Opposite Roles of Human Pancreatitis-Associated Protein and REG1A Expression in Hepatocellular Carcinoma: Association of Pancreatitis-Associated Protein Expression with Low-Stage Hepatocellular Carcinoma, β-Catenin Mutation, and Favorable Prognosis Free

Hepatocellular carcinoma (HCC) is one of the most frequent malignancies in south China, Taiwan, southeastern Asia, and sub-Saharan Africa. HCCs are genetically heterogeneous neoplasm and the genetic heterogeneity correlates with the variety of etiologic factors (1, 2). However, the molecular mechanisms remain largely unknown. Recent studies have unraveled many aberrant gene expressions, including activation of cellular oncogenes, inactivation of tumor suppressor genes, and overexpression of growth factors, in the neoplastic transformation of liver cells (3–5).

Regenerating gene (Reg), which is expressed in regenerating pancreatic islet, was first identified in the screening of regenerating islet-derived cDNA library taken from 90% depancreatized rat (6). Reg and Reg-related genes constitute a family within the superfamily of calcium-dependent lectin (C-type lectin; ref. 7). The C-type lectins are involved in several complex events, such as human malignancy and other diseases (8, 9). The Reg gene family consists of a group of acute phase reactants, lectins, antiapoptotic factors, or growth factors for pancreatic islet cells, neural cells, and epithelial cells in the digestive system (10, 11). Until now, 17 members of the Reg family have been identified and classified into four classes (Reg I-IV; ref. 12).

Reg I protein has many other synonyms, such as pancreatic thread protein, pancreatic stone protein, pancreatic stone protein, secretory, islet cells regeneration factor, etc. (13). There are two members of human REG I gene, REG1A and REG1B. The REGIA gene encodes a 166-amino-acid protein with a 22-amino-acid signal sequence. The REG1A protein is identical to the pancreatic thread protein, pancreatic stone protein, or protein X, and is highly represented in human pancreatic secretion (14). The REG1B gene codes for a transcript with 87% homology to the REG1A transcript, but the Reg1B protein has never been characterized and its expression in the pancreas remains questionable (15). The secretory Reg I protein is synthesized in the regenerating β cells (16), and through the Reg I receptor (17), stimulates the proliferation of pancreatic β cells, leading to an increase in the β cells mass in 90% depancreatized rats and nonobese diabetic mice and hence amelioration of experimental diabetes (18, 19).

Pancreatitis-associated protein (PAP), a member of the Reg III family, is also called HIP (hepatocarcinoma-intestine-pancreas), PAP1 (islet neogenesis associated protein INGAP), PAP-H (pancreatic β cell growth factor, human pancreatitis-associated protein), and REG-III. The human PAP cDNA encodes a 175-amino-acid protein with 49% identity with the human Reg protein (7). PAP protein is merely detectable in normal pancreas but remarkably increased representing up to 5% of secreted protein in acute pancreatitis and in some chronic pancreatitis (20).

In human, the REG1A, REG1B, RS (REG-related sequence), and PAP genes are clustered tandemly in a 95-kb region on chromosome 2p12. This gene cluster may have arisen from the same ancestral gene by gene duplication (21). The REG I mRNA was detected predominantly in the pancreas and at lower levels in gastric mucosa and kidneys but not in many other tissues (22, 23). PAP mRNA is expressed in the normal small intestine, pancreas, and pituitary gland but not in other tissues (7). REG I mRNA was detected in gastric cancer (24, 25), colorectal cancer (26), and cholangiocarcinoma (27). PAP mRNA can be detected in gastric cancer (28), colorectal cancer (26), pancreatic cancer (28, 29), and in about a quarter of primary human HCC (7, 30). A coordinate increase in pancreatic stone protein (Reg I) and PAP (Reg III) was found in experimental acute pancreatitis (31, 32), and in human colorectal cancer (26). In spite of these, the clinicopathologic role of REG1A and PAP expression and their interaction in HCC is not clear. In the present study, we show that PAP expression is associated with a subset of HCCs that is often low-grade, low-stage tumor and shows high frequency of β-catenin mutation, whereas a coexpression of REG1A leads to more advanced disease and poor prognosis.

