Allelic loss of 14q32 in the pathogenesis of gastrointestinal and ampullary malignancies: mapping of the target region to a 17 cM interval

Abstract

Purpose

The genetic basis for gastrointestinal and ampullary carcinomas remains uncertain. This study was performed to pinpoint novel chromosomal region involved in the tumorigenesis of gastrointestinal tract.

Methods

We screened the allelic status on 16 chromosomal arms in a patient with synchronous ampullary carcinoma and gastric cancer, but who had no family history of familial cancer syndrome. The significance of the shared 14q deletion was examined on clinical cohorts of sporadic gastric (n=12) and ampullary (n=10) carcinoma, respectively. Then, high-density allelotype mapping was performed on 14q32 by using 23 microsatellite markers for the synchronous tumors.

Results

The synchronous gastric and ampullary carcinomas had no frameshift mutations in the APC, MSH2, MSH3, and MSH6 genes. Among the microsatellite markers screened, only D14S267 showed identical loss in the synchronous tumors. The same allelic loss was also detected in one of ampullary carcinomas (10%) and two of gastric cancers (16.7%). Fine mapping of 14q determined a minimally deleted region between D14S65 and D14S1010 (17 centiMorgans) for the synchronous tumors.

Conclusions

This study illustrates a paradigm using molecular genetic approach in identifying chromosome 14q32 that may harbor a tumor suppressor gene involved in the pathogenesis of a subset of gastrointestinal and ampullary malignancies.

Introduction

Multiple primary neoplasms are a fairly common clinical occurrence. Although some cancers may co-occur by chance, a greater-than-chance association between two cancers can provide clues to a shared etiology (Neugut et al. 1992). Except for hereditary non-polyposis colorectal cancer syndrome, in which multiple colorectal tumors are observed in 20–40% of affected individuals, non-hereditary, multiple primary cancers of the gastrointestinal tract are uncommon. Synchronous primary colorectal and gastric cancer are among the best-known examples (Ohtani et al. 2000). Synchronous ampullary carcinoma (adenocarcinoma of the ampulla of Vater) and gastric cancer, however, is extremely rare. The co-existence of these two tumors thus gives us a good grasp of the shared genetic basis for multiple cancer development.

Carcinogenesis of gastrointestinal cancer is a multistep process characterized by accumulating alterations of oncogenes, tumor suppressor genes (TSG), and mismatch repair (MMR) genes (Fishel et al. 1993). Genetic instability has been implicated in the development of non-hereditary multiple primary cancers. The frequent microsatellite instability (MSI) in multiple synchronous epithelial carcinomas suggests that genetic instability due to certain defects in the MMR genes may play an important role in the development of multiple cancers (Horii et al. 1994).

Gastric cancer is common worldwide, but ampullary cancer is uncommon, and accounts for around 0.2% of gastrointestinal tract malignancies. Nevertheless, the genetic mechanisms of carcinogenesis for both types of cancer remain unclear. Allelotypic studies have provided information on the deleted chromosomal regions where important TSGs may exist. For example, loss of heterozygosity (LOH) on 4p, 7q, 14q, 17p, and 21q is a relatively early event in gastric carcinogenesis (Nishizuka et al. 1998), while allelic losses on the 1q, 5q, 7q, 9p, 11p, 11q, 13q, 16q, 17p, and 18q may play a role in the progression of gastric carcinoma (Kim et al. 2001). The frequently deleted chromosomes in biliary tract carcinoma are 4q, 5q, 9p, 14q, 17p, and 18q (Kang et al. 2000; Hidaka et al. 2001; Chang et al. 2002).

Because each patient exhibits a distinct pattern of allelic losses in the tumor (Chang et al. 2002), we screened the allelic status of synchronous ampullary carcinoma and gastric cancer in one patient on the chromosome arms 1p, 2, 3p, 4, 5, 8p, 9p, 9q, 11p, 11q, 12, 13q, 14q, 17p, 18q, and X. The clinical significance of the shared 14q deletion was analyzed on clinical cohorts of sporadic gastric and ampullary carcinoma. To further determine the minimally deleted region of 14q32, a high-density allelotype mapping was performed on the synchronous tumors by means of 23 microsatellite markers. The findings provide the genetic basis for the development and progression of gastrointestinal and ampullary malignancies.

