High-Resolution Array Comparative Genomic Hybridization in Sporadic and Celiac Disease–Related Small Bowel Adenocarcinomas

Purpose: The molecular pathogenesis of small intestinal adenocarcinomas is not well understood. Understanding the molecular characteristics of small bowel adenocarcinoma may lead to more effective patient treatment.

Experimental Design: Forty-eight small bowel adenocarcinomas (33 non–celiac disease related and 15 celiac disease related) were characterized for chromosomal aberrations by high-resolution array comparative hybridization, microsatellite instability, and APC promoter methylation and mutation status. Findings were compared with clinicopathologic and survival data. Furthermore, molecular alterations were compared between celiac disease–related and non–celiac disease–related small bowel adenocarcinomas.

Results: DNA copy number changes were observed in 77% small bowel adenocarcinomas. The most frequent DNA copy number changes found were gains on 5p15.33-5p12, 7p22.3-7q11.21, 7q21.2-7q21.3, 7q22.1-7q34, 7q36.1, 7q36.3, 8q11.21-8q24.3, 9q34.11-9q34.3, 13q11-13q34, 16p13.3, 16p11.2, 19q13.2, and 20p13-20q13.33, and losses on 4p13-4q35.2, 5q15-5q21.1, and 21p11.2-21q22.11. Seven highly amplified regions were identified on 6p21.1, 7q21.1, 8p23.1, 11p13, 16p11.2, 17q12-q21.1, and 19q13.2. Celiac disease–related and non–celiac disease–related small bowel adenocarcinomas displayed similar chromosomal aberrations. Promoter hypermethylation of the APC gene was found in 48% non–celiac disease–related and 73% celiac disease–related small bowel adenocarcinomas. No nonsense mutations were found. Thirty-three percent of non–celiac disease–related small bowel adenocarcinomas showed microsatellite instability, whereas 67% of celiac disease–related small bowel adenocarcinomas were microsatellite unstable.

Conclusions: Our study characterized chromosomal aberrations and amplifications involved in small bowel adenocarcinoma. At the chromosomal level, celiac disease–related and non–celiac disease–related small bowel adenocarcinomas did not differ. A defect in the mismatch repair pathways seems to be more common in celiac disease–related than in non–celiac disease–related small bowel adenocarcinomas. In contrast to colon and gastric cancers, no APC nonsense mutations were found in small bowel adenocarcinoma. However, APC promoter methylation seems to be a common event in celiac disease–related small bowel adenocarcinoma. Clin Cancer Res; 16(5); 1391–401

Small bowel malignancies account for <2% of all gastrointestinal cancers (1). Malignancies of the small intestine include primary adenocarcinomas (small bowel adenocarcinomas), neuroendocrine tumors, lymphomas, and soft tissue tumors (2). Although the incidence of small bowel cancers is low, cancer-protective and predisposing factors have been proposed. Cancer-protective factors include rapid intestinal transit, low levels of bacteria, alkaline pH, rapid cell renewal, increased IgA levels, and low levels of activating enzymes of carcinogens in the small intestine (3). Predisposing factors are dietary habits, smoking, and alcohol ingestion, which have been implicated in most cancers (3). Patients who present with familial adenomatous polyposis, Lynch syndrome, Crohn's disease, Peutz-Jeghers syndrome, cystic fibrosis, peptic ulcer, and celiac disease have an increased risk for developing small bowel adenocarcinoma (4–7). Small bowel adenocarcinoma accounts for 50% of all small intestinal cancers and are mainly found in the duodenum (8). The incidence of small bowel adenocarcinoma is higher in industrialized countries, increases with age, and is more frequent in males than females (3). Small bowel adenocarcinoma is usually diagnosed at a late stage of the disease, and despite complete surgical removal of the tumor, local recurrences occur in 50% of the cases (9). Treatment is currently similar to that used for patients with colorectal or gastric cancers, and an agreement on the protocols of postoperative adjuvant therapy after surgery does not exist (1, 10, 11). Nevertheless, prognosis of patients with small bowel adenocarcinoma is poor, with a 5-year survival of 40% to 65% after curative surgery (1).

