Alterations of p16 and Prognosis in Biliary Tract Cancers from a Population-Based Study in China

Biliary tract carcinomas are relatively uncommon in most parts of the world, although incidence is elevated in certain geographic locations and populations groups (1). From 1972 to 1994, biliary tract cancer was the most rapidly rising malignancy in Shanghai, the People’s Republic of China, with a 119% increase in men and 124% increase in women (2). Reasons for the rising incidence of biliary tract cancers in Shanghai are unclear, but nutritional changes and improvements in diagnosis and classification may contribute to this trend.

The p16 (INK4, MTS1, and CDKN2) tumor suppressor gene, which maps to chromosome 9p21-p22, is frequently inactivated in a wide variety of tumors by deletion, mutation, or CpG island methylation (3, 4, 5, 6, 7). p14 (ARF) and p15 genes are present in the same region of chromosome 9p as the p16 gene (8). p16 and p14 transcripts are produced from two separate promoters and use alternative first exons (1α and 1β, respectively) joined through the same splice acceptor site to exon 2 coding sequences but in a different reading frames (9, 10). p16 is a cyclin-dependent kinase (CDK) inhibitor involved in regulation of the G₁ checkpoint in the cell cycle (11). The p16 protein blocks the G₁-S transition in the cells by binding to and preventing association of CDK4 and CDK6 with cyclin D1, leading to dephosphorylation of the retinoblastoma gene product and repression of transcription factor E2F (11, 12). The CDK4/cyclin D complex phosphorylates the product of the retinoblastoma gene, thereby promoting entry into the cell cycle. The p16 gene is inactivated by homozygous or heterozygous deletion, mutations of coding and adjoining noncoding regions, or methylation of CpG island present at the 5′-end of the gene in a wide variety of primary cancers and tumor cell lines (5, 6, 7, 13, 14, 15, 16, 17). Loss of expression of p16 because of methylation or deletion or loss of functional p16 protein because of mutations result in unregulated CDK4 activity, leading to persistent retinoblastoma phosphorylation and therefore uncontrolled cellular proliferation.

p16 gene mutations (18, 19), loss of heterozygosity of chromosome 9p (19, 20, 21, 22) and loss of p16 protein expression (23) are reported in primary biliary tract cancers and cell lines derived from these cancers. To clarify the role of p16 inactivation in biliary tract carcinomas, we studied p16 gene alterations in a population-based study of these tumors by sequencing, methylation of CpG island present at the 5′ of the p16 gene, and loss of heterozygosity (LOH) of chromosome 9p. We also evaluated p16 gene alteration status as a prognostic marker for biliary tract cancers.

Patients with primary biliary tract cancer (ICD9 156) newly diagnosed between 1997 and 1999 were identified through a rapid reporting system established between the Shanghai Cancer Institute and 30 collaborating hospitals in urban Shanghai. This reporting system recruited over 95% of patients with biliary tract cancers in Shanghai. A total of 118 patients with biliary tract carcinomas, including 68 gallbladder cancers, 33 bile duct cancers, and 17 ampullary cancers, were identified for analysis as part of a population-based case-control study. Eligibility criteria for recruitment of patients included the following: residents of urban Shanghai between 18 and 74 years of age diagnosed after April 1997 with gallbladder, extrahepatic bile duct, or ampullary carcinomas.

Surgical pathology specimens were collected from patients with biliary tract cancers undergoing curative resection by pancreaticoduodenectomy or bile duct resection or biopsy of an advanced tumor. As part of the case-control study, 6 H&E-stained slides and 6 unstained slides (5 μ each) and 10 ml of peripheral blood were routinely collected for each patient from the surgical pathology and surgery departments of the participating hospitals. In addition, a structured questionnaire was used to elicit information on demographic, clinical, and epidemiological variables. The anatomical location where the tumor was taken was recorded in a diagram completed by the local pathologist at the participating hospital. The histopathological slides were reviewed by two pathologists from Shanghai and were independently reviewed by one of us (A. R.). Classification was based on the WHO classification of tumors of the biliary tract (24). Duodenal and pancreatic carcinomas were excluded from the study. Medical records were abstracted for all cancer cases. Patients’ follow-up status was obtained by the Shanghai Cancer Institute by contacting the patients or their relatives from the date of diagnosis to March 2001.