Materials and methods

Clinical summary

An 85-year-old male uremic patient previously on regular hemodialysis for 2 years was admitted for acute exacerbation of chronic renal failure. He was unintentionally found to have anemia (Hb 6.5 mg/dl) and occult blood in his fecal specimen. A panendoscopy revealed a 4×2-cm pedunculated polyp on the greater curvature side of the gastric antrum and a 3×2-cm ulcerated protruding tumor on the papilla of Vater. Microscopic examination showed both lesions to be adenocarcinoma. Chest plain films, an abdominal enecho, a CT scan, and a whole-body radionuclide scan demonstrated no evident metastatic lesions. Whipple’s procedure was then performed. One month later, the patient died from deteriorated cardiopulmonary function superimposed on pseudomonal septicemia.

Pathological examination

The surgical specimen showed a gray brown polypoid tumor measuring 3×2×1.5 cm at the anterior aspect of the gastric antrum and 4 cm proximal to the pyloric ring. Incision revealed an essentially intramucosal lesion (type I early gastric cancer) without invasion of the submucosa. The second tumor, 2.5×2×1.5 cm was at the ampulla of Vater and 12 cm distal to the pyloric ring. The ampullary tumor had invaded the adjacent duodenal wall and partially obliterated the distal portions of the pancreatic and common bile ducts. However, the remaining peripyloric and peripancreatic lymph nodes, the spleen, and the omental tissue were unremarkable.

DNA extraction

Somatic genomic DNA was obtained from fresh-frozen tumor tissues from the gastric cancer and carcinoma of the papilla of Vater. The process of tissue preparation has been described in detail previously (Eisenberger et al. 1999). Briefly, 5-mm-thick sections were cut and fixed in ethanol. The first section was stained with H&E and used as a template. The remaining unstained sections were mounted onto slides and micro-dissected. Tissue fragments were then collected and placed in 1% SDS/proteinase K for overnight digestion. The digested tissues were subjected to phenol-chloroform extraction and ethanol precipitation. Corresponding germline DNA was obtained from peripheral blood leukocytes according to the standard protocol.

Microsatellite analysis

Microsatellite markers from selected chromosomal arms for PCR analysis were obtained from Research Genetics (Huntsville, Ala., USA). The marker pairs included 1p (D1S160, D1S186, D1S163, D1S170), 2 (BAT26), 3p (D3S1317), 4 (FGA, D4S1615, D4S620, D4S392, D4S1545, D4S1554, D4S1598, D4S4920, D4S2954, D4S2943, D4S398, FABP2), 5 (D5S353, D5S417, ACTBP2), 8p (D8S307, D8S261, LPL), 9p (D9S162, D9S171, D9S1747, D9S104, D9S169, INFA), 9q (GSN, D9S195, D9S747), 11p (D11S569, D11S907, D11S1344, D11S554), 11q (D11S426, D11S938, WT1), 12 (D12S93), 13q (D13S133), 14q (D14S267, D14S288, D14S51, D14S72, MJD), 17p (D17S695, D17S654, TP53), 18q (D18S51, D18S67, D18S535, MBP), and X (DXS538) (Chang et al. 2002). Before amplification, one primer from each pair was end-labeled with [g-32P]ATP (20 mCi; Amersham) and T4 kinase (New England Biolabs, Beverly, Mass., USA) (Eisenberger et al. 1999). PCR amplification for each primer set was performed for 35 cycles consisting of denaturation at 95 °C for 30 s, annealing at 50-60 °C for one min, and extension at 70 °C for one min. PCR products were separated on 7% urea-formamide-polyacrylamide gels and exposed from 4 h to 24 h. For informative cases, loss of heterozygosity (LOH) was scored if the intensity of the tumor allele demonstrated a greater than 30% reduction in intensity compared to the corresponding allele in normal control DNA by two independent observers (Chang et al. 2002). Allelic loss was scored by observation of LOH at any informative marker mapped to the same chromosomal region (Eisenberger et al. 1999).

