Fax: (212) 639-4031
Combined hepatocellular and cholangiocarcinoma
Demographic, clinical, and prognostic factors
Article first published online: 28 MAR 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 7, pages 2040–2046, 1 April 2002
How to Cite
Jarnagin, W. R., Weber, S., Tickoo, S. K., Koea, J. B., Obiekwe, S., Fong, Y., DeMatteo, R. P., Blumgart, L. H. and Klimstra, D. (2002), Combined hepatocellular and cholangiocarcinoma. Cancer, 94: 2040–2046. doi: 10.1002/cncr.10392
- Issue published online: 28 MAR 2002
- Article first published online: 28 MAR 2002
- Manuscript Accepted: 29 NOV 2001
- Manuscript Revised: 1 NOV 2001
- Manuscript Received: 15 AUG 2001
- hepatocellular carcinoma;
- combined tumors;
Tumors with combined hepatocellular and cholangiocellular features are well known histopathologically but their clinical behavior is poorly understood. The objectives of the current study were to define the demographic profile of the patients in whom these uncommon tumors occur and to evaluate treatment outcome in comparison with that in patients with either hepatocellular carcinoma (HCC) or peripheral cholangiocarcinoma (CC) alone.
Twenty-seven patients with combined tumors were identified from a prospective database. Pathologic specimens were analyzed to confirm the diagnosis. Demographics, clinical data, and survival were analyzed. Outcome after resection was compared with that of patients with CC and with a matched group of patients with HCC.
The gender distribution of the combined tumors (52% men and 48% women) was intermediate between HCC (67% men and 33% women) and CC (30% men and 70% women) (P = 0.03). The incidence of positive hepatitis B or C serology and cirrhosis was similar in patients with combined tumors and those with CC (15% and 0% vs. 13% and 4%, respectively); similarly, patients of Asian heritage constituted 7% and 9%, respectively, of the patients with these tumors. By contrast, cirrhosis (41%) and positive hepatitis serology (56%) were far more common in patients with HCC, and 19% of these patients were of Asian heritage. Twenty-one of 27 patients with combined tumors (78%) underwent resection. All 6 patients with combined tumors that were not amenable to resection died of disease within 18 months. After resection, the 5-year survival was lowest in patients with combined tumors (24%) but was not significantly different from that in patients with CC (33%) or HCC (37%). The liver was the most common site of recurrence in all three groups.
The demographic and clinical features of patients with combined tumors were most similar to those of patients with CC. Most important, combined tumors were not found to be associated with chronic liver disease; consequently, the resectability rate was higher for these tumors than typically is reported for HCC. Resection was associated with long-term survival in some patients, but recurrent hepatic disease was common. The presence of cholangiocellular differentiation appeared to worsen the prognosis when compared with pure HCC, although this difference did not reach statistical significance. Cancer 2002;94:2040–6. © 2002 American Cancer Society.
Combined hepatocellular-cholangiocarcinoma is a rare form of primary liver carcinoma comprised of cells with histopathologic features of both cholangiocarcinoma (CC) and hepatocellular carcinoma (HCC). It first was described comprehensively in 1949 by Allen and Lisa1 but many of the demographic and clinical features of these tumors remain unclear. In particular, it has been reported that combined tumors are diagnosed more commonly in those patients with cirrhosis and chronic hepatitis.2–4 In addition, some investigators have demonstrated a conspicuous male predominance,2, 3 but to our knowledge this has not been a consistent finding.4 Likewise, the reported results of resection have been inconsistent. Nakamura et al.,5 in an analysis of 6 patients, reported a median survival of 48 months after resection whereas other authors have suggested a much less favorable prognosis of < 12 months.3 Furthermore, comparing the outcome of patients with a combined tumor with that of patients with HCC has yielded conflicting results in a number of studies.2–5 Therefore, the demographic features and clinical behavior of these tumors remain ill-defined, which reflects their rarity.