Genomic DNA was extracted from tumor tissue by microdissection from three H&E-stained slides without a coverslip from formalin-fixed, paraffin-embedded blocks and prepared as described in previous studies (25). DNA from tumor tissue represented at least 50% tumor DNA. DNA from peripheral blood was used if available.

A RNase cleavage assay was used to screen for mutations of p16 gene according to the manufacturer’s instructions (Nonisotopic RNase cleavage assay; Ambion, Austin, TX). Exons 1 and 2 of the p16 gene were amplified. Exon 1 of the p16 gene was amplified using the forward and reverse amplification primers 5′-GAAGAAAGAGGAGGGCTG-3′ and 5′-GCGCTACCTGATTTCAATTC-3′, respectively. The cycling conditions were as follows: 95°C for 10 min; 37 cycles of 95°C for 30 s; and 55°C for 45 s. Exon 2 of the p16 gene was amplified using the forward and reverse amplification primers 5′-ACACAAGCTTCCTTTCCGTC-3′ and 5′-TCTGAGCTTTGGAAGCTCTC-3′, respectively. The cycling conditions were as follows: 95°C for 10 min; 38 cycles of 95°C for 30 s; and 60°C for 45 s. The T7 promoter sequence was introduced at both ends of the PCR product by nested PCR using 2 μl of the PCR product and forward and reverse T7 primers containing the 20 bp of T7 promoter sequence at the 5′-end of each primer. The forward and reverse primers for nested PCR were 5′-TAATACGACTCACTATAGGGAGGGGCTGGCTGGTC-3′ and 5′-TAATACGACTCACTATAGGGTCCCCTGCAAACTTCGTC-3′ for exon 1, and 5′-TAATACGACTCACTATAGGGCTTCCTTTCCGTCATGCCG-3′ and 5′-TAATACGACTCACTATAGGGCTTTGGAAGCTCTCAGGGG-3′, respectively. The cycling conditions were as follows: 95°C for 10 min; 40 cycles of 95°C for 30 s; and 60°C for 45 s. The RNA product was made by in vitro transcription from the nested PCR product using T7 RNA polymerase. RNA was denatured at 95°C and hybridized at room temperature. The duplex RNA products were cleaved by RNase I and RNase T1 if mismatch nucleotide(s) were present in the duplex RNA. The cleaved product was directly loaded on 3% agarose gel and visualized by ethidium bromide. Samples from lanes with the shorter bands indicating a mismatch in nucleotide sequence or absence of bands indicating a technical failure or multiple mismatched nucleotide sequences were selected for DNA sequencing. No bands or aberrant bands were present in the RNase cleavage product of 18 carcinomas for exon 1 and 53 carcinomas for exon 2. All carcinomas with no bands or aberrant bands were sequenced. Forty carcinomas without any aberrant bands were sequenced, and no mutations were found. PCR amplification of exon 1, exon 2, or both exons failed for 26 carcinomas.

DNA sequencing was performed using the PCR product and amplification primers using a commercial DNA sequencing kit according to the manufacturer’s instructions (Sequitherm Excel; Epicenter Technologies, Madison, WI). The forward and reverse primers for sequencing were 5′-GCTGGCTGGTCACCAGAG-3′ and 5′-ACTTCGTCCTCCAGAGTC-3′ for exon 1, and 5′-CTTCCTTTCCGTCATGCGG-3′ and 5′-AGCTTTGGAAGCTCTCAGG-3′, respectively. Mutations were confirmed by reamplification of tumor DNA and sequencing in both directions. Germ-line mutations were excluded by sequencing DNA extracted from blood cells if available and by comparison with the previous reports of germ-line mutations (16, 17).

