Jae Yoon Seok is currently affiliated with the Department of Pathology, Kwandong University College of Medicine, Myongji Hospital, Goyang, Korea. These authors contributed equally.
Article first published online: 29 MAY 2012
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 55, Issue 6, pages 1776–1786, June 2012
How to Cite
Seok, J. Y., Na, D. C., Woo, H. G., Roncalli, M., Kwon, S. M., Yoo, J. E., Ahn, E. Y., Kim, G. I., Choi, J.-S., Kim, Y. B. and Park, Y. N. (2012), A fibrous stromal component in hepatocellular carcinoma reveals a cholangiocarcinoma-like gene expression trait and epithelial-mesenchymal transition. Hepatology, 55: 1776–1786. doi: 10.1002/hep.25570
Potential conflict of interest: The authors declare that there are no conflicts of interest.
This work was supported by a Korea Science and Engineering Foundation grant funded by the Korean government (MOST; 20100008075), a National Research Foundation of Korea grant funded by the Korean government (MEST; no. 2011-0030707), partly by grants (A101201 and A111574) of the Korea Healthcare Technology R&D Project, Ministry of Health and Welfare, Republic of Korea, and partly by Associazione Italiana Ricerca sul Cancro (AIRC) (grant no. 10247; to M.R.).
The raw data of the microarray experiments are available in the GEO database (http://www.ncbi.nlm.nih.gov/geo) with an accession number (GSE31370).
- Issue published online: 29 MAY 2012
- Article first published online: 29 MAY 2012
- Accepted manuscript online: 11 JAN 2012 04:49AM EST
- Manuscript Accepted: 7 DEC 2011
- Manuscript Received: 11 AUG 2011
- Korea Science and Engineering Foundation
- Korean government. Grant Number: MOST; 20100008075
- National Research Foundation of Korea
- Korean government. Grant Numbers: MEST; no. 2011-0030707, A101201, A111574
- Korea Healthcare Technology R&D Project, Ministry of Health and Welfare, Republic of Korea
- Associazione Italiana Ricerca sul Cancro (AIRC). Grant Number: 10247
Hepatocellular carcinoma (HCC) and cholangiocarcinoma (CC) are the major primary liver cancers in adults. The phenotypic overlap between HCC and CC has been shown to comprise a continuous liver cancer spectrum. As a proof of this concept, a recent study demonstrated a genomic subtype of HCC that expressed CC-like gene expression traits, such as CC-like HCC, which revealed the common genomic trait of stem-cell–like properties and aggressive clinical outcomes. Scirrhous HCC (S-HCC), a rare variant of HCC, is characterized by abundant fibrous stroma and has been known to express several liver stem/progenitor cell markers. This suggests that S-HCC may harbor common intermediate traits between HCC and CC, including stem-cell traits, which are similar to those of CC-like HCC. However, the molecular and pathological characteristics of S-HCC have not been fully evaluated. By performing gene-expression profiling and immunohistochemical evaluation, we compared the morphological and molecular features of S-HCC with those of CC and HCC. S-HCC expresses both CC-like and stem-cell–like genomic traits. In addition, we observed the expression of core epithelial-mesenchymal transition (EMT)-related genes, which may contribute to the aggressive behavior of S-HCC. Overexpression of transforming growth factor beta (TGF-β) signaling was also found, implying its regulatory role in the pathobiology of S-HCC. Conclusion: We suggest that the fibrous stromal component in HCC may contribute to the acquisition of CC-like gene-expression traits in HCC. The expression of stem-cell–like traits and TGF-β/EMT molecules may play a pivotal role in the aggressive phenotyping of S-HCC. (HEPATOLOGY 2012;55:1776–1786)
Hepatocellular carcinoma (HCC) and cholangiocarcinoma (CC) are the major primary liver cancers in adults. Most HCCs and CCs are derived from hepatocytes and cholangiocytes, respectively. Both hepatocytes and cholangiocytes originate from common liver stem cells with the potential to differentiate into both types of cells.1, 2 Thus, the primary liver cancers that arise from different developmental stages of liver stem cells are thought to harbor common genomic traits between HCC and CC. The existence of combined hepatocellular-cholangiocarcinoma (CHC) also supports the phenotypic overlap between HCC and CC.3, 4