The molecular pathogenesis of small bowel adenocarcinoma is not well understood. Some studies suggested that small bowel adenocarcinomas develop through an adenoma-to-carcinoma progression pathway similar to colorectal cancer (12), whereas others refuted this hypothesis (13). Fifteen to twenty percent of small bowel adenocarcinomas show microsatellite instability (8, 14), and patients with microsatellite-unstable small bowel adenocarcinomas seem to have a better prognosis than patients with microsatellite stable tumors (15). Several studies have also documented mutations of SMAD4, APC, KRAS, and β-catenin (13, 16–18), as well as epigenetic silencing of APC and gains on chromosomes 7, 8, 12, 13q, and 20q, and losses on chromosomes 2, 4, 5q, 6q, 8, 9, 15q, 17p, 18, and 21 (19, 20). Studies evaluating the chromosomal aberrations in patients with concomitant small bowel adenocarcinoma and Crohn's disease or familial adenomatous polyposis have shown that the chromosomal aberrations of sporadic small bowel adenocarcinoma do not differ from patients with these predisposing conditions (19, 20).

Celiac disease is a chronic autoimmune disorder caused by intolerance to gluten in genetically predisposed individuals (21, 22). When gluten peptides reach the lamina propria in celiac disease patients, this triggers an immune response that leads to lymphocytic infiltration of the intestinal epithelium (Marsh I), crypt hyperplasia (Marsh II), and villous atrophy (Marsh III; ref. 23). It has been shown that an abnormal immune response is activated when HLA-DQ2/DQ8 molecules present gliadin to CD4+ T cells (24). However, the exact pathways leading to increased proliferation of the crypts and flattening of the villi are not fully understood (21). Treatment consists of life-long, gluten-free diet, which resolves the histologic and clinical symptoms and seems to be associated with a decreased incidence of malignancies (25, 26). The factors promoting higher risk for small bowel adenocarcinoma in celiac disease remain unknown. It is expected that genetic predisposition and molecular mechanisms involved in the pathogenesis of celiac disease could be related.

We hypothesized that unraveling the molecular mechanisms playing a role in the pathogenesis of small bowel adenocarcinoma will help us to better understand the clinicopathologic characteristics of these cancers and to identify potential markers that may be used to guide clinicians when deciding on treatment strategies for patients with small bowel adenocarcinoma. Therefore, the aim of our study was to determine the pattern of DNA copy number changes, microsatellite status, and APC mutation and promoter methylation status in a large set of sporadic small bowel adenocarcinomas and to compare the molecular profile of celiac disease–related and non–celiac disease–related small bowel adenocarcinomas.

Forty-eight sporadic small bowel adenocarcinomas (15 patients with small bowel adenocarcinoma showing histologic and clinical features of celiac disease and a comparable number of sporadic, that is, non–celiac disease–related small bowel adenocarcinoma cases) were collected retrospectively from the Pathology archives of the Leeds General Infirmary, Leeds, United Kingdom (n = 6), through the British Society of Gastroenterology National Survey (n = 39) and at the Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands (n = 3). Individuals with a diagnosis or family history of Lynch or Peutz-Jeghers syndrome, Crohn's disease, or familial adenomatous polyposis were excluded from this study. Clinicopathologic features of the 48 patients are listed in Table 1.

DNA from matched nonneoplastic mucosa and cancer was isolated using the QIAmp microkit (Qiagen) as previously described (27).

Ethical approval for the study was obtained from the Radboud University Nijmegen Medical Center (the Netherlands), the Leeds Research Ethics Committees (United Kingdom), and the North West Multicentre Research Ethics Committee (United Kingdom).

Array comparative genomic hybridization was done on 30K oligonucleotide arrays as described before (28, 29). Briefly, 600 ng of tumor and normal DNA were differently labeled by random priming (Bioprime DNA Labelling System, Invitrogen). Unincorporated nucleotides were removed with Sephadex columns (ProbeQuant G-50 Micro Columns, Amersham Biosciences), and 50 μL of tumor and nonneoplastic DNA were mixed with 10 μg Cot-1 DNA and precipitated by adding 2.5 volumes of ice-cold 100% ethanol and 0.1 volume of 3 mol/L sodium acetate (pH 5.2). The DNA was collected by centrifugation at 14,000 rpm for 30 min at 4°C. The obtained DNA pellet was dissolved in 130 μL hybridization solution and incubated at 73°C for 10 min to denature the DNA followed by 60 to 120 min incubation at 37°C. Thereafter, the hybridization mixture was added to the array in a hybridization station (HybArray12^TM, Perkin Elmer Life Sciences) and incubated for 38 h at 37°C. After hybridization, the slides were washed and dried by centrifugation for 3 min at 1,000 × g.