Another likely reason for the conflicting results of many clinical series is the inconsistent use of a standardized pathologic classification for combined tumors. To our knowledge two histopathologic classification schemes for these tumors have been described to date(Table 1).1, 2 However, previous attempts to define the clinical behavior of these tumors are difficult to interpret because a wide spectrum of lesions have been included, many of which do not meet the definition of combined tumors. To our knowledge only Allen and Lisa type C and Goodman type II tumors have been reported to exhibit a mixture of the two lines of differentiation throughout and appear to be true combined tumors. However, to our knowledge many of the published series either include all other types or do not define the diagnostic criteria of the study population.4–8 In addition, the majority of reports are from Asia, and only the series of Goodman et al.2 analyzed patients with combined tumors from a Western cancer center. Fong et al.9 recently demonstrated that patients with HCC who were treated at a Western cancer center differed from their Asian counterparts in that underlying cirrhosis was found to be less common and its etiology more varied. It is likely that such differences also exist for patients with combined tumors.
|Allen and Lisa1||Description||Goodman et al.2||Description|
|Type A||HCC and CC are present at different sites within the same liver.||Type I||“Collision tumor” or an apparently coincidental occurrence of HCC and CC within the same liver.|
|Type B||HCC and CC are present at adjacent sites and mingle with continued growth.||Type II||“Transitional tumor” in which there is transition from elements of HCC to elements typical of CC.|
|Type C||HCC and CC are combined within the same tumor.||Type III||“Fibrolamellar tumor” which resembles the fibrolamellar subtype of HCC but which contains mucin-producing pseudoglands.|
The current study defines the demographic features of patients with combined HCC/CC and documents the results of resection at a single Western oncology center. To define more fully the clinical behavior of these lesions, patients with combined tumors were compared with those patients with “pure” peripheral CC and HCC. A rigorous histopathologic definition was used to ensure uniformity of the study population. The results demonstrate that the finding of combined histology has clinical implications.
MATERIALS AND METHODS
Review of a prospective database begun in 1985 of patients with primary liver tumors who were seen at Memorial Sloan-Kettering Cancer Center identified 27 patients with a diagnosis of combined HCC/CC between 1985 and December 1999. Tumor pathology was reexamined by two pathologists (D.K. and S.T.). The morphologic criteria proposed by Goodman et al.2 were used to classify the tumors. All tumors had histopathologic features intermediate between HCC and CC throughout or demonstrated a gradual transition from glandular CC areas toward moderate to well differentiated areas of HCC (Fig. 1). Tumors with features consistent with fibrolamellar HCC were not included. The current study therefore included only Allen and Lisa type C or Goodman type II tumors (Table 1). For each specimen, biliary differentiation was confirmed either by mucin positivity or immunoreactivity for cytokeratins characteristic of bile duct differentiation (CK7, CK19, and AE1); hepatocellular differentiation was confirmed by in situ hybridization for albumin mRNA (the specific methodology was described by D'Errico et al.10). The clinical, pathologic, surgical, and follow-up data for these patients were reviewed. Patient demographics, the pathology of the nonneoplastic liver, the incidence of cirrhosis and chronic hepatitis, serum tumor marker levels (α-fetoprotein [AFP] and carcinoembryonic antigen [CEA]), the surgical procedures performed, and survival were analyzed. The presence of positive serology for hepatitis B core antigen was interpreted as indicating chronic hepatitis B carrier status. Positive serology for hepatitis C was interpreted as indicating chronic hepatitis C infection. For patients seen and treated before the advent of routine testing for hepatitis C, none were found to have a diagnosis of chronic non-A, non-B hepatitis. The current status of all patients with combined tumors was determined by personal contact with the patient, the patient's family, or the attending physician. Patients with combined tumors were compared with patients with CC who underwent resection between March 1992 and June 2000. This group of patients with CC represents a subgroup of a larger series that recently was analyzed and published.11 In addition, comparisons also were made with 27 patients with HCC matched to the combined group with respect to age, tumor size, and location, and, in those patients who underwent resection, the type of partial hepatectomy performed.