The methylation status of p16 was determined by bisulfite treatment of DNA followed by methylation-specific PCR as described, with modifications (26). In brief, 1 μg of microdissected genomic DNA was denatured with 2 m NaOH at 37°C for 10 min, followed by incubation with 3 m sodium bisulfite (pH 5.0) at 50°C for 16 h in the dark. After treatment, DNA was purified using the DNA Cleanup Kit (Promega) as recommended by the manufacturer, incubated with 3 m NaOH at room temperature for 5 min, precipitated with 10 m ammonium acetate and 100% ethanol, washed with 70% ethanol, and finally resuspended in 20 μl of distilled water. The primers and PCR conditions for p16 were the same as those reported by Herman et al. (26). In brief, 2 μl of bisulfite-treated DNA were used as template for PCR reactions using primers specific for methylated and unmethylated alleles. PCR products from methylated and unmethylated reactions were electrophoresed on 10% acrylamide gels and visualized by ethidium bromide staining (examples in Fig. 1 A). Representative samples were reanalyzed to confirm the methylation status. DNA for bisulfite treatment was unavailable for 17 carcinomas.

Fluorescent-labeled PCR amplification was performed using the markers on the short arm of chromosome 9 (D9S161, D9S259, D9S171, and D9S285, centromeric to telomeric). D9S161, D9S259, and D9S171 at 9p21 are centromeric to p16 gene locus, and D9S285 at 9p22-9p23 is telomeric to p16 gene locus. The fluorescent dye-labeled and unlabeled primers were obtained (Life Technologies, Inc., Gaithersburg, MD). The 5′-oligonucleotide was end labeled with 6-FAM (D9S161, D9S259) or HEX (D9S171, D9S285) fluorescent dyes. PCR was performed with 40 ng of DNA in reaction mix consisting of 1× GeneAmp PCR Gold buffer (Applied Biosystems, Foster City, CA), 2.5 mm MgCl₂, 200 μm deoxynucleotide triphosphates, 0.83 μm of each primer, and 2 units of AmpliTaq Gold DNA polymerase (Applied Biosystems) in a final volume of 15 μl. PCR was performed using a GeneAmp PCR system 9700 thermal cycler (Applied Biosystems) with the following conditions: 95°C for 7 min; 3 cycles of 94°C for 1 min, 58°C for 30 s, and 72°C for 45 s; 42 cycles of 93°C for 45 s, 54°C for 30 s, 72°C for 40 s; and final extension at 72°C for 30 min. A 0.25-μl aliquot of each fluorescent-labeled PCR product was analyzed on an ABI 3700 Genetic Analyzer using GeneScan Analysis software (Applied Biosystems). Each sample included GeneScan 500 (TAMRA) size standard for accurate size calling. The ratio of the peak areas of the two alleles from the tumor DNA was compared with the ratio of the peak area of the two alleles from the peripheral blood DNA (examples in Fig. 1 B). Samples showing difference of ≥50% were classified as LOH at that marker. Representative samples were reanalyzed to confirm the chromosome 9p loss status. Nontumor DNA from either slides or blood was unavailable for 27 patients, and 11 patients were noninformative for the markers used in this study.

χ² or Fisher’s exact tests were performed for comparison of frequency of each clinical or pathological characteristic. The t test was used for comparison of means. Overall survival time was measured from the date of surgery and compared using the product limit method (27). The simultaneous effects of more than one prognostic factors were estimated by multiple regression in the proportional hazards model (28). The overall survival time of patients with somatic p16 alterations was compared with those without p16 alterations according to at least two different assays using the log-rank statistic (Epistat Services, Richardson, TX). All three carcinomas with silent mutations without other p16 alterations were excluded because these carcinomas had p14 missense mutations.

p16 gene alterations including p16 mutations (including silent mutations and/or p14 gene mutations due to a shared exon), methylation of the CpG island in the 5′-region of the p16 gene, and loss of chromosome 9p21-22 that targets p14, p15, and p16 genes were present in 72 (61.0%) of 118 biliary tract carcinomas, including 42 of 68 (61.8%) gallbladder cancers, 18 of 33 (54.5%) bile duct cancers, and 12 of 17 (70.6%) ampullary cancers (Fig. 2). The frequency of these alterations in gallbladder, bile duct, and ampullary carcinomas were 13 of 53 (24.5%), 5 of 25 (20.0%), and 3 of 13 (23.1%) for nonsilent p16 gene mutations, 8 of 54 (14.8%), 5 of 31 (16.1%), and 6 of 15 (40.0%) for p16 methylation, and 32 of 44 (72.7%), 12 of 21 (57.1%), and 8 of 16 (50.0%) for LOH of chromosome 9p, respectively. LOH of chromosome 9p was more prevalent in gallbladder carcinomas and bile duct carcinomas than were p16 mutations or methylation of CpG islands in the 5′-region of the p16 gene (P = 0.000001 and P = 0.02, respectively). The frequency of LOH at D9S161, D9S259, D9S171, and D9S285 was 31 of 45 (68.9%), 23 of 31 (74.2%), 20 of 26 (76.9%), and 17 of 40 (42.5%) of informative cases, respectively.

