Telephone: 33158411734; FAX: 33158411743
Hepatocellular carcinoma developed on compensated cirrhosis: Resection as a selection tool for liver transplantation
Article first published online: 28 MAY 2008
Copyright © 2008 American Association for the Study of Liver Diseases
Volume 14, Issue 6, pages 779–788, June 2008
How to Cite
Scatton, O., Zalinski, S., Terris, B., Lefevre, J. H., Casali, A., Massault, P.-P., Conti, F., Calmus, Y. and Soubrane, O. (2008), Hepatocellular carcinoma developed on compensated cirrhosis: Resection as a selection tool for liver transplantation. Liver Transpl, 14: 779–788. doi: 10.1002/lt.21431
- Issue published online: 28 MAY 2008
- Article first published online: 28 MAY 2008
- Manuscript Accepted: 6 DEC 2007
- Manuscript Received: 26 SEP 2007
The objective of this study was to evaluate the histological profile obtained from primary resection of hepatocellular carcinoma (HCC) as a selection tool for liver transplantation (LT). The natural history of HCC depends on its histological features. The clinical effectiveness of resection as a selection tool for salvage or de principe LT has been previously advocated. Between 1987 and 2006, 20 patients underwent a resection prior to LT. Long-term survival of these 20 patients was compared to that of 73 patients who underwent primary LT. Histological features of the resected specimen were compared to those of the recurrences. Feasibility, morbidity, and mortality of LT following primary resection were also analyzed. Mean follow-up was 3.8 ± 4.4 and 2.7 ± 4.5 years from resection and LT, respectively; 6 patients died. The mean 1-, 3-, 5-, and 10-year overall survival rates were 71%, 61%, 55%, and 45% and 74%, 66%, 66%, and 40% after primary transplantation and primary resection, respectively (not significant). At LT, 14 patients had a recurrence, but histological study of the recurrence was not possible in 2 (complete necrosis). For 9 patients (75%), histological features of both primary and recurrent tumors were exactly the same. Four patients had recurrence following LT; in each case, primary and recurrent nodules shared the same histological markers of poor prognosis. De principe transplantation was proposed to 6 patients because of poor prognosis histological features on the resected specimen. All these patients are alive without recurrence with a mean follow-up of 55 months. In conclusion, the natural history of HCC can be predicted on the basis of the histological profile of the resected specimen, which may be used as a selection tool for LT. De principe LT in patients within Milan criteria with poor prognosis histological features may be an optimal strategy. Liver Transpl 14:779–788, 2008. © 2008 AASLD.
Hepatocellular carcinoma (HCC) is the fifth most common neoplasm worldwide and the fourth most common in mortality. Although more common in Asia and Africa because of a high prevalence of hepatitis B–related and hepatitis C–related cirrhosis, its incidence has doubled in Western countries in the past 20 years.1 Despite a close observation of patients with liver cirrhosis, HCC is often diagnosed at an advanced stage at which no optimal treatment has been established,2–4 and only 20%–25% of patients will benefit from resection or liver transplantation (LT),5 the only chance to improve life expectancy. Optimal treatment depends on the severity of the underlying liver disease. For patients with Child-Pugh B/C cirrhosis whose HCC fulfills the Milan criteria,6 LT is widely accepted as the treatment of choice.7–10 Optimal treatment for HCC patients with compensated cirrhosis (Child-Pugh A) is still controversial, and whether resection or LT offers the best prognosis still remains an issue of debate. More recently, benefits of resection prior to transplantation have been argued in the hepatobiliary surgeons' community.11, 12 Conflicting results have been published by Adam et al.11 and Belghiti et al.,12 so the feasibility and long-term results of this sequential strategy have not been well established. LT following resection for HCC can be performed either de principe or as a salvage approach. A salvage approach is to be considered in the case of recurrence meeting the Milan criteria or liver function deterioration. On the other hand, de principe transplantation has been proposed on the basis that early recurrence may be expected according to the histological analysis of the resected specimen. Although this latter strategy is speculative in terms of effectiveness, there are relevant arguments supporting the idea that histological characteristics of the primary tumor may have an impact in terms of prognosis and recurrence following LT. Long-term prognosis (survival and recurrence) discrepancies observed between patients display the limit of the Milan criteria as a unique selection tool of patients eligible for LT. Although the Milan criteria remain the most widely used selection tool in transplant centers,13–15 this scoring system is inevitably limited as it does not consider histological features of tumors. Actually, tumor differentiation, satellite nodules, and microvascular invasion, which are now universally recognized as prognostic factors,16, 17 may certainly be highly significant in terms of patient selection for LT and graft allocation improvement. Finally, given that histological features may help to predict the natural history of HCC and given the donor shortage, resection may be doubly useful; it is the key to relevant histological information, and above all, it is the best treatment before salvage or de principe transplantation. Interestingly, to date, no study has compared histological features of the primary resected tumor and survival after salvage transplantation, and a fortiori histological features of the recurrent tumors have not been compared to those of the primary one.