Morphologically, all HCC cases had a predominantly trabecular or solid plate-like growth pattern and cells containing abundant eosinophilic cytoplasm and large vesicular nuclei with prominent nucleoli (Fig. 2). All peripheral CCs had prominent glandular formations lined by cuboidal cells (with or without mucin production), and typically demonstrated significant desmoplasia (Fig. 3); in situ hybridization for albumin mRNA was uniformly negative.
Numeric data are presented as the mean ± the standard deviation, unless otherwise indicated. Continuous variables were compared using the two-tailed Student t test and categoric variables were compared with a chi-square or Mann–Whitney U test. Survival probabilities were estimated using the Kaplan– Meier methodand compared using a log-rank test.12, 13
The clinical data are presented in Table 2. During the 15-year study period, 27 patients with combined tumors were evaluated.
|Age (yrs) (mean ± SEM)||61 ± 2||63 ± 2||62 ± 2|
|Age (yrs) (median [range])||61 (41–79)||65 (27–84)||64 (45–76)|
|Asian, n (%)b||2 (7%)||5 (19%)||2 (9%)|
|Cirrhosis (%)c||0||11 (41%)||1 (4%)|
|Hepatitis B or C (+)d||4 (15%)||15 (56%)||3 (13%)|
|AFP (ng/mL)||187 ± 6||1273 ± 28||12 ± 5 (n = 10)|
|CEA (ng/mL)||5 ± 5||3.8 ± 2||7 ± 3 (n = 11)|
The median age of the patients with combined tumors was 61 years (range, 41–79 years) and there were 14 men (52%) and 13 women (48%). Twenty patients were of European descent (81%) and 2 were Asian (7%). No patient had histopathologic evidence of cirrhosis and 4 patients (15%) had positive serology for either hepatitis B or hepatitis C (Table 2). Two patients had documented intrahepatic infection with Clonorchiasis sinensis and Schistosoma mansoni, respectively. In patients with CC, the median age was 64 years (range, 45–76 years) and there were 7 men and 16 women. Nineteen patients were of European descent (83%) and 2 were Asian (9%). One patient had histopathologic evidence of cirrhosis (4%), 1 patient had pathologically documented sclerosing cholangitis, and 3 patients (13%) had positive serology for either hepatitis B or C. By contrast, the median age of the patients with HCC was 65 years (range, 27–84 years), 67% were male, 19% were Asian, 41% had pathologic evidence of cirrhosis, and 56% had positive serology for hepatitis (Table 2). All patients with HCC and cirrhosis were Child Class A13 and Okuda Class I.14 These results for the HCC group are similar to those of a larger series of patients reported previously.9
Patients with combined tumors tended to have lower serum AFP levels and slightly higher CEA levels compared with patients with HCC, but these differences were not significantly different. In contrast, in the CC patients whose tumor markers were measured, the AFP levels were normal and the CEA levels were modestly elevated (Table 2).
The average size of resected combined tumors, CC, and HCC was 9.2 ± 4.1 cm (median, 8.8 cm; range, 2.9–16 cm), 7.8 ± 4.2 cm (median, 6 cm; range, 1.2–to 16 cm), and 13.2 ± 5.2 cm (median, 12.5 cm; range, 4–28 cm), respectively (P value was not significant [NS] for combined tumors vs. CC). Satellite tumors were present in 29% of the patients with combined tumors, 29% of the patients with CC, and 43% of the patients with HCC (P = NS). Similarly, the incidence of vascular invasion (7 patients each with combined tumors and CC and 12 patients with HCC) was not significantly different among the 3 groups. Hilar lymph node metastases were present in one patient with a combined tumor, four patients with CC, and two patients with HCC.
Twenty-one of 27 patients with combined tumors (78%) underwent resection with curative intent whereas 6 patients were found to have unresectable disease and were managed with supportive care alone. Of the 21 hepatic resections performed, there were 9 lobectomies and 12 trisegmentectomies (right or left hepatectomies).15 The hepatic resections performed in patients with CC and HCC were similar in magnitude to those performed in patients with combined tumors. The median blood loss for all resections was 870 mL (range, 25–4000 mL). The blood loss associated with the resection of combined tumors (median, 1150 mL; range, 450–3800 mL) was higher than that for CC (median, 700 mL; range, 100–3000 mL) (P = 0.024) and that for HCC (median, 800 mL; range, 25–4000 mL) (P = 0.055). There were three perioperative deaths reported overall (4%): two in the combined tumor group (one case of myocardial infarction and one case of hepatic failure) and one in the HCC group (hepatic failure).