p16 gene mutations are tabulated in Fig. 2. There were 36 mutations present in 18 gallbladder carcinomas, 6 bile duct carcinomas, and 3 ampullary carcinomas. The mutations included a noncoding mutation in the promoter region, 8 mutations in exon 1 and 27 mutations in exon 2 of p16 gene. Twelve carcinomas had silent mutations in p16 gene, but 9 were associated with other p16 alterations. As exon 2 of p16 gene also codes for exon 2 of p14 gene in a different reading frame, all missense and silent mutations in exon 2 of p16 gene were missense or silent mutations in exon 2 of p14 gene. There were 21 missense and 5 silent mutations of the p14 gene.

Among 32 coding single-base substitutions, all were present at G or C nucleotides; 30 (93.7%) were transitions (19 G to A and 11 C to T) and only 2 (6.3%) transversions (1 C to A and 1 G to T). Furthermore, 20 (62.5%) of the 32 mutations were present at non-CpG dinucleotides and 12 (37.5%) at CpG dinucleotides. Eleven of these 12 single-base substitutions at CpG dinucleotides were G-A or C-T transitions.

Three patients with gallbladder carcinomas, 2 with bile duct carcinomas, and 1 with ampullary carcinoma had two or more p16 and/or p14 mutations. In addition, of 98 carcinomas that were studied by two or more methods, 12 gallbladder carcinomas, 3 bile duct carcinomas, and 5 ampullary carcinomas had two or more p16 alterations (Fig. 2). The prevalence of p16 alterations was more frequent if all three methods were used: 24 of 31 (77.4%) gallbladder carcinomas, 12 of 16 (75.0%) bile duct carcinomas, and 9 of 12 (75.0%) ampullary carcinomas had p16 alterations in patients whose carcinomas were examined by all three methods—DNA sequencing, methylation, and LOH.

The clinicopathological characteristics of biliary tract carcinomas, p16 alteration status, and survival times were evaluated for 101 patients, including 57 patients with gallbladder carcinomas, 28 with bile duct carcinomas, and 16 with ampullary carcinomas with and without p16 alterations according to at least two different methods (Fig. 2). Two gallbladder carcinomas and one bile duct carcinoma with silent p16 mutations (with p14 mutations) without other p16 alterations were excluded. None of 4 (0%) adenosquamous gallbladder carcinomas had p16 alterations compared with 40 of 53 (75.5%) gallbladder carcinomas of other histological types (P = 0.001). The vital status, age or gender of patient, presence of cholelithiasis, and histological type of bile duct and ampullary carcinomas were not associated with p16 status.

Patients whose carcinomas lacked p16 alterations had a better mean survival compared with those with alterations. The mean survival of patients whose carcinomas lacked p16 alterations was 20.5 ± 13.4 months, compared with 14.2 ± 11.9 months for those with alterations (P = 0.04; Fig. 3 A).

As expected, clinical stage and histology of the carcinomas were significant prognostic factors. The overall mean survival of patients with stage I, II, III, and IV disease was 26.5 ± 11.5, 17.9 ± 12.1, 13.5 ± 12.0, and 12.0 ± 11.4 months, respectively (P = 0.009). The mean survival of patients with p16 alterations versus those without was 26.5 ± 11.5 versus 29.2 ± 12.3 months for stage I carcinomas, 17.9 ± 12.1 versus 22.1 ± 12.5 months for stage II carcinomas, 13.8 ± 12.3 versus 18.3 ± 15.4 months for stage III carcinomas, and 12.0 ± 11.4 versus 17.5 ± 13.5 months for stage IV carcinomas, respectively (P = 0.002, Fig. 3 B). Patients without p16 alterations had a better prognosis at each stage, particularly those with stage II–IV tumors. Thus, the mean survival of stage III and IV patients with biliary tract cancers lacking p16 alterations resembled that for stage II patients with p16 alterations.