Images of the arrays from the array comparative genomic hybridization studies were acquired by scanning (Agilent DNA Microarray scanner, Agilent Technologies), and Bluefuse software version 3.4 (BlueGenome) was used for automatic feature extraction. Spots were excluded when the quality flag was <1 or the confidence value was <0.1. Log₂ tumor to normal ratios were calculated for each spot and normalized against the mode of the ratios of all autosomes. For determining copy number gains and losses, the R package comparative genomic hybridization call was used (30). Output of the comparative genomic hybridization call analysis was used for comparative genomic hybridization region analysis to reduce the number of data points while maintaining the genomic variance measured over the genome and thus increasing statistical power using a threshold for average error rate of 0.001. Log₂ tumor to reference ratio > 1 was regarded as amplification. Hierarchical cluster analysis was done using the WECCA program and the parameter total linkage (31). Analysis was done according to the oligonucleotide position from the UCSC May 2004 freeze of the Human Golden Path⁷

. Array comparative genomic hybridization data is available at Gene Expression Omnibus, accession number GSE15470.

Expression of MLH1 and MSH2 was investigated by immunohistochemistry using existing tissue microarrays. Five-micrometer sections were deparaffinized in xylene, washed in 100% ethanol, and rehydrated in tap water. Endogenous peroxidase was blocked using 3% hydrogen peroxidase in methanol, and antigen retrieval was done by boiling the slides in a pressure cooker for 2 min in Antigen Unmasking Solution (Vector Laboratories Ltd.). Slides were incubated with 10% casein (Vector Laboratories Ltd.) to block nonspecific protein binding followed by incubation with a mouse monoclonal antibody for MLH1 (BD Biosciences) diluted 1:70 overnight at 4°C and a mouse monoclonal antibody for MSH2 (Oncogene) diluted 1:100 for 1 h at room temperature. The DAKO EnVision kit (DAKO EnVision, HRP Rabbit/Mouse, DakoCytomation) was used as detection system according to the manufacturer's protocols. 3,3'-Diaminobenzidine (DAKO) was used as chromogen, and the reaction product was enhanced by incubating the slides in 0.5% copper sulphate/0.9% saline, followed by counterstaining with Mayer's hematoxylin. The slides were scored by two independent observers (C. Verbeke and K. Maude). Tumors were categorized as “negative” when there was no nuclear expression of MLH1 or MSH2 in the tumor cells in the presence of positive nuclear staining in normal control tissue.

Microsatellite status was established using DNA extracted from the paraffin blocks and the MSI Analysis System (MSI Multiplex System Version 1.2, Promega) containing the BAT-25, BAT-26, NR-21, NR-24, MONO-27 monomorphic markers following the manufacturer's instructions. The obtained PCR products were separated on an ABI 3130 DNA sequencer (Applied Biosystems) and analyzed by GeneScan 3100 (Applied Biosystems). Tumors were considered as microsatellite unstable when two or more markers showed length changes (instability).