Our aim was to evaluate the principle of resection as a selection tool for LT to assess the value of de principe LT in terms of survival and graft allocation optimization, that is, if a histologically based selection of patients for LT may improve the survival of those presented with poor prognosis histological features on the resected specimen.
PATIENTS AND METHODS
All of the patients analyzed in this retrospective cohort study underwent a curative treatment of an HCC at our institution. From March 1987 to April 2006, among 93 patients who underwent LT for HCC, 20 patients (15 males; 75%) underwent primary liver resection as a first-line treatment. Mean age at transplantation was 52 ± 8 years (median age 53, range from 30 to 66 years). All these patients had compensated cirrhosis, a Child-Pugh score of A, and a Model for End-Stage Liver Disease score of 8 (n = 20). The underlying liver disease was related to hepatitis C virus in 10 cases (50%), hepatitis B virus in 7 cases (35%), and alcohol in 3 cases (15%). On preoperative radiological assessment, a median number of 1 nodule (range from 1 to 5 nodules) was diagnosed; the mean size of the largest tumor was 4.2 ± 2 cm (median 4, range from 1.5 to 7.5). Two patients received a transarterial chemoembolization (as a neoadjuvant treatment prior to resection). The patients' characteristics at both resection and LT are summarized in Table 1. The study population and allocated treatment are depicted in Fig. 1.
|At Resection||At Transplantation|
|Male||15 (75%)||15 (75%)|
|Age||52.2 ± 8||53.5 ± 8|
|Virus-related||10 C, 7 B||10 C, 7 B|
|Child A||20 (100%)||17 (85%)|
|Child B and C||0||3|
|Mean tumor number||1.6 ± 1||2.7 ± 2|
|Maximal size (cm)||4.1 ± 1.3||2 ± 1|
|Milan criteria +||12 (60%)||12 (60%)|
|Time interval (months)||—||15.2 ± 9|
Surgical Procedures and Histological Features
All hepatectomies but one were performed through the J-shaped approach; one was performed laparoscopically. Major resection was defined as resection of 3 or more anatomical segments as described by Couinaud. An anatomical resection was performed in 17 patients. Three patients had a major hepatectomy (2 right hepatectomies and 1 right lateral sectionectomy with resection of segment 1). Among the 17 patients who had a minor resection, 4 had a left lateral sectionectomy, 1 had a IV-V bisegmentectomy, and 8 had a segmentectomy (Table 2). Patients selected for LT fulfilled the Milan criteria6 (solitary liver nodule not exceeding 5 cm in maximum diameter or 2–3 nodules not exceeding 3 cm in diameter) on the basis of a preoperative morphological assessment.