In previous histological studies, a rare variant of HCC, characterized by abundant fibrous stroma between tumor nests, was reported in the absence of any preoperative treatment.5-8 These HCCs, namely scirrhous HCC (S-HCC), comprise up to 4.6% of the total cases of HCC.6 S-HCC has been known to express several liver stem/progenitor cell (SPC) markers, such as keratins (K) 7 and K19 and epithelial cell adhesion molecule (EpCAM).7, 9 This suggests that S-HCC may harbor intermediate traits between HCC and CC, including stem-cell traits. A recent study has demonstrated a novel subtype HCC that expresses CC-like traits, namely, CC-like HCC.10 This subtype, which has intermediate features between HCC and CC, showed distinct pathobiological characteristics enriched with stem-cell–like gene traits and therefore poorer clinical outcomes. However, the histopathological characteristics of these intermediate tumors have not yet been fully evaluated. Therefore, we hypothesized that the HCC phenotype with a fibrous stromal component (S-HCC) may share common genomic features with CC-like HCC, such as CC-like and/or stem-cell–like expression traits. To evaluate our hypothesis, we performed a gene-expression profiling analysis of S-HCC and compared the profiles of S-HCC with those of HCC (i.e., classical HCC without fibrous tissue) and CC.
In addition, it has been reported that scirrhous-type cancers, including gastric cancer, become invasive and metastatic by stimulating epithelial-mesenchymal transition (EMT).11 EMT is the differentiation switch from polarized epithelial cells to contractile and motile mesenchymal cells.12 EMT is thought to be a fundamental process that governs morphogenesis during embryogenesis and is reactivated in fibrogenic disease, provoking tumor progression.13 Snail and Twist, the key molecules of EMT, are associated with invasion and tumor metastasis and are also independent markers for worse prognosis in HCC.14, 15 A recent study also demonstrated that EMT induction by ectopic expression of either Snail or Twist transcription factors is able to generate cancer stem-cell properties in human breast cancer cells.16 These results suggest that EMT may play a critical role in the aggressive behavior and acquisition of stem-cell–like traits in S-HCC. With respect to these notions, we further evaluated whether EMT is involved in the HCC phenotype with fibrous stroma (S-HCC).
Materials and Methods
Tissue Specimens and Pathological Examination.
A total of 57 cases of primary liver carcinomas showing the following features were studied: (1) 14 patients with S-HCC, which had fibrous stroma in ≥50% of the tumor area by histological morphometry in the greatest dimension of cut surface9, 10, 17; (2) 24 patients with HCC, which had very little or no fibrous stroma; and (3) 19 patients with typical CCs. The tumor specimens were fixed in 10% buffered formalin, and representative sections were submitted for histological examination and immunohistochemistry (IHC). Fresh liver tissues were sampled from the same patients, snap-frozen in liquid nitrogen, and stored at −80°C. None of the case patients had preoperative treatment. For the pathological evaluation, histological grading of tumor differentiation was evaluated as follows: well, moderately, poorly, and undifferentiated. Tumor-capsule formation was categorized as “complete capsule” if more than 50% of the tumor circumference showed capsule formation, “partial capsule” for less than 50% capsule formation, and “none” if there was no capsule formation. Tumor invasion of the portal vein and microvessels was evaluated as “frequent” if found in more than five foci, “present” if one to five foci of vascular invasion were observed, and “absent” if no invasive foci were detected. Intracellular mucin formation was evaluated by mucicarmine stain.