APC mutation status was analyzed by sequencing the mutation cluster region (MCR) by four flanking PCRs. Subsequently, each of these flanking PCRs was reamplified in two seminested PCRs, in which the first seminested PCR shared the forward primer of the flanking PCR. Annealing temperatures are listed in supplementary Table 1. The PCRs included 2.5 μL of GeneAmp 10× PCR Buffer II, 0.25 μL of AmpliTaq Gold (5 U/L), 1.5 μL MgCl₂ (25 mmol/L; Applied Biosystems), 2.5 μL dNTPs (2 mnol/L), 50 ng of DNA, and 1 μL of each of the primer sets (10 pmol/μL) in a final volume of 25 μL. PCRs were done following an initial denaturation at 94°C for 2 min; an annealing step of 35 cycles, including 30 s at 94°C, 30 s at the corresponding annealing temperature, and 30 s at 68°C; and an extension step of 5 min at 68°C. The PCR product was purified by mixing 5 μL of the PCR product, 1 unit/μL of Shrimp Alkaline Phosphatase, and 1 unit/μL of Exonuclease I enzymes (USB Corporation) and incubating the mix for 30 min at 37°C, followed by 15 min at 80°C. Subsequently, samples were prepared for sequencing using the ABI PRISM BigDye terminator cycle sequencing ready kit (Applied Biosystems) according to manufacturer's protocol. Sequencing reactions were further purified by precipitation with 50 μL 100% ethanol, 2 μL sodium acetate (3 mol/L; pH 5.3), and 10 μL H₂O. Samples were incubated at −80°C during 1 h and then washed with 100 μL ethanol 70%. Lastly, the pellets were dissolved in 10 μL of HiDi formamide (Applied Biosystems) and then heated 3 min at 95°C and incubated 10 min in ice. The sequencing PCR products were separated on a 3130 DNA sequencer (Applied Biosystems), and analyses were carried out with the sequencer navigator and Vector NTI software (Invitrogen).

Promoter methylation status of APC and MLH1 was investigated by methylation-specific PCR. Before methylation-specific PCR, samples were subjected to sodium bisulphite treatment following the protocol of the EZ DNA Methylation kit (Zymo Research Corporation) using 500 ng of DNA as starting material and eluted in 50 μL 1 mmol/L Tris-HCl (pH 8.0). Primers and annealing temperatures are listed in Supplementary Table 2. Reactions were done in a final volume of 20 μL containing 1 μL of bisulphite-treated DNA, 1.88 μL of 10× PCR buffer including MgCl₂ (20 mmol/L), 0.65 μL 10× PCR buffer without MgCl₂, 0.125 μL of Faststart TaqPolymerase (Roche Molecular Biochemicals), 2.5 μL dNTPs (2 mmol/L), and 0.125 μL of each set of primers (20 pmol/μL). Amplification conditions included an initial denaturation step for 8 min at 95°C; an annealing step of 35 cycles at 95°C for 3 s, the corresponding annealing temperature for 30 s, and 45 s at 72°C; and an extension step at 72°C for 4 min. All methylation-specific PCR reactions included bisulphite-treated DNA from the Colo205 cell line and DNA derived from human placenta as positive controls for unmethylated alleles and human placenta DNA treated in vitro with SssI methyltransferase (New England Biolabs) for the methylated alleles. All methylation-specific PCR products were analyzed by electrophoreses using a 2% agarose gel.

Comparative genomic hybridization test, which comprises a dedicated Wilcoxon–Mann-Whitney two-sample test for array comparative genomic hybridization, was used for analyzing the differences in DNA copy number aberrations between celiac disease–related and non–celiac disease–related small bowel adenocarcinomas⁸

. Identified chromosomal regions with P < 0.05 and false discovery rate < 0.15 were considered to be significant. Other statistical analyses were done using SPSS 12.0.1 for Windows (SPSS, Inc.). Pearson χ² test was used for analyzing the relationship between cluster membership and clinicopathologic variables, as well as celiac disease status. Univariate survival analyses were done using the Kaplan-Meier method and calculating a long-rank test (SPSS, Inc.). The survival time was calculated from the day of surgery to the day of death due to small bowel adenocarcinoma (event) or the last day of follow up (censored). Patients who died within 30 d were classified as postoperative mortality (n = 1) or categorized as palliative treatment (n = 6) and excluded from survival analyses. Multivariate analysis was done using a Cox's proportional hazard regression model in a forward stepwise method for variable selection. Gender, age, tumor stage, lymph node stage, and celiac disease state were entered into the analysis. P < 0.05 was considered to be significant.