|Type of Resection*||n|
|Right lateral sectionectomy + I||1|
|Left lateral sectionectomy||4|
|Right lateral sectionectomy||1|
Liver Transplantation (LT)
Transplantation has been defined as a salvage procedure when patients' enlistment was decided on the basis of a proven hepatic recurrence or liver function deterioration and as a de principe procedure when enlistment was done on the basis of histological criteria of the resected specimen, without evidence of residual disease. By the time of LT, the mean patient age was 54 ± 8 years (median age 54, range from 33 to 65 years). Seventeen patients had a Child-Pugh score of A, whereas 3 were scored B/C. Eighteen of the patients who underwent primary resection received a whole cadaveric liver graft, whereas 2 had a split liver graft (right lobe). Radiological assessment prior to LT showed a mean number of 2.7 ± 2 nodules; the mean size of the largest was 2 ± 1 cm. In order to assess the feasibility and morbidity of LT following primary resection, the following data were recorded: operative mortality, operative time, red cell transfusion, postoperative morbidity (defined as any complication related to the surgical procedure within the 30 postoperative days), intensive care unit length of stay, and reoperation rate. In terms of postoperative morbidity, special attention was paid to postoperative bleeding, bile leak, sepsis, intra-abdominal collection, severe ascitic effusion, hepatic arterial thrombosis, portal vein thrombosis, primary liver dysfunction, and acute rejection. Survival of the 20 patients who underwent primary resection followed by LT was compared to survival of the 73 patients who underwent an LT as a unique treatment during the same period (1987–2006).
All histological analyses of resected specimens, liver explants, and recurrences were repeated for this study and performed by a single pathologist (B.T.) who was unaware of the clinical data. Histological features of the resected specimen were compared to those of recurrences found in the explanted liver for each patient. Microscopic tumoral features were as follows: Edmondson score, predominant architectural growth pattern (microtrabecular, microtrabecular and acinar, macrotrabecular, or compact), nuclear grade (mild, moderate, or severe),18 cholestasis (absent, moderate, or severe), and steatosis (absent or present if more than 25% of tumoral cells were involved). Stratification of the recurrent tumors as related to the primary tumor or de novo was based on the comparison of 3 microscopic histological features. When all were similar on both specimens, we considered the tumors to be related. When 1 of these features appeared different, we considered the recurrent tumor to be de novo. We considered related tumors to correspond to vascular dissemination of the primary tumor, whereas de novo tumors were considered to be a new lesion arising in an underlying oncogenic parenchyma. Microvascular invasion and macrovascular invasion were also specifically assessed and compared both at the time of resection and at LT.
Categorical data were analyzed, after multiple 2 × 2 contingency tables were set up, with the chi-square test or Fischer's exact test as appropriate. Continuous parametric data were analyzed with the Student t test, and nonparametric data were analyzed with the Mann-Whitney U test. Continuous variables are reported as the mean ± standard error of the mean, and the median (range) is specified when required. P < 0.05 was considered statistically significant.
Feasibility of Transplantation Following Primary Resection
The mean delay from resection to transplantation was 15 ± 9 months (median 13 months, range from 2 to 43 months). A temporary portocaval anastomosis was performed in 8 patients (40%). A termino-terminal cavocaval anastomosis was performed in 11 patients (55%); a piggyback procedure was performed otherwise. Mean duration of LT was 8 ± 2 hours (median time 7.5 hours, range from 4.5 to 13 hours) with a mean liver explantation time of 3.4 ± 1 hours (median time 3 hours, range from 1 to 5 hours). In 8 patients (40%), no blood or plasma transfusion was needed; otherwise, the median number of transfused blood cell packs was 8 ± 11 (median 3, range from 2 to 45), and the median number of plasma transfusions was 5.3 ± 10 (median 3, range from 2 to 45). Two patients died within 1 month following LT (1 from primary graft nonfunction and 1 from multiorgan failure), and 1-month mortality was 10%. The postoperative course revealed that 14 patients (70%) had at least 1 postoperative complication, which occurred within the first month after LT in 11 cases (55%). Two patients underwent reoperation because of postoperative complications: 1 patient had an intra-abdominal collection (biliary leak) requiring surgical drainage, and 1 patient underwent reoperation for acute intra-abdominal bleeding. The mean stay in the intensive care unit was 8 ± 8 days (median 5 days, range from 0 to 35 days). The mean follow-up was 4 ± 4 years (median 3 years, range from 1 to 21 years). Details of medical and surgical events are reported in Table 3.