The clinical features including follow-up data were obtained from hospital charts. The tumor node metastasis stage of each patient was evaluated according to the 7th American Joint Committee on Cancer staging system. This study was approved by the Institutional Review Board of Severance Hospital, Yonsei University College of Medicine (Seoul, Korea), and liver specimens were provided by the Liver Cancer Specimen Bank, National Research Resource Bank Program, Korea Science and Engineering Foundation of the Ministry of Science and Technology.
Total RNA Extraction and Gene-Expression Profiling.
Total RNA was extracted from tumor specimens using the Mirvana RNA isolation kit (Ambion, Inc., Austin, TX), according to the manufacturer's instructions. For complementary RNA (cRNA) production, 500 ng of the total RNA per sample was used employing the Illumina TotalPrep RNA amplification kit (Ambion). Integrity and quantity of the total RNA were assessed by the NanoDrop (Thermo Fisher Scientific, Wilmington, DE) and the Bioanalyzer (Agilent Technologies, Santa Clara, CA), respectively. cRNA was used for hybridization of a human HT12-v4 Illumina Beadchip gene-expression array (Illumina, San Diego, CA), according to the manufacturer's protocol. The hybridized arrays were scanned and fluorescence signals were obtained using the Illumina Bead Array Reader (Illumina). After quantile normalization of the raw data, the fold-difference values in each tumor against the average gene-expression values in five nontumoral surrounding tissues were used for further analysis.
Real-Time Quantitative Reverse-Transcription Polymerase Chain Reaction.
Primer sets for specific reverse transcription (RT) were used with the high-capacity RNA-to-cDNA Kit (Applied Biosystems Inc., Foster City, CA), according to the manufacturer's protocol. Assay IDs of the primers were as follows: K19 (Hs_00761767_s1), EpCAM (Hs_00901887_m1), cluster of differentiation (CD)133 (Hs_00195682_m1), Oct3/4 (Hs_00999632_g1), cMET (Hs_01565582_g1), transforming growth factor beta (TGF-β) 1 (Hs_00171257_m1), TGF-β receptor I (TGFβRI) (Hs_00610319_m1), TGF-β receptor II (TGFβRII) (Hs_00559661_m1), Smad4 (Hs_00232068_m1), Snail (Hs_00195591_m1), Twist (Hs_01675818_s1), and glyceraldehyde-3-phosphate dehydrogenase (Hs_99999905_m1). All reagents and instruments were purchased from Applied Biosystems. The reaction master mix, containing 2× RT Buffer, 20× Enzyme Mix, and nuclease-free water, was briefly mixed with 20 ng of each total RNA sample. Mixtures were incubated for 60 minutes at 37°C, 5 minutes at 95°C, and then kept at 4°C. Real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) was carried out using the Applied Biosystems 7500 Real-Time PCR System. The PCR master mix, containing TaqMan 2× Universal PCR Master Mix, 20× TaqMan assay, and RT products in a 20-μL reaction volume, was processed as follows: 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds, and then 60°C for 60 seconds. The signal was collected at the endpoint of every cycle. The mean values of the Ct, obtained in duplicate or triplicate, were used for data analysis.
Representative sections of formalin-fixed, paraffin-embedded tissues were used for IHC. Primary antibodies against K7, K19, EpCAM, CD56, alpha-fetoprotein (AFP), HepPar1, Smad4, and Snail were used (Supporting Table 1). We used the DAKO Envision Kit (Dako, Glostrup, Denmark) for IHC with a single primary antibody, then applied 3,3-diaminobenzidine (Dako). For double IHC staining, the EnVision AP system (Dako) and Vector Blue Alkaline Phosphatase Substrate Kit III (SK-5300; Vector Laboratories, Burlingame, CA) were used to detect the first primary antibody, then the EnVision DuoFLEX Doublestain System (SK110; Dako) and Vector NovaRED Substrate Kit (SK-4800; Vector Laboratories) were used to detect the second primary antibody. The expression of each marker was evaluated as positive when it was detected in more than 5% of tumor cells and was scored as follows: 1+ for detection in 5%-10% of tumor cells, 2+ for 11%-50%, and 3+ if detected in over 50% of tumor cells.