The histology of carcinomas was also associated with prognosis and with p16 alteration status. The overall mean survival of patients with adenocarcinomas not otherwise specified, papillary adenocarcinomas, mucinous adenocarcinomas, adenosquamous carcinomas, and small cell carcinomas was 13.9 ± 11.2, 26.9 ± 12.5, 25.6 ± 9.8, 26.8 ± 16.3, and 6.3 ± 3.4 months, respectively (P = 0.00002). The mean survival of patients with p16 alterations versus those without was 13.3 ± 11.2 versus 15.6 ± 11.4 months for adenocarcinomas not otherwise specified, 23.3 ± 12.8 versus 35.3 ± 7.8 months for papillary adenocarcinomas, 24.5 ± 11.0 versus 30 months for mucinous adenocarcinomas, 6 versus 31.0 ± 14.2 months for adenosquamous carcinomas, and 4.7 ± 1.5 versus 11 months for small cell carcinomas, respectively (P = 0.00007). In the multiple regression proportional hazards model, only stage (P = 0.001) and p16 alterations (P = 0.045) were independent predictors of survival.

The difference in survival associated with p16 alterations status of carcinomas was present for all three subsites of biliary tract cancers but was statistically significant only for gallbladder carcinomas (Fig. 3,C and Table 1). The mean survival of patients whose gallbladder carcinomas lacked p16 alterations was 21.5 ± 14.8 months compared with 12.1 ± 11.4 months for those with alterations (P = 0.02). Patients whose gallbladder carcinomas lacked p16 alterations had a better survival for clinical stages II–IV and each histological type, but the latter was not statistically significant. The mean survival of patients with p16 alterations versus those without was 29.6 ± 8.0 versus 29.3 ± 12.3 months for stage I carcinomas, 11.4 ± 9.6 versus 26.5 ± 21.9 months for stage II carcinomas, 7.2 ± 8.1 versus 12.7 ± 11.6 months for stage III carcinomas, and 6.9 ± 5.2 versus 19.4 ± 16.0 months for stage IV carcinomas, respectively (P = 0.0004).

Few studies of the molecular genetics of biliary tract carcinomas have addressed clinical and pathological associations in well-characterized population samples. We therefore evaluated alterations of the p16 gene in 118 gallbladder, extrahepatic bile duct and ampullary cancers from a population-based study. Patients with gallbladder carcinomas associated with p16 gene alterations and/or alterations of other tumor suppressor genes on chromosome 9p had a poorer survival, stage for stage, compared with those without p16 alterations.

It is possible that other genes such as p14 or p15 may contribute to the survival disadvantage associated with the presence of p16 mutations or LOH at chromosome 9p in our study. As mutations in exon 2 of p16 gene are also p14 gene mutations in a different reading frames (9, 10), missense or silent mutations in exon 2 of the p16 gene were nonsense or missense mutations of the p14 gene in our study. Similarly, p14 and p15 genes are present in the same region of chromosome 9p as the p16 gene (8), and may be involved by allelic loss of chromosome 9p21–22 in our study. However, no p15 mutations were present in a previous study of biliary tract carcinomas (18), but involvement of p14 has not been evaluated. Overexpression of p14 induces G₁- and G₂-cell cycle arrest by a mechanism that is independent of CDKs (10). Chromosome 9p has additional tumor suppressor genes, which are deleted in a variety of other tumors (29), but their role in biliary tract cancer remains to be determined.

p16 gene mutations, methylation of CpG islands, and/or heterozygous deletions were present in 61.0% of the biliary tract carcinomas in our study. Homozygous deletion is another mechanism of p16 inactivation, but detection requires pure tumor DNA from cell lines or tumor xenografts (14). In our study, nonamplification of p16 gene was present in ten carcinomas by multiples PCR but also had LOH of chromosome 9p (data not shown). Other techniques such as real-time PCR or fluorescence in situ hybridization can be used to detect homozygous deletions in tumor tissue but require frozen tissue (30, 31, 32, 33). Previous studies have reported p16 mutations in biliary tract carcinomas (18, 19) and LOH of chromosome 9p in gallbladder carcinomas (19, 20, 21, 22). Most of the p16 mutations in the previous study were missense point mutations or silent mutations (18). Although these alterations predominated as well in our study, we also noted nonsense and silent mutations, along with a deletion resulting in frameshift mutation. Two or more mutations and/or alterations were present in 27 biliary tract carcinomas in our study, which probably represent inactivation of both alleles of the p16 gene.