DNA copy number changes were observed in 37 (77%) of the 48 small bowel adenocarcinomas. The mean number of chromosomal events (gains and losses) per tumor was 5 (range, 0-32), consisting of 3.4 gains (range, 0-23) and 1.6 losses (range, 0-9). The following aberrations were observed in >10% of small bowel adenocarcinomas: gains on 5p15.33-5p12, 7p22.3-7q11.21, 7q21.2-7q21.3, 7q22.1-7q34, 7q36.1, 7q36.3, 8q11.21-8q24.3, 9q34.11-9q34.3, 13q11-13q34, 16p13.3, 16p11.2, 19q13.2, and 20p13-20q13.33 and losses on 4p13-4q35.2, 5q15-5q21.1, and 21p11.2-21q22.11. A detailed overview of frequencies of gains and losses is shown in Fig. 1. DNA amplifications were found on 6p21.1, 7q21.1, 8p23.1, 11p13, 16p11.2, 17q12-q21.2, and 19q13.2. The 31 genes included in these amplified regions are listed in Table 2 (Ensembl release 49). Expression of some of the genes identified in these amplified regions [e.g., A kinase anchor protein 9; LIM domain only 2 (rhombotin-like 1); glia maturation factor, γ; sterile α motif domain containing 4B; plexin domain containing 1; calcium channel, voltage-dependent, β1 subunit; and GATA-binding protein 4] has been described in the normal small bowel (Human Atlas project version 4.0).

Forty-five small bowel adenocarcinomas were selected based on positive nuclear immunohistochemical staining of MLH1 or MSH2 of neoplastic cells. To confirm the microsatellite status of these tumors, further microsatellite analyses using five monomorphic markers was carried out. Results were available for 41 of the 45 small bowel adenocarcinomas. Thirty-eight of 41 small bowel adenocarcinomas with MLH1 or MSH2 staining and microsatellite instability analysis data showed concordance in microsatellite status by both protein expression of MLH1 or MSH2 and the monomorphic markers analysis, and three cases showed contradictory results. In addition, three additional celiac disease–related small bowel adenocarcinomas were investigated only by the monomorphic markers and found to be microsatellite-unstable cancers. In total, 20% of 44 small bowel adenocarcinomas with monomorphic analysis available was microsatellite unstable, of which three (33%) were non–celiac disease–related tumors and six (67%) were celiac disease–related small bowel adenocarcinomas. All microsatellite-unstable cases showed promoter methylation of the MLH1 gene. No differences in survival were found between patients with microsatellite-unstable and microsatellite-stable cancers.

DNA copy number changes were observed in 24 (72.7%) of the 33 non–celiac disease–related small bowel adenocarcinomas. Thirteen (87%) of the 15 celiac disease–related small bowel adenocarcinomas showed chromosomal aberrations. The mean number of chromosomal events (gains and losses) per tumor in the non–celiac disease–related small bowel adenocarcinomas was five (range, 0-25), consisting of 3.2 gains (range, 0-20) and 1.8 losses (range, 0-9). The mean number of chromosomal events (gains and losses) per tumor in the celiac disease–related small bowel adenocarcinomas was five (range, 0-32), consisting of 3.8 gains (range, 0-23) and 1.2 losses (range, 0-9). Chromosomal aberrations common to non–celiac disease–related and celiac disease–related small bowel adenocarcinomas were gains on 7p22.3-p22.1, 7q22.1-q34, 8q12.3-8q24.3, 20p13-q13.33, and 20q11.21-q13.33 (Fig. 2A and B). The non–celiac disease–related small bowel adenocarcinomas showed more commonly a loss of 4q12-q35.2 compared with celiac disease–related small bowel adenocarcinomas (Fig. 2A). Celiac disease–related small bowel adenocarcinomas showed more frequently a gain on 5p13.33-p15.1, 7p14.3, 7q36.1, 9q34.3, 13q12.11-q13.34, and 20p13-20q11.21 and a loss on 5q15-q23.1 compared with non–celiac disease–related small bowel adenocarcinomas (Fig. 2B). Unsupervised hierarchical cluster analysis of all 48 small bowel adenocarcinomas did not show significant association between cluster membership and celiac disease–related or non–celiac disease–related small bowel adenocarcinomas. There was also no significant association between cluster membership and celiac disease–related or non–celiac disease–related small bowel adenocarcinomas after excluding the microsatellite-unstable cases. No significant association was found between cluster membership and other clinicopathologic variables such as age, gender, location, survival, or tumor stage. Patients with celiac disease–related small bowel adenocarcinoma did not show significant differences in survival compared with patients with non–celiac disease–related small bowel adenocarcinoma. None of the celiac disease–related small bowel adenocarcinomas showed DNA amplifications.