Statistical analysis was performed using the SAS software (version 9.1.3; SAS Institute Inc., Cary, NC) and R package (http://www.r-project.org). We assessed the IHC stain results using the chi-square test, and the Student's t test was used to compare the results of the real-time qRT-PCR. The bivariate correlation test was used to analyze correlations among the qRT-PCR results. Survival analysis was carried out using Kaplan-Meier's method, and differences were analyzed using the log-rank test.
Clinicopathological Features in S-HCCs, HCCs, and CCs.
In histological evaluation, S-HCCs showed abundant fibrous stroma between trabeculae or solid nests of tumor cells, and CCs also showed marked fibrous stroma between tumor glands, whereas HCCs showed trabecular or adenoid patterns with no or little fibrous stroma (Fig. 1). The centers of the S-HCC nests were composed of polygonal cells with abundant cytoplasm resembling mature hepatocytes, whereas the periphery of the tumor nests facing the fibrous stroma was composed of small, oval-shaped tumor cells with a high nuclear cytoplasmic ratio (Fig. 1B). These oval-shaped tumor cells mimic hepatobiliary cells with ductular reactions, which are thought to be indicative of liver SPCs. This pattern was focally found in rare cases of HCCs and not in any case of CCs. None of the S-HCC or HCC samples showed intracellular mucin, whereas all the CC samples showed mucin formation.
Clinicopathological features of S-HCCs, HCCs, and CCs were compared (Table 1). Most S-HCCs (11 of 14; 79%) formed no tumor capsule, as was the case for the majority of CCs (18 of 19; 95%), whereas most HCCs (23 of 24; 96%) showed a partial or complete tumor capsule. S-HCCs showed significantly less tumor-capsule formation than HCCs (P < 0.001). S-HCCs showed more frequent invasion of microvessels than HCCs (P = 0.025). Tumor stage was higher in S-HCCs than in HCCs (P = 0.023) at diagnosis, although there was no significant difference in tumor size between the two groups (P = 0.244). Lymph-node metastasis and portal-vein invasion were more frequent in CCs than in S-HCCs (P < 0.05). Etiologies of S-HCCs were hepatitis B in 11 patients, alcohol in 1 patient, and unknown in 2 patients; those of HCCs were hepatitis B in 18 patients, hepatitis C in 4 patients, and unknown in 2 patients. For CCs, 7 patients had hepatolithiasis and 12 patients showed no specific etiology.
|Clinicopathological Features||S-HCC (n = 14)||HCC (n = 24)||CC (n = 19)||S-HCC Versus HCC (P Value)||S-HCC Versus CC (P Value)|
|Age (years)||53 ± 12 (27∼67)||58 ± 10 (40∼81)||61 ± 9 (41∼81)||0.066||0.012‡|
|Tumor size (cm)||3 ± 1.8 (1∼7)||4 ± 1.7 (2∼9)||7 ± 2.3 (2∼12)||0.244||<0.001‡|
|Differentiation (%)||Well||0 (0)||0 (0)||1 (5)||0.393||0.119|
|Moderate||14 (100)||22 (92)||14 (74)|
|Poor||0 (0)||2 (8)||4 (21)|
|Capsule formation* (%)||Complete||0 (0)||10 (42)||0 (0)||<0.001‡||0.193|
|Partial||3 (21)||13 (54)||1 (5)|
|None||11 (79)||1 (4)||18 (95)|
|Portal vein invasion† (%)||Frequent||0 (0)||0 (0)||4 (21)||0.050||0.016‡|
|Present||5 (36)||2 (8)||11 (58)|
|Absent||9 (64)||22 (92)||4 (21)|
|Microvessel invasion† (%)||Frequent||4 (29)||0 (0)||13 (68)||0.025‡||0.052|
|Present||4 (29)||7 (29)||4 (21)|
|Absent||6 (42)||17 (71)||2 (11)|
|Satellite nodule (%)||Present||1 (7)||0 (0)||5 (26)||0.368||0.171|
|Absent||13 (93)||24 (100)||14 (74)|
|Lymph node metastasis (%)||Present||1 (7)||1 (4)||8 (42)||0.607||0.030‡|
|Absent||13 (93)||23 (96)||11 (58)|
Comparison of Gene-Expression Profiles and Survival Analysis of S-HCCs, HCCs, and CCs.