Tissue samples from clinical cohort

To confirm the importance of LOH on 14q32 in the ampullary and gastric cancers, we collected the files for 12 cases of sporadic gastric carcinoma and ten cases of sporadic ampullary carcinoma from the pathology departments of two hospitals in different areas of the country, as indicated in the affiliations of the authors (details in Table 1). The histological grading of tumors was determined according to WHO system, and stage classification according to the TNM system (4th edn., 1987). The archival samples were reviewed for histology and then sectioned for allelic loss assessment. After microdissection of non-neoplastic elements such as germline control, the tumor cells were cut separately for DNA extraction as described above.

Table 1.

Clinicopathological profiles of patient cohort with sporadic gastric cancer and amullary carcinoma.

No.	Age	Sex	Histologya	Stageb	LOH on D14S267
Sporadic gastric carcinoma
1	64	M	Well	T2aN0M0	-
2	71	M	Well	T2aN0M0	-
3	67	M	Well	T2aN1M0	+ (ST1 in Fig. 5)
4	75	M	Moderate	T2aN1M0	-
5	73	M	Moderate	T2aN1M0	-
6	66	F	Moderate	T2bN1M0	-
7	75	F	Moderate	T2bN1M0	-
8	72	M	Poor	T2bN1M0	+ (ST2 in Fig. 5)
9	80	M	Poor	T3N0M0	-
10	51	F	Poor	T3N0M0	-
11	53	M	Poor	T2bN2M0	-
12	49	M	Poor	T3N1M0	-
Sporadic ampullary carcinoma
1	64	F	Moderate	T2N0M0	-
2	40	F	Moderate	T2N1M0	+ (AV1 in Fig. 5)
3	73	M	Well	T1N1M0	-
4	48	M	Well	T2N1M0	-
5	63	M	Moderate	T2N1M0	-
6	62	M	Moderate	T2N1M0	-
7	64	M	Poor	T3N1M0	-
8	52	F	Poor	T3N1M0	-
9	43	F	Poor	T3N1M0	-
10	57	M	Poor	T4N1M0	-

^aThe histological grading was determined according to criteria of WHO system

^bThe stage classification was determined according to TNM system

High-density deletion mapping on chromosome 14q32

We then performed LOH analysis on chromosome 14q32 using tissue of the ampullary and gastric carcinomas based on the results of the allelotype study. Twenty-three microsatellite markers were selected around the D14S267 locus for deletion mapping of the physical boundaries of 14q32. The order of marker arrangement from centromeric to telomeric side was as follows: D14S51 at 115.6 centiMorgans (cM), D14S979, D14S611, D14S1067, D14S65, D14S998, D14S1019, D14S267, D14S250, D14S543, D14S78, D14S1426, D14S1006, D14S985, D14S1051, D14S293, D14S118, D14S557, D14S272, D14S260, D14S1010, D14S292, D14S1007, and D14S544 at 138.18 cM, all determined by Marshfield Master Map (Invitrogen, Carlsbad, Calif., USA). These two loci are ~22.58 cM apart.

Bioinformatic analysis of the human chromosome 14q32 region

Microsatellite markers were chosen to search the human genome sequence database (NCBI) for all putative human genes within the critical region. With information of gene ontology (e.g., apoptosis, cell adhesion, developmental regulation, and cell cycle control) for each gene, the promising candidates for gastrointestinal tract malignancies thus could be determined.