Tissue samples. From January 1983 to December 1997, 1,033 surgically resected primary and 188 recurrent HCCs were pathologically assessed at the National Taiwan University Hospital. The tissues were immediately cut into small pieces, snap frozen in liquid nitrogen, and stored in deep freezers. 80% of these cases had been followed for >8 years or until death. Multifocal HCCs were excluded from this correlation because of incomplete sampling of the tumor nodules for genetic analysis and their variation in pathologic features. Of these, 265 patients who already had mRNA samples taken from resected primary HCC were selected for this study. The diagnosis of unifocal HCCs consisting of multiple nodules was made by morphology and verified by HBV integration patterns, α-fetoprotein mRNA expression, and mutation pattern of β-catenin and p53 genes, as described previously (33, 34). The 265 patients included 209 males and 56 females with a mean age of 55.6 years (range, 14-88 years). Serum hepatitis B surface antigen (HBsAg) was detected in 187 cases and anti-HCV antibody in 79 cases, including 26 cases positive for both. All of the patients had adequate liver function reserve at the time of surgery and all of the tumors were surgically resectable. None of these patients had received transhepatic arterial embolization or chemotherapy before surgery. To study the tissue distribution, we also examined the mRNA expression of PAP and REG1A in multiple adult tissues and multiple fetal tissues obtained from autopsy.

Histologic study and tumor staging. Tumor grade was divided into three groups, well differentiated (grade 1, 61 cases), moderately differentiated (grade 2, 102 cases), and poorly differentiated (grade 3 and 4, 102 cases). HCC tends to spread in the liver via the portal vein invasion even in advanced stage. At the time of operation, no evidence of regional lymph node or distant metastasis was noted, and minute HCC (≤2 cm) has excellent prognosis. HCC with complete fibrous encapsulation has a favorable four-survival, and vascular invasion, the most crucial step of intrahepatic tumor metastasis, is a crucial prognostic factor for HCC (34). Therefore, a modified tumor staging with special emphasis on the extent of vascular invasion, tumor size (≤2 or >2 cm), and encapsulation was adopted (35, 36). This modified staging correlated with survival (P < 0.0001) in 781 cases of unifocal surgical HCCs analyzed (37). Stage I to II HCCs had no vascular invasion, whereas stage III to IV HCCs had various extent of vascular invasion. Stage I HCC (six cases) included completely encapsulated minute HCC ≤2 cm with no liver invasion. Stage II HCC (115 cases) included minute HCC with liver invasion and/or microscopic satellite close to the main tumor; or larger HCC without or with liver invasion and/or minute satellite close to the main tumor. Stage IIIA HCC (43 cases) had invasion of thin-walled vessels in the tumor capsule but no portal vain invasion or satellite deep in the liver parenchyma. Stage IIIB HCC (33 cases) had invasion of small portal vein in portal tract near the main tumor but no invasion of major portal vein branch and satellite deep in the liver parenchyma. Stage IV HCC (68 cases) had invasion of major portal vein branches, satellites extending deeply into the surrounding liver, tumor rupture, or invasion of the adjacent organs.

The intrahepatic tumor recurrence was based on imaging diagnosis with ultrasonography and/or computed tomography, supplemented with elevated serum α-fetoprotein. Among the 265 patients studied, 236 (89.1%) were eligible for the evaluation of early intrahepatic tumor recurrence (≤1 year). Twenty-nine patients who died within 1 year after resection and had no information or were negative for intrahepatic tumor recurrence were excluded from the evaluation of early recurrence.