|Type of graft||Split = 2, whole = 18|
|Operative mortality||2 (10%)|
|Mean number of red blood cell transfusions (units)||7 ± 9|
|Operative time (hours)||8 ± 2|
|Liver resection duration||3.4 ± 1|
|Postoperative complications||11/20 (55%)|
|Large amount of ascites||2|
|Hepatic artery thrombosis||0|
|Primary liver graft nonfunction||1|
|Acute rejection||4 (20%)|
|ICU stay (days)||8 ± 7|
Impact of Resection on Survival
After a mean follow-up of 4 ± 4 years from resection (median 3 years, range from 1 to 21 years) and a mean follow-up of 3 ± 4 years from LT (median 1 years, range from 6 months to 20 years), 6 patients had died: 4 patients from recurrence, 1 patient from severe cholangitis, and 1 patient from hepatitis C virus recurrence. The mean 1-, 3-, 5-, and 10-year overall survival rates in patients treated with primary LT and resection prior to LT were 71%, 61%, 55%, and 45% and 74%, 66%, 66%, and 40%, respectively (Fig. 2). There was no statistically significant difference in survival between the 2 groups.
Impact of Milan Criteria at Resection (Fig. 3)
Among the 20 patients who underwent primary resection, 12 were within the Milan criteria (M+) at resection. The mean delay to LT was 14 ± 6 months (median 13 months, range from 5 to 24 months) for these patients. At LT, 9 of 12 (75%) patients had a recurrence that was still within the Milan criteria, whereas the last 3 patients had no recurrence (de principe LT). There was a complete correlation between pretransplant morphological evaluation and histological findings. None of them was beyond the Milan criteria.
For the 8 patients who were beyond the Milan criteria (M−) at resection, the mean delay to LT was 16.7 ± 13 months (median 15 months, range from 2 to 43 months). Four of them (50%) still remained beyond the criteria at LT on pathological examination. Preoperatively, these 4 patients had been evaluated within the Milan criteria on the basis of radiological assessment. Three had no recurrence (de principe transplantation), and only 1 of 8 (12%) patients had a recurrence within the Milan criteria. Although not statistically significant, M+ patients at resection had a better 5-year survival rate (56% versus 25%, not significant) than patients who where M− (Fig. 4). Furthermore, M− patients at LT had a significantly worst prognosis compared to M+ patients at LT (P = 0.04; Fig. 5). The 1-, 3-, and 5-year survival rates in M+ or M− patients at LT were 86.7%, 86.7%, and 65% and 80%, 20%, and 0%, respectively.
Homologies Between the Primary Resected Tumor and Its Recurrence
A histological comparison between initial and recurrent nodules is reported in Table 4. For 2 patients, histological study of the recurrent tumors was not possible because of complete necrosis of the lesion related to the pretransplant treatment (1 radiofrequency and 1 transarterial chemoembolization). Mostly (9 of 12 patients; 75%), microscopic histological features of the recurrent tumor were exactly the same as those of the initial nodule. No local recurrence on the site of the resected tumor was observed. Vascular invasion at the time of resection and LT is reported in Table 4. When present at the time of resection, microvascular invasion was observed in all patients, highlighting the absence of phenotype downstaging. Four patients without initial vascular involvement secondarily developed microvascular invasion at the time of LT. Moreover, 6 patients underwent de principe transplantation, and all of them had vascular invasion at the time of resection. Histological features found in these patients are depicted in Table 5. Overall, most of the patients presenting with initially poor prognostic factors (phenotype and vascular invasion) at the time of resection showed recurrence with the same histological features or worse. Finally, 4 patients experienced recurrence following LT (Table 6); in each case, primary and recurrent nodules shared the same histological markers of poor prognosis. Two of them also had microvascular invasion progression.
