferation marker) and E-cadherin (metastatic potential marker) were retrospectively studied on explanted livers. Findings showed that survival in five patients with low Ki67 (< 5%) and regular E-cadherin staining was significantly better than in 12 patients with high Ki67 or aberrant E-cadherin expression (100% vs. 0% at 7 years). However, these promising results were not confirmed in a prospective series (21). Immunohistochemical data were not available in our retrospective study.

Although not specifically mentioning prognostic factors, several recent reports described outcomes of LTx for MET in function of selection criteria (Table 8). A group in Milan (12,22) obtained excellent results using the following selection criteria for LTx with curative intent: age ≤ 55 years, carcinoid MET, primary in portal system, <50% liver involvement, resection of primary tumor prior to LTx and stability of lesions during the observation period. However, with such stringent criteria they probably excluded many patients who might have benefited from LTx according to our criteria. Moreover, one can rightly wonder what the 5-year survival in this carefully selected cohort of patients including 90% asymptomatic cases would have been without LTx. Another group from Göteborg (19,20) reported good results in a group of older patients with larger well differentiated tumors exhibiting a Ki67 < 10%. The Mayo Clinic group (21) used the following selection criteria: well differentiated tumor, pretransplant assessment using octreoscan®, systematic laparotomy and resection of the primary tumor within 6 months. The main drawback of the study was short follow-up. In another report focusing on patient selection for LTx, the Essen group (18) contraindicated 10 of 28 (36%) eligible patients after assessment using octreoscan® detected extrahepatic metastases not diagnosed by standard imaging procedures. However, all patients were symptomatic due mainly to carcinoid hormonal secretion or tumor bulk.

Table 8.  Liver transplantation for metastatic endocrine tumors. Literature review: 2000–2007
Author, year, (ref)Study periodNo. of patientsPrimary tumor site% Liver involvementPatients with symptomsOperative mortalityPatient survival
  1. DP = duodenum/pancreas; DT = digestive tract; UD = undetermined; NR = not recorded;1 Two other primary tumors were undetermined at the time of LTx; 2Possible primary hepatic tumor; 3Including 10 cases of LTx only and 5 cases of multi-visceral resection/transplantation for pancreatic head tumor; 4One other primary tumor was undetermined at the time of LTx; 5Survival rate calculated from the data of the paper; it was 90% for the 10 patients undergoing LTx only.

Rosenau, 2002, (16)1982–19971910 8 11NR90%5%80% at 5 years
Florman, 2004 (17)1992–200211 7 3 1NR100%27%36% at 5 years
Frilling, 2006 (18)1992–200415 7 7 1median 88%100%27%67% at 5 years
VanVilsteren, 2006 (21)1998–20051911 7 12NRNR 5%77% at 2 years
Olausson, 2007 (20)1997–2005 15310 4 14>50% in 12 cases# 60%13%71% at 5 years5
Mazzaferro, 2007 (22)1987–200624 024 0<50% in all cases10%NR90% at 5 years
Present series1989–200585403114median 50%58%14%47% at 5 years

Based on the results of our retrospective study and review of the literature, LTx for MET appears useful in the following cases: (i) Operable patients regardless of age. Heart function in patients is a special concern for patients with carcinoid tumors (18) who may require valve replacement prior to LTx (21); (ii) Patients with unresectable MET causing incapacitating hormone- or tumor-related symptoms (14,15,17–19) or even low-grade symptoms (22,33) as ‘oncological indications’; (iii) Patients with well differentiated MET with a Ki67 index <10% (16,18,19,22); (iv) Patients in whom no extrahepatic disease can be documented by thorough assessment using octreoscan® and exploratory laparotomy (5,15,18,21); (v) Patients who have undergone resection of the primary tumor, thus providing a period of observation before undertaking LTx (5,21,22) and avoiding the risk of performing major resection concurrently with LTx (5,33).

Our analysis also allowed us to develop a useful patient selection tool based on primary tumor location and liver size. A survival rate much higher than the 50% threshold needed to validate LTx in cancer patients (10) can be achieved in patients with scores of 0 or 1, i.e. patients presenting MET with bronchial, digestive or undetermined primaries (with some restrictions in the latter case) regardless of liver size, or patients presenting MET with pancreatic primaries without hepatomegaly. Conversely LTx seems futile in patients with a score of 2, i.e. presenting MET with pancreatic primaries in association with hepatomegaly. If this patient selection scoring tool had been used in the present series, it would have allowed 70% of patients to benefit from LTx with a 5-year survival rate of 68%. In the 30% of patients that would have been contraindicated for LTx only two 5-year survivors would have been wrongfully excluded. However, due to the retrospective analysis and heterogeneous population of our study, this selection tool will require further confirmation in a larger population of MET patients.


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

The authors thank Delphine Laurenceau, MSc, Clinical Trial Center, Hôpital La Timone, Marseille, for her expert contribution to statistical analysis, and all the others who assisted data collection: Olivier Chazouillieres, Department of Hepatology, Hôpital Saint Antoine, Paris; Guy Benhamou, MD, Department of Surgery, Hôpital Bichat, Paris; Francis Navarro, MD, Department of Digestive Surgery, Hôpital Saint Eloi, Montpellier; Karim Boudjema, MD, Department of Surgery, Hôpital Pontchaillou, Rennes; Christian Letoublon, MD, Department of Surgery, Hôpital Michallon, Grenoble, France.


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