Results

Microscopic findings

The ampullary tumor was a moderately differentiated adenocarcinoma in a complex tubulopapillary pattern (T2N0M0, stage II by AJCC) (Fig. 1). The adjacent duodenal submucosa and part of the muscular layer of the distal pancreatic and common bile ducts were all involved by the tumor. Neither lymphovascular nor perineural invasion was discernible. The spleen, omentum, and dissected lymph nodes were free of metastasis. The gastric antrum revealed a well-differentiated tubular-type adenocarcinoma invading the muscularis mucosae (early gastric cancer type I, T1N0M0, stage I by TCOG) (Fig. 2). Dysplasia of varying degrees and intestinal metaplasia were visible in the adjacent mucosa, which indicated that the primary origin of the adenocarcinoma was the gastric mucosa. In addition, there were some Helicobacter pylorus bacilli in the foveola.

Fig. 1.

Histopathology of ampullary carcinoma. The tumor shows a complex tubulo-papillary pattern (right half). The muscular layer of the common bile ducts was invaded by the tumor (left half). No evidence of lymphovascular or perineural invasion was found (magnification ×150).

Fig. 2.

Histopathology of gastric cancer. The gastric carcinoma revealed an irregular tubular pattern with mild to moderate cellular atypism and nuclear pleomorphism (magnification ×300).

Clinical examination

The occupational history, personal habits, residence, and medical history of the patients were essentially unremarkable. In addition, they had no apparent benign neoplasms and their first-degree relatives showed a similar array of benign and malignant single and multiple tumors, thus providing information about any patient’s predisposition to cancer.

Allelotype study

Both tumors retained the BAT locus, the most sensitive marker for MMR phenotype (Zhou et al. 1998). Only one instance of MSI on D18S535 was observed in gastric cancer (Fig. 3). Otherwise, LOH was seen mostly in one of the tumors or a different allele in both tumors. For example, LOH on p53 was detected in ampullary but not in gastric cancer. Taken together, discordant chromosomal losses in 1p, 4q, 5, 8p, 9p, 11p, 17p, and 18 were demonstrated between ampullary and gastric carcinomas in this allelotype screening (Fig. 4). Interestingly, both ampullary and gastric cancers showed identical loss of D14S267 (14q32.1–32.2) (Fig. 3). In contrast, there was retention of D14S72 (14q11.1), D14S51 (14q32.1), and D14S288 (14q32.1) in both tumors (Fig. 3). LOH on MJD (14q21) was detected only in gastric cancer (data not shown). Taken together, LOH on 14q32 is the commonly deleted region in both tumors.

Fig. 3.

Representative results of microsatellite analysis for synchronous ampullary and gastric cancers. Loss of heterozygosity (LOH) was scored if the tumor allele (AV ampullary carcinoma, ST stomach cancer) demonstrated a greater than 30% reduction in intensity compared to the corresponding allele in the non-malignant control (N). Results from D1S170, MBP, D4S4920, TP53, and D14S267 represent LOH, while microsatellite instability on D18S35 was observed in the gastric cancer. The arrows indicate relative loss of alleles in tumor samples.

Fig. 4.

Genome-wide loss of heterozygosity (LOH) screening for synchronous ampullary and gastric carcinomas. Allelotyping was performed using the 23 microsatellite markers listed in “Materials and methods”. Markers are labeled above each: non-cancerous (N), ampullary carcinoma (ampulla); and stomach cancer (stomach). 1 LOH on the upper allele: black box shows retention of both alleles; 2 LOH on the lower allele; dotted box: not informative.

LOH Analysis at the D14S267 locus in sporadic gastric and ampullary carcinoma

We then assessed whether the alteration of chromosome 14 also occurs in sporadic cases of gastric and ampullary carcinomas. LOH analysis was carried out on paired samples of cancer and corresponding non-cancerous elements of ampullary (n=10) and gastric cancer (n=12) at the D14S267 locus (details in Table 1). The reason that D14S267 was chosen for screening comes from the initial deletion mapping of our study and its deletion in 20% of cardiac carcinoma in a prior report (Yanagi et al. 2000). Apparent LOH at the locus was detected in one (10%) of the ampullary and two (16.7%) of the gastric cancers (16.7%) tested, respectively (Fig. 5, Fig. 6).