|Patient Number||Edmondson||Nuclear Grade*||Architectural Growth Pattern||Vascular Invasion||Tumor on Explant||Recurrence After OLT||Death||Survival (Months)|
|Patient Number||Edmondson||Vascular Invasion||Nuclear Grade*||Architectural Growth Pattern||Follow-Up (Months)||Alive Without Recurrence|
|Patient Number||Milan/Edmonson/Vascular Invasion/Nuclear Grade||Histological Comparison Between Resection and Liver Transplantation|
|At Resection||At Liver Transplantation|
Optimal treatment of HCC in patients with cirrhosis is debated. Yet, controversies are limited to patients with small HCC and compensated liver function. For patients whose liver function precludes any liver resection, transplantation is universally accepted to be the treatment of choice because their HCC fulfills the Milan criteria.6–10, 19 Ongoing debates about the optimal strategy in the case of preserved liver function (Child-Pugh score A) is partly related to organ shortage. Prolonged waiting time is associated with tumor progression and dropout from the waiting list, and this curtails the real benefit of transplantation over resection.20 Dropout rates of 15%–33% have been reported even in countries in which the waiting time for cadaveric liver is relatively short. Some authors have advocated surgical resection to be a good first-line treatment to stop tumor progression and delay the need for transplantation in order to the drawbacks of a long waiting time and donor shortage. Thus, a new strategy has been proposed for patients with preserved liver function and HCC: hepatic resection prior to salvage transplantation, that is, close follow-up of patients upon recurrence and LT enlistment when recurrence is diagnosed.21, 22 This strategy has been fully supported by Poon et al.,23 who reported that nearly 80% of patients who underwent a primary liver resection were still eligible for LT at the time of recurrence. Yet, other groups are not so optimistic when considering the applicability rate of such a strategy, reporting that, because of tumor recurrence, not more than 20% of the primary resected patients are candidates for salvage surgery.20, 24, 25 In fact, tumor recurrence, which is the major drawback after resection of HCC, is the consequence of either primary tumor dissemination or de novo HCC arising in a liver parenchyma with underlying disease.26 As described by Imamura et al.,26 recurrences due to tumor dissemination peak at 1 year postoperatively and are usually associated with an aggressive histological phenotype of the tumor, whereas de novo tumors occur later and are usually associated with underlying disease severity. Considering this later point, to optimize patient selection for LT and avoid an excessive dropout rate from the waiting list among patients who had previously been resected, Sala et al.24 proposed de principe LT for patients classified as high-risk for recurrence on the basis of gross and microscopic examination of the resected specimen.
The first concern of this strategy, resection prior to LT, is its impact on posttransplant survival. As reported by Belghiti et al.,12 we did not observe primary resection to impair overall survival following LT. However, these results should be carefully considered according to the Milan criteria at the time of resection. Indeed, we found that Milan criteria at the time of resection were as relevant as Milan criteria at the time of LT. No patient staged within the Milan criteria at resection had a recurrence staged beyond Milan criteria, and these patients were good candidates for salvage transplantation. On the other hand, salvage transplantation cannot be proposed to patients beyond Milan criteria at resection, given that most of them have recurrence exceeding the Milan criteria on histological analysis. For this subgroup of patients, de principe LT should be discussed according to histological findings on the resected specimen. At the time of resection, the Milan criteria appeared to be the most clinicomorphologically relevant criteria to select patients for salvage or de principe strategy.
Indeed, Milan criteria, which seem to be a good first-line selection tool, are inevitably limited in their sensitivity by the accuracy of the morphological assessment. It is not uncommon to find incidental HCC in liver explants when preoperative imaging did not show evidence of liver tumors. Yet, a histological profile obtained from a resection allows an objective analysis of relevant prognostic factors27 and could be used as an optimal selection tool for LT, that is, to systematically enlist patients whose HCC has poor histological prognostic factors for de principe LT prior to any recurrence. Tumor differentiation, satellite nodules, and microvascular invasion, which are now universally recognized as prognostic factors,16, 17 could certainly be highly significant in terms of patient selection for LT and graft allocation improvement. An original and important finding of our study is that the tumoral phenotype of the recurrence may be predicted on the basis of histological findings of the primary resected tumor. In this study, we found that microscopic histological features observed in the recurrent tumor were often similar to those of the primary one (75%). Moreover, this important result highlights that a poor initial histological prognosis is to be expected for the recurrent tumor if it is present at resection. Indeed, we observed recurrence after LT to be related to poor prognostic factors described both at the time of LT and at resection. In this latter situation, tumors at resection and transplantation shared exactly the same poor histological profile, suggesting that the recurrence could have been predicted on the basis of microscopic examination of the resected specimen. We observed that microvascular invasion was almost always present at the time of LT if present at the time of resection. Moreover, 2 patients who had initially poor histological features presented microvascular progression at the time of LT.