Next, to address the heterogeneous genomic features of HCC and CC, we performed gene-expression profiling on the subset (21 of 57 cases) of liver cancers, including 9 S-HCC, 6 HCC, and 6 CC. Gene-expression profiling was also performed on 5 cases of nontumoral surrounding tissues to normalize the profiles of tumor tissues. We first identified 293 differentially expressed gene features between S-HCC and HCC, with the cutoff of more than 2-fold difference and P < 0.01 (Student's t test). Gene-ontology analysis with these genes was performed by using the DAVID bioinformatics resource (http://david.abcc.ncifcrf.gov), which showed significant up-regulation of cell adhesion, development, migration, and proliferation-related gene functions in S-HCC, suggesting the aggressive phenotype of S-HCC, compared to that of HCC (Supporting Table 2).
Comparison of gene-expression profiles among the three groups of S-HCC, HCC, and CC was performed by analysis of variance (P < 0.001), which yielded a total of 612 differentially expressed gene features. Interestingly, most of the gene features showed intermediate expression levels between HCC and CC, and no significant expression patterns specific to S-HCC were found. This finding may be indicative of the intermediate phenotype of S-HCC between HCC and CC (Fig. 2A). This also suggests that S-HCC harbors a CC-like gene-expression trait (i.e., CC signature), which has been previously identified as representing a subtype of CC-like HCC.5 Therefore, we examined the expression of CC signature in S-HCC using the gene set enrichment analysis (GSEA) method.18 This showed significant enrichment of both CC_UP and CC_DOWN signatures on S-HCC, compared to those on HCC (Fig. 2B,C). This suggests that the expression of a CC-like trait in HCC might be attributable, at least in part, to the existence of intratumoral fibrous stroma in HCC (i.e., S-HCC).
Next, we compared the disease-free survival (DFS) of those three tumor types. Follow-up data were available in all cases, with an average duration of 29 ± 24 months (mean ± standard deviation). Consistent with a previous report, which showed poorer clinical outcomes for CC-like HCC,5 the DFS rate was the highest in HCCs and the lowest in CCs, with S-HCCs falling in between (P < 0.001) (Fig. 2D). A comparison of DFS between S-HCCs and HCCs showed that the difference was not significant, which may have been a result of the small sample size. However, DFS was significantly worse in S-HCCs than in HCCs when large tumors (≥5 cm) were analyzed separately (P = 0.038; Fig. 2E). A recent study has shown the aggressive pathological features of S-HCC, which has less tumor-capsule formation and hypervascularity than conventional HCC, supporting our findings.17
Expression of Liver SPC Markers in S-HCCs, HCCs, and CCs.
CC-like HCCs have shown the expression of stem-cell–like traits, implying their cellular origin from liver SPCs.5 Similarly, S-HCCs have also been reported to express several SPC markers, such as K7, K19, and EpCAM.8, 10 With respect to this finding, we further evaluated the expression of stem-cell–like traits in S-HCC. The expression of differential markers, including EpCAM, K7, K19, CD56, AFP, and HepPar1, was evaluated using IHC stain. Strikingly, most cases of S-HCCs (13 of 14; 93%) were immunostained by at least one of the K19, EpCAM, and CD56 markers (Fig. 3A-D). In addition, the topographical expression patterns of K19/EpCAM (liver SPC markers) and HepPar1 (a hepatocyte marker) were evaluated by double IHC stain. In most S-HCCs (8 of 11 double-positive cases; 73%), K19 and/or EpCAM protein was expressed in the small peripheral tumor cells adjacent to the fibrous stroma, whereas HepPar1 protein was expressed mainly in the eosinophilic polygonal tumor cells with ample cytoplasm at the center of the tumor-cell nests (Fig. 3D). These findings may indicate that the expression of CC-like and stem-cell–like traits is closely related to the fibrous stromal component in S-HCC.