Fig. 5.

LOH analysis on D14S267 in sporadic ampullary and gastric cancers. Markers are labeled above each non-cancerous (N) and corresponding tumor (T) lanes. Autoradiographs depicting LOH in positive cases are shown as AV1 (ampullary carcinoma), ST1 (stomach cancer, case 1), and ST2 (stomach cancer, case 2). The arrows indicate relative loss of alleles in tumor samples.

Fig. 6.

Results of the fine mapping of the commonly deleted region on chromosome 14q32 in both tumors. The results of LOH analysis and relative position for ten of 23 microsatellite markers are illustrated in relation to the ideogram of chromosome 14 (the panel on the right-hand side). The arrows on the right side indicate relative loss of alleles in tumor samples. Retention of both alleles was observed in D14S51 and D14S292, whereas it was not informative for D14S293 and D14S544. Except for D14S293, there is identical LOH on the upper allele from D14S65 to D14S1010. The dark arrow indicates the critical border markers of minimally deleted region.

Deletion mapping on the chromosome 14q32

To further determine the minimally concordant LOH in these two tumors, we performed a detailed fine mapping on the chromosome 14q32 region. Typical patterns of LOH analysis are shown in Fig. 4. Identical LOH was present at the D14S65, D14S267, D14S250, D14S543, D14S78, D14S985, D14S557, and D14S1010 loci, and both alleles were retained at the D14S51, D14S1067, D14S998, D14S1426, D14S1006, and D14S118 loci. The remaining ten markers, however, were not informative. The critical border markers of 14q32 were determined to be D14S65 and D14S1010 (~17 cM by Marshfield Master Map).

Bioinformatic analysis of the chromosome 14q32 region

To search for candidate TSGs, a publicly available NCBI database was searched using D14S65 and D14S1010 as queries. This chromosome region corresponds to the human genome sequence contig NT_026437.10 (starting from sequence 77541501 to 84134543). Within the minimal deletion area, a total of 66 genes were identified within the genetic region using Map View (URL: http://www.ncbi.nlm.nih.gov/mapview/maps.cgi). A total of 30 genes have been labeled, i.e., BCL11B, FLJ23027, KIAA1822, CYP46A1, EML1, EVL, C14orf66, YY1, C14orf69, C14orf68, WARS, MGC4645, KIAA1446, MEG3, PPP2R5C, DNCH1, WDR20, RAGE, CINP, KIAA0329, RCOR, TRAF3, CDC42BPB, TNFAIP2, MARK3, MGC2562, KNS2, XRCC3, ZFYVE21, and PPP1R13B; there are 36 hypothetical genes. The functions of FLJ23027, MGC4645, KIAA0329, MGC2562, and ZFYVE21 are not known.

Discussion

Cancers of the biliary tract, including the gallbladder, extrahepatic bile duct, and ampulla of Vater, are uncommon and generally have a poor prognosis. The carcinogenic mechanisms underlying their induction are poorly understood, except for a close relationship with gallstones. Despite that, biliary tract cancers have been reported to co-occur with cancers of the prostate, ovary, breast, esophagus, rectum, pancreas, liver, colorectum, and stomach (Renault et al. 1978; Kitano et al. 1984; Schlossberg et al. 1988; Takayasu et al. 1992).

In this study, we found a differential pattern of allelic loss between our patient’s ampullary (LOH on 1p, 4q, 11p, 17p, and 18) and gastric cancers (LOH on 5, 8p, 9p, and 17p). These results essentially agree with the reported genetic alterations for gastric carcinoma (Nishizuka et al. 1998; Kim et al. 2001) and biliary tract cancer (Shiraishi et al. 1999; Kang et al. 2000; Hidaka et al. 2001; Shiraishi et al. 2001). Interestingly, loss of 14q was the only shared genetic alteration for both tumors, consistent with prior studies showing that LOH on 14q is one of the early events in the carcinogenesis of gastric (Nishizuka et al. 1998; Kim et al. 2001) and biliary tract cancer (Shiraishi et al. 1999; Kang et al. 2000; Hidaka et al. 2001; Shiraishi et al. 2001).