After resection, the median survival of patients with combined tumors, CC, and HCC was 32 months, 37 months, and 46 months, respectively; these differences were not found to be significant. The actuarial 5-year survival was 24% for combined tumors, 33% for CC, and 37% for HCC (P = NS) (Fig. 4). Survival for patients in the HCC group was similar to that of a larger cohort of HCC patients previously reported from this center.9 There were 3 actual 5-year survivors after resection in the combined tumor group. The presence of satellite lesions (P = 0.0005) and vascular invasion (P = 0.007) (Table 3) were found to be significant predictors of poor outcome; none of the 6 patients with satellite tumors survived > 3 years after resection. The median survival in 4 patients with positive resection margins was 28 months versus 32 months in patients with negative resection margins (P = NS). Other variables that were not found to be associated with reduced survival after resection included tumor location, tumor size, number of tumors, perineural invasion, blood loss, surgical procedure, and surgical time.
|Characteristic||No.||1-yr survival (%)||3-yr survival (%)||5-yr survival (%)||P value|
|Mixed tumor (n = 21)|
The median disease-free survival was 16 months for patients with combined tumors, 14 months for patients with CC, and 18 months for patients with HCC; again, these differences were not found to be significantly different (Fig. 5). In those patients with combined tumors, the only factor that was predictive of disease recurrence was the presence of vascular invasion (P = 0.004), whereas perineural invasion, a positive resection margin, number of tumors, and tumor size were not found to be significant variables. The liver was the first site of recurrence in 11 of 15 patients with combined tumors (73%), 12 of 14 patients with CC (86%), and 15 of 19 patients with HCC (79%). In the current series, all patients with recurrent intrahepatic HCC were managed with either hepatic artery embolization or percutaneous alcohol injection. Of the patients with recurrent intrahepatic CC, two underwent ablative procedures (one alcohol injection and one radiofrequency ablation) and one underwent attempted reresection and was found to have extrahepatic disease.
Primary hepatic tumors comprised of elements of HCC and CC first were described in 1903 by Wells,16 but to our knowledge the first comprehensive description was not published until 1949.1 These tumors are encountered infrequently, and therefore meaningful clinical experience with them has been limited. To our knowledge the largest published clinical series to date is that of Maeda et al.,17 who described the pathologic and clinical features of 36 patients with combined tumors. However, the results of this and other studies are difficult to interpret because all histopathologic subtypes were included and analyses were not restricted to solitary tumors with mixed cellular features of both HCC and CC. Furthermore, the relatively recent availability of in situ hybridization techniques has allowed much more precise characterization of hepatocellular differentiation. The current study is an attempt to define combined tumors more fully and precisely by using strict histopathologic diagnostic criteria (Allen and Lisa type C and Goodman type II) (Table 1) and by comparing the demographic and clinical features of patients with these tumors with those of patients with pure CC and HCC managed over the same time period.
The reported frequency of combined tumors in series of primary hepatic malignancies varies widely, from a rate of 2.4% reported by Goodman et al.2 to a rate of 5.3% reported by Ng et al.3 to a high of 14.2% from the series of Allen and Lisa.1 During the 15 years in which patients in the current study were accumulated, a total of 742 patients with primary tumors of the liver were evaluated. This includes 27 patients with combined tumors (as defined earlier), resulting in an incidence of 3.6% in all patients with primary hepatic malignancies.