The differential expression of SPC markers was further confirmed at both the mRNA and protein levels. Messenger RNA (mRNA) levels of EpCAM, CD133, K19, Oct3/4, and cMET were significantly higher in S-HCCs than in HCCs (P < 0.05), whereas those of CD133 and K19 were lower in S-HCCs than in CCs (P < 0.05) (Fig. 3E-I; Supporting Table 3). The expression of those SPC markers correlated well with one another, indicating the modular coexpression of SPC markers (Fig. 3J; Supporting Fig. 1A-E). Similarly, protein expression levels of EpCAM, K19, K7, CD56, and AFP were more prevalent in S-HCCs than in HCCs, whereas HepPar1 was less prevalent in S-HCCs (Fig. 3K-P). Moreover, frequent coexpression of EpCAM and K19 proteins was observed in S-HCCs (8 of 14; 57%), but not in HCCs (1 of 24; 4%), indicating the distinct expression of SPC markers in S-HCC, compared to that in HCC (Supporting Fig. 1F,G).
Expression of EMT Core Network in S-HCCs.
As described above, it has been reported that the scirrhous type of cancers undergo EMT.11 To evaluate whether or not the EMT-related genes are expressed in S-HCCs, we applied the core EMT interactome genes that have been previously identified elsewhere.19 When we performed the GSEA, significant enrichment of the EMT core gene signature (EMT_CORE_UP; n = 91) was observed in S-HCCs, compared to HCCs (Fig. 4A). This finding was further validated by examining the expression of individual EMT-related genes by the real-time PCR method. The expression of the key EMT molecules, Snail and Twist, was significantly higher in S-HCCs than in HCCs (P < 0.05) (Fig. 4B,C; Supporting Table 3). Protein expression of Snail was significantly more prevalent in S-HCCs (10 of 14; 71%) and CCs (11 of 19; 58%) than in HCCs (6 of 24; 25%) (P = 0.006) (Fig. 4D). In immunohistological evaluation, the expression of Snail protein was mainly localized to the periphery of the tumor nests, next to the fibrous stroma (Fig. 4E; Supporting Fig. 2). In addition, Snail and K19/EpCAM expression was colocalized in most S-HCCs (8 of 10 double-positive cases; 80%), which were mainly located on the periphery of the tumor nests, next to the fibrous stroma cells (Fig. 4F). These findings support that the expression of EMT genes are not merely the attributes of fibrotic components, but the results of the cross-talk interaction between epithelial and fibrotic components within the tumors. We also observed that mRNA levels of Snail and Twist were well correlated with those of SPC markers, including K19, EpCAM, cMET, CD133, and Oct3/4, suggesting the distinctive association of EMT with the stem-cell-like features of S-HCC (Supporting Fig. 3).
TGF-β Signaling Is a Potential Key Regulator for EMT Induction in S-HCCs.
TGF-β binds two types of receptors, TGFβRI and TGFβRII, to form active signaling complexes. The phosphorylated TGFβRI transmits signals intracellularly by phosphorylating the transcription factors, Smad2 and Smad3, which then forms a complex with Smad4. The Smad complex moves into the nucleus, where it regulates the expression of target genes.20 TGF-β is known to promote tumor-cell invasion and metastasis as a potent stimulator of EMT.21-23 With this concern, we next evaluated whether the TGF-β signaling is differentially expressed among tumor subtypes according to the expression of EMT-related molecules. We observed that the expression of TGF-β-signaling molecules, including TGF-β, TGFβRI, and Smad4, was significantly higher in S-HCCs than in HCCs or CCs (P < 0.05) (Fig. 5A-D; Supporting Table 3). Smad4 protein was expressed in most S-HCCs (12 of 14; 86%), but not significantly more than other types (Fig. 5E). Particularly, the Smad4 protein in S-HCCs (6 of 12; 50%) was mainly expressed in small, oval-like, and peripherally located tumor cells facing the fibrous stroma and was also focally expressed in some stromal cells (Fig. 5F; Supporting Fig. 2). In addition, the expression of TGF-β-signaling molecules correlated well with that of liver SPC markers, including K19, EpCAM, and cMET (Supporting Fig. 4). Snail and Twist were also well correlated with TGF-β-signaling molecules at mRNA levels (Supporting Fig. 5). These results clearly reinforce the idea that TGF-β signaling may play a pivotal role in the induction of EMT in S-HCCs.