In our study, the p16 alterations status of gallbladder carcinomas was associated with prognosis. Patients whose carcinomas lacked p16 alterations had a better mean overall survival compared with those whose carcinomas had p16 alterations. This difference in survival was associated with the stage of gallbladder cancer. In our study, the frequency of p16 alterations was similar in patients with gallbladder, bile duct, and ampullary carcinomas, but p16 alteration status was a significant prognostic factor only for patients with gallbladder carcinomas. In contrast, a recent study has shown loss or heterogeneous p16 protein expression was associated with poor survival among patients with bile duct and ampullary carcinomas (23). This difference in prognosis may be due to the geographical or racial differences in the study population and/or differences in the methodology of the two studies.

Comparison of p16 gene alterations in biliary tract cancers versus other malignancies of the hepatopancreaticobiliary system suggests site-specific differences in the mechanisms of p16 inactivation (34). Inactivation of p16 is frequent in pancreatic malignancies and occurs usually by homozygous or intragenic deletions and occasionally by promoter region hypermethylation (35, 36, 37). However, in hepatocellular carcinomas and intrahepatic cholangiocarcinomas, p16 is inactivated frequently by methylation of the promoter region rather than by homozygous deletions or gene mutations (38, 39, 40, 41).

The majority of mutations present in our study have been described in previous compilations of p16 gene mutations (16, 17). However, there are interesting differences in the p16 mutations in our study as compared with other studies. All single-base substitutions were present at C or G nucleotides, whereas 93.8% were transitions at G or C. In contrast, transitions at G or C appear to be responsible for 45% of the base substitutions in a number of malignancies, and for 57% of the base substitutions in melanomas (17). The type of mutation depends on the gene, site of tumor, and type of mutagenic exposure; for example, G to T transversions are common in the p53 gene but not in p16 in smoking-related lung carcinomas (16, 42). Distinctive mutational signatures from exposure to a specific carcinogens have been reported in the tandem base substitutions at dipyramidine sequences in UV-related skin cancers (42), G to T transversions at codon 249 of the p53 gene in hepatocellular carcinomas due to aflatoxins (43), and G to T transversions in p53 gene due to benzo(a)pyrene adducts in smoking-related lung carcinomas (44). C to T transitions due to deamination of 5-methylcytosine residues in methylated CpG dinucleotides can arise spontaneously or due to reactive oxygen species or nitric oxide (45, 46, 47). In our study, the majority of point mutations were present at non-CpG dinucleotides. G/C to A/T transitions are seen after exposure to alkylating agents, primarily because of O⁶-methylguanine DNA adducts (48, 49, 50). O⁶-Methylguanine-DNA methyltransferase protects against the mutagenic effects of alkylating agents by removing O⁶-methylguanine adducts from the DNA, and this activity is modulated by in vitro and in vivo levels of O⁶-methylguanine-DNA methyltransferase (50, 51, 5253). It is possible that exposure to an alkylating agent may account for the increased G/C to A/T transitions in our study.

In conclusion, our study revealed that alterations of p16 gene and/or other tumor suppressor genes on chromosome 9p are frequent in biliary tract cancers and may cause p16 inactivation through gene mutations, methylation of the promoter region, and/or LOH of chromosome 9p. Patients with p16 alterations in their gallbladder carcinomas have a worse survival when compared with those without p16 alterations.

We thank Lu Sun and the staff at the Shanghai Cancer Institute; surgeons and pathologists at the collaborating hospitals; the staff at Westat, Inc., for study management and data preparation; Janis Koci at the National Cancer Institute for management of biological samples; and Patrick Scott Houlihan and Dr. Annie On-On Chan of the M. D. Anderson Cancer Center for assistance with figures.