Reverse transcription-PCR. Reverse transcription-PCR was used to determine the expression of PAP and REG1A in 265 samples of HCCs and 219 corresponding nontumorous liver parenchyma, as was described elsewhere (38). S26 ribosomal protein mRNA, a housekeeping gene, was used as the internal control (39). Two micrograms RNA was transcribed to c-DNA in 20 μL reaction containing 0.5 μg random hexamer. Two microliters synthesis buffer (10×), 1 μL of 10 mmol/L deoxynucleotide triphosphate mix, 2 μL of 0.1 mol/LDTT, and 10 units Moloney murine leukemia virus reverse transcriptase (EPICENTRE, Madison, WI) at 37°C for 60 minutes. Two microliters reverse transcription product, 1.25 units Pro Taq polymerase (Protech Technology Enterprise, Taipei, Taiwan), 1× Pro Taq buffer, and 200 μmol/L (each) dATP, dCTP, dGTP, and dTTP were mixed with primer pairs for target gene and internal control gene in a total volume of 30 μL. PCR was done in an automatic DNA thermal cycler 480 (Perkin-Elmer Co., Wellesley, MA), with initial heating at 94°C for 2 minutes followed by 29 cycles of 94°C for 30 seconds, 60°C for 1 minute, 72°C for 1 minute, and final 72°C for 10 minutes for PAP. S26 primers were added at cycle 8. For REG1A, initial heating at 94°C for 2 minutes was followed by 28 cycles of 94°C for 30 seconds, 58°C for 1 minute, 72°C for 1 minute, and final 72°C for 10 minutes. S26 primers were added at cycle 7. The primers used are summarized in Table 1. PCR was stopped at the exponential phase of the amplified genes, 29 cycles for PAP, 28 cycles for REG1A, and 22 cycles for S26. After PCR, 5 μL of the reaction product were electrophoresed on a 2% agarose gel.

Analysis of p53 and β-catenin mutations. Mutations of the p53 tumor suppressor gene and β-catenin gene were detected by direct sequencing of exons 2 to 11 of p53 (40) and exon 3 of β-catenin (33). Cases with incomplete study were excluded from statistical analysis.

Follow-up observation. During the follow-up period up to 175 months, 213 patients (80.4%) had been followed for >10 years or until death. At the end of follow-up, 64 patients remained alive, 18 of who had survived for >10 years.

Statistical analysis. The analyses are carried out using the Statistica for the Windows software (Statsoft, Inc., Chicago, IL). Two-tailed χ² and Fisher exact tests were used for univariate analysis. The cumulative survival after tumor removal was calculated with log-rank test. Ps < 0.05 are considered statistically significant.

Expression of pancreatitis-associated protein and REG1A in hepatocellular carcinoma and liver. We used reverse transcription-PCR for large-scale analysis of PAP and REG1A mRNA levels in 265 unifocal primary HCCs. PAP and REG1A were overexpressed in 97 (36.6%) and 55 (20.8%) tumors, respectively. Among them, 46 tumors expressed both PAP and REG1A, with a high concordance rate (77.4%, P < 0.00001), whereas only nine HCCs expressed REG1A alone. Both genes were not detectable in 219 nontumorous liver tissues (Fig. 1).

Clinicopathologic correlation and relation to p53 and β-catenin mutations in hepatocellular carcinoma. As shown in Table 2, PAP overexpression in HCC showed a positive correlation with low α-fetoprotein level (≤200 ng/mL), and low-stage (stages I-II) HCCs that had no vascular invasion, P = 0.039 and P = 0.013, respectively, whereas REG1A overexpression did not. The expression of PAP or REG1A did not correlate with age, gender, chronic hepatitis infection, tumor size, tumor grade, or early tumor recurrence.

Mutations of p53 and β-catenin are the two major genetic alterations in HCCs. In this series, p53 mutation was found in 99 of 221 cases (44.8%), whereas β-catenin mutation was detected in 37 of 248 tumors (14.9%). We found that PAP and REG1A expression showed strong association with β-catenin mutation (P < 0.00001 and P = 0.00005, respectively) but not with p53 mutation (Table 3).

Opposite effects of pancreatitis-associated protein and REG1A expression in the tumor progression of hepatocellular carcinoma. Because of the frequent coexpression of PAP and REG1A, we did a combination analysis to further characterize the effects of PAP and REG1A expression in the tumor progression of HCC. We divided these cases into four groups according to presence or absence of PAP and REG1A overexpression: PAP(+)/REG1A(+), PAP(+)/REG1A(−), PAP(−)/REG1A(+), and PAP(−)/REG1A(−). Among the four groups, HCCs expressing PAP alone were associated with the highest frequencies of low-grade (grade 1) and low-stage tumors (P < 0.007 and P < 0.001, respectively) and hence the lowest early tumor recurrence (P = 0.051; Table 4). Consistent with results shown in Table 2, there was also no significant difference in age, gender, α-fetoprotein level, and chronic hepatitis infection between these four groups (data not shown).