Our results fully support those previously published by Sala et al.,24 who proposed de principe LT to 8 patients who were at high risk of recurrence according to the pathological features of the resected specimen. Among them, 7 presented recurrence, compared to only 2 of the 9 patients estimated to be at low risk of recurrence (P = 0.012). After LT, 1 patient died of extrahepatic dissemination. The others were free of disease after a median follow-up of 45 months. Although small, this series of patients confirms that histological features of the resected specimen help identify patients at higher risk of recurrence and allow optimal enlistment among patients presented with HCC. The authors concluded that their policy is clinically effective and could further improve the outcome of resected patients. On the basis of our results, we also would recommend systematically enlisting patients for de principe LT if microscopic analysis of the resected specimen reveals poor prognosis histological features. This is to be considered only in patients entering the Milan criteria at the time of resection. On the other hand, if poor histological findings are found in patients exceeding Milan criteria at resection, transplantation should be precluded because they are at high risk of recurrence. Patients exceeding Milan criteria at the time of resection but who have good histological prognostic factors may benefit from de principe LT. Finally, if Milan criteria are respected at the time of resection and if the specimen shows good prognostic factors, a salvage transplantation could be decided. Accordingly, 6 patients underwent de principe transplantation on the basis of the histological profile of the resected specimen with apparently good results because they are all still alive without recurrence after a mean follow-up of 55 months.
Finally, as recently reported by Pawlik et al.,28 histological grading of HCC based on needle core biopsy does not correlate with histological features of the resected specimen. Therefore, we think that selection of candidates for LT should be decided on the basis of final histological analysis rather than preoperative biopsy. The primary resection may represent the best selection tool for LT in Child-Pugh score A patients with small resectable HCC.
A second limitation concerning this concept has been the expected increased technical difficulty during LT. Conflicting results have been published concerning the feasibility of transplantation following partial resection in patients with cirrhosis and HCC. Belghiti et al.12 studied 18 secondary LTs and concluded that primary resection did not increase intraoperative and postoperative LT morbidity and mortality. Interestingly, Adam et al.11 published opposite results, with an increased operative mortality of secondary LT compared to that of primary LT (28.6% versus 2.1%; P = 0.0008) and increased intraoperative bleeding. The first main finding of our study was that primary resection impaired neither the LT feasibility nor the postoperative course. These results are consistent with those published by Beaujon's group.12 One of the main differences observed between Paul Brousse11 and Beaujon that could hypothetically explain the increased technical difficulties during LT and worsen the postoperative course was the technical approach used for primary resection. Belghiti et al. reported thoracotomy to be less invasive than laparotomy because of diaphragmatic adhesions, which led to less blood loss during transplant hepatectomy. In our study, all patients but one were operated on through laparotomy, and no significant increase in bleeding or blood transfusion requirement was noted during LT. Given that our results are comparable to those of Beaujon's group and that most of our patients had a minor hepatectomy (85% in our series and 72% in Beaujon's group), it appears to us that the extent of the resection is the main factor that could impair LT feasibility. However, a comparative study should have been made to draw firmer conclusions, but we considered that the long time interval of this study (1987–2006) probably biased the results, given the evolution and refinement of LT techniques and postoperative care during this period.
In conclusion, in most cases, the primary tumor and its recurrence share the same histological features. The natural history of HCC can be predicted on the basis of the histological profile of the resected specimen, which may be used as a selection tool for LT. In stringently selected patients, on the basis of Milan criteria at resection and histological findings on the resected specimen, de principe LT may be an optimal strategy. In addition, we confirmed that primary resection before LT is feasible and has no deleterious consequences in terms of morbidity, mortality, and survival after LT.
- 21Primary liver resection and salvage transplantation or primary liver transplantation in patients with single, small hepatocellular carcinoma and preserved liver function: an outcome-oriented decision analysis. Hepatology 2000; 31: 899–906., , , .