It is now generally accepted that primary liver cancer forms a continuous spectrum from HCC and CC, mimicking each other's morphological and phenotypic properties.24 In the present study, we have shown that a variant of HCCs with scirrhous components (i.e., S-HCC) has an intermediate phenotype, expressing both CC-like and stem-cell-like traits. These results suggest the acquisition of the CC-like trait in HCCs might be related to, at least in part, the existence of the fibrous stroma in tumors. The cross-talk between the fibrous-stroma and the tumor-cell components may contribute to poor prognostic outcome. However, there are several conflicting studies that have shown the overall survival of S-HCCs to be better than7, 8, 25 or similar to10, 17 that of HCCs. This might be a result of the limited sample sizes of the studies, revealing the need for further large-scale evaluations. In contrast, we observed, in this study, that DFS of the patients is worse in S-HCCs and CCs than in HCCs. This result was obtained by applying a stringent criterion for S-HCCs that the fibrous stroma is more than 50% of tumor area without any preoperative treatment. Indeed, S-HCCs showed more aggressive phenotype of frequent invasion of microvessels and less frequent tumor-capsule formation than HCCs (Table 1). The enrichment of tumor aggressiveness-related gene functions in S-HCCs also supports the clinical characteristics (Supporting Table 2). More likely, our finding is consistent with the previous findings of the poorer clinical outcome in the intermediate phenotype tumors, including CC-like HCC or CHC.
It has been well established that HCCs with stem-cell-like traits (e.g., EpCAM, CD133, and K19) have poorer prognoses and higher recurrence rates than those without.26-31 We observed that S-HCCs express liver SPC markers (e.g., K7, K19, EpCAM, CD56, CD133, Oct3/4, and cMET), which was in agreement with the previous results. Interestingly, the liver SPC markers in S-HCCs were mainly detected in the small and oval-like tumor cells, which are peripherally located at the tumor nests and are considered to be associated with “stemness.” This finding also supported the idea that cancer stem-like cells might be more involved in S-HCCs than in HCCs.
Tumor-stroma cross-talk in HCC was recently highlighted by the finding that stromal myofibroblasts provide TGF-β and induce a characteristic EMT at the tumor border.32 TGF-β was reported to regulate CD133 expression by inhibiting the expression of DNA methyltransferase 1 and DNA methyltransferase 3b and subsequent demethylation of promoter-1. Also, the TGF-β1-induced CD133+ Huh7 cells were reported to be tumorigenic.33 In accord with these findings, we observed that TGF-β-signaling molecules as well as EMT-related genes were significantly up-regulated in S-HCCs. Topographically, S-HCC showed colocalized expression of Snail and K19/EpCAM molecules, particularly in the small and oval-like tumor cells, which may support the idea that TGF-β signaling and EMT play a critical role in the acquisition of stem-cell-like traits. Similarly, it has been reported that S-HCCs express TGF-β1 at the periphery of tumor nests, next to the fibrous stroma as well as myofibroblasts, and also coexpress CD56 and K7.8 Taken together, we suggest that the EMT might be involved in the acquisition of stem-like and CC-like genomic features in S-HCC through the up-regulation of TGF-β signaling, which may contribute to the presence of an invasive pathological property (illustrated in Fig. 6A).