Further support for 14q32 in pathogenesis of gastrointestinal malignancies comes from recent molecular cytogenetic analyses (Bando et al. 1999; Yanagi et al. 2000; Ihara et al. 2002). LOH on D14S267 or D14S292 was observed in seven (17.9%) and two (5.1%) of 39 early colorectal cancer patients, respectively (Bando et al. 1999). The highest reported frequency at D14S267 was 50.0% in a cohort of 66 colorectal cancers, and the minimally deleted region was bounded by D14S65 and D14S250, spanning approximately 8 cM (Ihara et al. 2002). LOH on D14S267 or D14S65 was observed in ten (38.5%), and eight (30%) of 26 informative cases of esophageal carcinoma, respectively (Yanagi et al. 2000). LOH on D14S51 (14q32) was observed in 41.7% of intrahepatic cholangiocarcinoma (Kang et al. 2000). In the present study, LOH on D14S267 was discovered in 10% and 16.7% of sporadic ampullary and gastric cancers, respectively, suggesting that a TSG in 14q32 may play a role in the pathogenesis of a subset of gastrointestinal malignancies.

It should be noted that identical LOH pattern in eight consecutive microsatellite markers (D14S65, D14S267, D14S250, D14S543, D14S78, D14S985, D14S557, and D14S1010) for the synchronous tumors points toward a non-random basis for the allelic losses. Assuming equal probability of loss of paternal or maternal allele for each marker, the odds of losing these markers is small [P=(0.5)ⁿ⁻¹=0.5^8–1=0.0078]. The chance of identical LOH thus is therefore very remote in the same patient (P <0.001), suggesting that it is very unlikely that allelic loss of 14q32 in the synchronous cancers develops by chance.

Both TSHR (14q31) and AKT1 (14q32.3) have been proposed as candidate TSGs for stomach cancer, however (van Dekken et al. 1999; Rosenberg et al. 2002). To clarify the discrepancy, a detailed functional genomic investigation is mandatory. Currently, bioinformatic analysis has identified BCL11B, FLJ23027, KIAA1822, CYP46A1, and EML1 as the potential TSGs between D14S65 and D14S1010. CYP46A1 and EML1, as well as seven hypothetical genes, together with the region bounded by D14S65 and D14S250 in colorectal cancer, appear to be the current consensus genes involved in gastrointestinal tract carcinogenesis. A clarification may provide molecular targets for early diagnosis, disease prevention, and therapeutic intervention for sporadic gastrointestinal tract cancer.

The development of multiple primary cancers of the gastrointestinal tract and other organs is influenced mainly by environmental risk factors, such as heavy smoking and the excessive consumption of alcohol (Shimada et al. 1995). Uremic patients are known to have an increased incidence of malignancy, especially urologic cancer (Chen et al. 1995). A recent study revealed LOH on D14S267 in the hyperplastic parathyroid glands in uremic patients (4/23, 17.4%), implying a clonal change (Nagy et al. 2001). Because our patient also had uremia for 2 years before being diagnosed with cancer, it is conceivable that metabolic alterations associated with uremia might play a role in the development of multiple gastrointestinal malignancies.

The present study has produced several important conclusions. First, we have defined a minimal common region of deletion, 17 cM, at 14q32 in a subset of human ampullary and gastric carcinomas. Taken together with the results of bioinformatic analysis, our result should facilitate the cloning of candidate tumor suppressor genes at 14q32. Identification of the target genes may help greatly in early diagnosis, disease prevention, and therapeutic intervention for gastrointestinal tract cancer. Second, this study also illustrates a paradigm using molecular genetic methodologies to pinpoint candidate genes (s) involved in the pathogenesis of multiple primary gastrointestinal cancers.