The results of the current study demonstrate that, in many respects, patients with combined tumors most closely resemble those with CC. Patient age, tumor size, and ethnic background were similar in both groups of patients, although CC was more common in women.4 However, most important, positive serology for hepatitis B and hepatitis C and histopathologic evidence of cirrhosis were much less common in patients with combined tumors and CC compared with patients with HCC. This finding has obvious implications for therapy, because underlying chronic liver disease is a common reason that HCC is not amenable to resection. In previous reports from the study institution, the overall resectability of patients with HCC and CC was 37%9 and 62%,11 respectively, compared with 78% for patients with combined tumors in the current study. Therefore, for those patients with combined tumors and CC, the extent of disease generally is the predominant determinant of resectability, in contrast to HCC in which coexisting hepatic dysfunction is an equally important variable and often dictates therapy.
The apparent lack of an association between chronic liver disease and combined tumors would suggest that the pathogenesis of these lesions differs from that of pure HCC, at least in the West. However, similar to HCC,9 there appear to be etiologic differences between Western and Asian patients with combined tumors. In Hong Kong, histologic evidence of cirrhosis or chronic hepatitis was present in 78% of patients with combined tumors and a large number of these patients were positive for the hepatitis B surface antigen.3 In Japan, chronic hepatitis C infection appears to be common in patients with combined tumors,18 and Taguchi et al.4 reported that nearly 40% of patients had evidence of cirrhosis and the majority had positive hepatitis serology.
Western and Asian patients also appear to differ with regard to tumor size at the time of diagnosis. In the current study, patients with combined tumors presented with relatively large lesions, as did patients with CC. The large tumor size is reflected in the surgical procedures required for resection, all of which were a lobectomy or extended lobectomy in the combined group, and also may explain the high incidence of vascular invasion and satellite lesions in both groups.19 By contrast, Asian patients with combined tumors generally have smaller lesions at the time of presentation, often ≤ 5 cm.4 This difference also has been observed in patients with HCC,9 and is nearly certainly related to the more extensive hepatoma screening programs in many Asian countries compared with Europe and North America.
The current study shows that resection of combined tumors can result in long-term survival in some patients. By contrast, patients with combined tumors not amenable to resection fared poorly, with no survivors reported beyond 18 months. Vascular invasion and the presence of satellite lesions, which are related and associated with tumor size, were significant predictors of poor outcome after resection of combined tumors. This observation also has been reported after resection of HCC and CC.4, 9 In contrast to the findings of the majority of studies analyzing outcome after hepatic resection for malignant disease, the current study failed to demonstrate a significant influence of microscopically involved resection margins, even though no patient with involved margins survived > 5 years. This finding nearly certainly reflects the small number of patients, and complete resection with clear surgical margins must remain the goal of surgery.
The overall median and 5-year survival rates were lower in patients with combined tumors and CC compared with patients with HCC, suggesting again that the clinical behavior of combined tumors more closely approximates that of CC. However, because of the small size of the study group, the differences did not reach statistical significance, and meaningful conclusions in this regard are not possible. It may be that the trend toward improved overall survival in HCC patients is related to the treatment rendered after disease recurrence. The initial site of recurrence in all patients with HCC was the liver, and all patients underwent additional treatment with hepatic artery embolization and/or alcohol injection. Conversely, patients with combined tumors or CC who had recurrence in the liver usually received systemic therapy, which is limited in its impact, or best supportive care. The similar disease-free survival reported among the three groups would support this observation. Peripheral CCs and many combined tumors are less vascular and much more fibrotic than HCC, and thus are less likely to respond to embolization or ethanol injection. However, other approaches, such as radiofrequency ablation or cryoablation, might be useful to treat recurrent disease in some of these patients, given the lack of effective chemotherapeutic alternatives and the high frequency of hepatic recurrence reported with both tumors.
The findings from the current series from a Western oncology referral center demonstrate that combined tumors share many demographic and clinical similarities with CC. The most important of these similarities is the absence of underlying chronic liver disease, a finding that is in marked contrast to patients with pure HCC. Complete resection is the only effective therapy and can result in long-term survival, but intrahepatic recurrence is common. More effective approaches for treating recurrent hepatic disease currently are available for these tumors and may improve outcome after resection.
- 151957.. Etudes anatomiques et chirurgicales. Paris: Mason;