In the two groups of HCCs with PAP expression, HCCs with coexpression of REG1A showed more frequent high-grade (84.8% versus 58.8%, P = 0.005), high-stage tumors (60.9% versus 29.4%, P = 0.002) and hence high-early tumor recurrence rate (61.8% versus 31.3%, P = 0.006) compared with HCCs showing PAP expression alone (Table 4).

To elucidate the reasons for the more aggressive tumors and early tumor recurrence in PAP(+)/REG1A(+) HCCs than in PAP(+)/REG1A(−) HCCs, we further analyzed the role of p53 and β-catenin mutations. As shown in Table 5, the former group showed significantly higher frequency of p53 mutation (46.2% versus 25.6%, P = 0.036), whereas the two groups did not differ significantly in β-catenin mutation.

Then, we further analyzed the role of p53 and β-catenin mutations in PAP(+)/REG1A(+) and PAP(+)/REG1A(−) HCCs. As shown in Table 6, HCCs coexpressed PAP and REG1A tended to have more frequent high-stage tumors but not of statistical significance, and the majority of HCCs with β-catenin mutation had low-stage tumors, regardless of the presence or absence of REG1A.

Correlation of pancreatitis-associated protein and REG1A expression with tumor progression in hepatocellular carcinomas without mutations of p53 and β-catenin. Despite the frequent p53 and β-catenin mutations, approximately half of HCCs are negative for both mutations. We then analyzed potential role of PAP and REG1A expression in this subset of HCC. In the subset of HCCs expressing PAP, the coexpression of REG1A had adverse effect in the groups of HCCs with PAP expression. As shown in Table 6, HCCs with coexpression of REG1A and PAP had >3-fold high-stage tumor than those with PAP expression alone (P < 0.005).

Prognostic significance of pancreatitis-associated protein and REG1A expression. To elucidate the roles of PAP and REG1A in HCC, we analyzed the survival rates of the four different groups of HCCs. According to the expression pattern of PAP and REG1A, HCCs with PAP expression alone had the best 5-year survival (P = 0.044) significantly better than HCC with coexpression of PAP and REG1A (P < 0.0002; Fig. 2).

PAP and regenerating protein 1 α (Reg1A) are reported to be expressed in various types of human cancer, such as gastric cancers (24, 25), colorectal cancers (22, 26), pancreatic cancers (29, 41), cholangiocarcinoma (27), and hepatocellular cancers (7, 30) but not in normal adult tissues, such as liver (7) and pancreas (29, 41). PAP can increase DNA synthesis and acts as a growth factor in vivo to enhance liver regeneration (42). PAP is induced by tumor necrosis factor α and cells overexpressing PAP show significantly less apoptosis on exposure to tumor necrosis factor α (43). Therefore, PAP seems to have dual mitogenic and antiapoptotic functions. REG1A is also shown to act as a factor to reduce epithelial apoptosis in inflammation (44). These observations indicate that PAP and REG1A play important roles in the tumorigenesis of various types of human cancer, including HCC. Their clinicopathologic significance in HCCs remains largely unknown because of the small number of cases analyzed. In this study of 265 surgically resected unifocal primary HCCs, PAP was expressed in 97 HCCs (36.6%) and REG1A in 55 (20.8%) but not in any of the 219 nontumorous livers. Furthermore, the interrelation between PAP and REG1A expressions and their significance in the tumor progression are not clear.

In contrast to the view that PAP overexpression is an indicator of tumor aggressiveness in pancreatic cancer (29, 41), we showed that PAP expression was associated with more frequent low-stage (stages I and II) HCCs that had no vascular invasion (P = 0.013). REG protein expression has been shown to correlate with frequent vascular invasion, lymph node metastases, and worse prognosis in gastric cancers (24, 25). Other investigators have shown that REG 1 protein is more frequently expressed in well-differentiated cholangiocarcinoma (27), and Reg gene is up-regulated during the differentiation of colorectal cancer cell lines (45). In this study, we failed to show a positive correlation of REG1A expression with tumor progression. The reasons for these discrepancies need to be further explained.