S-HCC is different from CHC, although small and oval-like tumor cells and abundant fibrous stroma are found in both tumors.4 Unlike CHC, the classical type, S-HCC, does not show mucin or CC components. In addition, S-HCC is composed mainly of hepatocyte-like tumor cells, whereas CHC with stem-cell features has a main component of small oval-like tumor cells that have stem-cell-like features.4, 34-37 More important, similar fibrotic stroma can occur after chemotherapy, radiation, or transarterial chemoembolization. Such cases should not be confused with S-HCCs. Therefore, in this study, we included all the cases that had no preoperative treatment.
In summary, the overall features of our findings can be viewed from the perspective of a spectrum of primary liver cancer composed of HCC, CC, and CHC in the middle (Fig. 6B). Variant HCCs, including S-HCC, are positioned next to CHC because they harbor more HCC-like features than CHC. The molecular characteristics of S-HCC are highlighted by TGF-β signaling and EMT. Our results may provide new pathobiological insights regarding the scirrhous phenotype of HCC and its contribution to the primary liver cancer spectrum.
- 4Combined hepatocellular-cholangiocarcinoma. In: Bosman FT Carneiro F, Hruban RH, Theise ND, eds. WHO classification of Tumours of the Digestive System, 4th ed. Lyon, France: IARC; 2010: 225-227., , , .
- 5Tumors of the Liver and Intrahepatic Bile Ducts. Washington, DC: Armed Forces Institute of Pathology; 2001: 356., , .
- 24Overview of cholangiocarcinoma and evidence for a primary liver carcinoma spectrum. In: Wang XW, Grisham JW, Thorgeirsson SS, eds. Molecular Genetics of Liver Neoplasia. New York, NY: Springer; 2011: 19-33., , .
- 37Histopathological classification of hepatocellular carcinoma. Digest Liver Dis 2010; 42: S228-S234., , .
Additional Supporting Information may be found in the online version of this article.
|HEP_25570_sm_SuppFig1.tif||5653K||Supplementary Figure 1. A-E) Correlation among log transformed mRNA levels of liver stem/progenitor cell markers [epithelial cell adhesion molecule (EpCAM), keratin (K) 19, CD133 and Oct3/4]. Venn diagram depicting protein expression levels of EpCAM, K19 and CD56 in scirrhous hepatocellular carcinomas (HCC) (F) and HCCs (G). K19-positive and/or CD56-positive HCCs are also immunoreactive for EpCAM.|
|HEP_25570_sm_SuppFig2.tif||4651K||Supplementary Figure 2. The expression of Snail and Smad4 protein in S-HCCs. A) Snail protein is mainly found in the peripheral cells of tumor nests and it is also focally expressed in some stromal cells in S-HCCs. B) Smad4 protein is mainly found in the tumor cells and focally in some stromal cells in S-HCCs.|
|HEP_25570_sm_SuppFig3.tif||5062K||Supplementary Figure 3. Correlation of log transformed mRNA levels of liver stem/progenitor cell markers [epithelial cell adhesion molecule (EpCAM), keratin (K) 19, cMET, CD133, Oct3/4] with those of Snail and Twist.|
|HEP_25570_sm_SuppFig4.tif||4580K||Supplementary Figure 4. Correlation of log transformed mRNA levels of liver stem/progenitor cell markers [keratin (K) 19, epithelial cell adhesion molecule (EpCAM), Oct3/4 and cMET] with those of transforming growth factor β (TGF-β), TGF-β receptor I (TGFβRI) and Smad4.|
|HEP_25570_sm_SuppFig5.tif||4664K||Supplementary Figure 5. Correlation of log transformed mRNA levels of transforming growth factor β (TGF-β), TGF-β receptor I (TGFβRI), TGF-β receptor II (TGFβRII) and Smad4 with those of Snail and Twist.|
|HEP_25570_sm_SuppTab1.doc||76K||Supplementary Table 1. Antibodies.|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.