We showed that PAP expression closely correlated with REG1A expression (P < 0.00001), with a concordance of 77.4%, and 46 of 55 HCCs (83.6%) coexpressed REG1A and PAP. Coexpression of PAP and REG 1 has been shown in colorectal cancer (46). These data suggest a close interrelation between PAP and REG1A expressions. However, the significance is not fully understood. We then did a combination analysis of the four expression patterns of both genes in HCC. We showed that HCCs with PAP expression alone had the highest frequencies of well differentiated (grade 1) and low-stage tumors (P < 0.007 and P < 0.001, respectively), the least frequent early tumor recurrence, and the best 5-year survival compared with the other three groups (P = 0.051 and P = 0.044, respectively). These data suggest that PAP expression is associated with a subset of less aggressive HCC with favorable prognosis, whereas REG1A is associated with more advanced disease and may contribute to tumor progression of HCC. This suggestion is further supported by the finding that HCCs with coexpression of PAP and REG1A exhibited more frequent high-grade and high-stage tumors and more frequent early tumor recurrence than HCCs expression PAP alone (P = 0.005, P = 0.002, and P = 0.006, respectively). This result is consistent with the observation that Reg I alone or coexpression of Reg I with PAP have a significantly worse survival in early-stage colorectal cancer after curative surgery (46).

The p53 and β-catenin are two most frequently mutated genes in HCC (33, 40). HCCs have been classified into two major groups: one is characterized by chromosome stability and β-catenin mutation; the other is characterized by chromosomal instability and p53 mutation (47). Inactivation of p53 results in centrosome hyperamplification, leading to aberrant mitosis and chromosomal instability (48). We have shown that p53 mutation is associated with more aggressive HCC ( 49, 50), whereas β-catenin mutation is associated with low-grade, low-stage HCC, and better 5-year survival, and may possess tumor metastasis suppression activity (33). In this study, we showed that PAP and REG1A expression were closely associated with more frequent β-catenin mutation in HCCs (P < 0.00001 and P = 0.00005, respectively) but not with p53 mutation. Indeed, majority of HCCs expressing PAP alone, which had frequent β-catenin mutation, had low tumor stage (70.6%) and better prognosis (P = 0.044). However, HCCs coexpressing PAP and REG1A, which also had high frequent β-catenin mutation, had more frequent p53 mutation, high-stage tumors, and poor prognosis, suggesting that the REG1A expression abolishes the beneficial effects of PAP. Further analysis showed that the majority of HCCs with β-catenin mutation had low-stage tumors, regardless of the presence or absence of REG1A (13 of 17 versus 10 of 11, P > 0.05). This finding suggests that mutant β-catenin exerts stronger negative effect on tumor progression and can not be abolished by REG1A expression. We then exclude the interferences of p53 and β-catenin mutations and analyzed the roles of PAP and REG1A expression in HCCs without mutation of these two genes. In this subset of HCCs, the coexpression of PAP and REG1A was associated with more frequent high-stage tumors than PAP expression alone (P < 0.005). Although small in number of cases, these data further strengthen our observations that PAP and β-catenin mutations are associated with low-stage HCCs with favorable outcome, whereas REG1A possesses opposite effects that abolish the beneficial effects of PAP in this subset of HCC, resulting in more advanced disease. These data highlight the importance of combined analysis to better understand the significance of PAP and REG1A expressions in various disease processes, particularly the tumor progression. Thus, the REG1A expression might provide a valuable indicator for the identification of patients at increased risk of tumor recurrence after curative surgery for HCC and other cancers (46). The molecular mechanisms for the opposite roles of PAP and REG1A in tumor progression and metastasis need further studies for clarification.

In conclusion, these observations provide in vivo evidence that PAP and REG1A have divergent effects in the tumor progression of HCC. PAP expression plays a role in the development of subset of HCCs with less aggressive behavior and frequent association with β-catenin mutation, whereas coexpression of REG1A had a deleterious effect leading to more advanced disease and poor prognosis.

We thank Po-Lin Lai and Dr. Chiao-Ying Lin for their technical support.