To further investigate the contribution of 5q15-q23.1 loss in the pathogenesis of small bowel adenocarcinoma, we examined the mutation and promoter methylation status of the tumor suppressor gene APC (5q21-q22). APC promoter hypermethylation was observed in 46% of 46 small bowel adenocarcinomas, with a higher frequency in the celiac disease–related cancers (73%) compared with the non–celiac disease–related cancers (48%; P = 0.009). Information about APC mutation status was available for 31 small bowel adenocarcinomas, of which 12 were celiac disease–related and 19 were non–celiac disease–related small bowel adenocarcinomas. Seven missense and two silent mutations were found, but no nonsense mutations were detected. APC mutations were detected in 8 (42%) of the 19 non–celiac disease–related small bowel adenocarcinomas and in 1 (7%) of the 12 celiac disease–related small bowel adenocarcinomas (P = 0.04). Patients with small bowel adenocarcinomas showing no mutation in the APC gene had a significantly better survival than patients with mutated APC gene in univariate (P = 0.006; log-rank, 7.5) and multivariate (P = 0.04; hazard ratio, 6.0) analysis, whereas none of the other variables showed additional prognostic value (Fig. 3). No significant differences in survival were found between patients with small bowel adenocarcinoma with and without APC promoter hypermethylation.

The molecular characteristics of small bowel adenocarcinoma are still largely unknown. A number of small studies have shown alterations in the DNA mismatch repair system, chromosomal aberrations, mutations, and methylation changes in a small number of genes in small bowel adenocarcinoma (8, 13, 14, 16, 20, 34). It has been shown that the risk for developing small bowel adenocarcinoma is increased in patients with celiac disease (relative risk, 60- to 80-fold; refs. 33, 34). However, the mechanisms that predispose celiac disease patients to small bowel adenocarcinoma are not known. The aim of our study was to comprehensively characterize small bowel adenocarcinoma using array comparative genomic hybridization methodology to analyze the status of the mismatch repair system and to investigate the APC gene (32, 35).

DNA copy number changes were observed in 77% of 48 small bowel adenocarcinomas. This observation is in agreement with previous studies, which identified chromosomal aberrations in 82% to 86% of small bowel adenocarcinomas (19, 20). Our study identified for the first time gains on chromosome 5p, 9q, 16p, and 19q and loss on chromosome 21q in small bowel adenocarcinomas. Most of the other regions with gains and losses identified in our study overlap with previously published studies (19, 20, 36). However, by using high resolution array comparative genomic hybridization, we were able to narrow down the regions with chromosomal gains to 7p22.3-7p22.1, 7p14.3, 7q22.1, 8q22.3-24.3, 13q11-q34.3, and 20p13-q13.33. These chromosomal regions have also been implicated in the pathogenesis of colorectal cancer and gastric cancer, indicating possible common mechanisms in the pathogenesis of gastrointestinal malignancies (37–40). The oncogene c-myc, located on 8q24.21, has been implicated in several cancer types, including colorectal cancer and gastric cancer (41, 42). On the 13q region, two candidate oncogenes have been studied in colorectal cancer, the miR-17-92 cluster and the CDK8 gene. The miR-17-92 cluster has been implicated in the progression of colorectal adenoma to carcinoma due to the transcriptional activity of c-myc and the DNA copy number gain of the miR-17-92 locus (43) and, recently, the CDK8 gene, which has been proposed to be involved in the pathogenesis of colorectal cancer as an oncogene by coactivating β-catenin (44). Multiple oncogenes contributing to the colorectal adenomacarcinoma progression have been proposed to be located on the 20q amplicon. Nevertheless, none of these genes have been identified as putative oncogenes in small bowel adenocarcinoma. Because of the use of high resolution oligoarrays, we were able to identify seven highly amplified regions in small bowel adenocarcinoma containing genes expressed in the normal small bowel (45).

Our study showed that 20% of small bowel adenocarcinomas show microsatellite instability, which is inconcordance with the 10% to 20% shown in previous reports in small bowel adenocarcinoma (8, 14, 15, 19). Our assessment of the mismatch repair status by immunohistochemistry and microsatellite markers provided further evidence for the suggestion that microsatellite markers may be more sensitive than the immunohistochemistry approach as 92.7% were microsatellite unstable by immunohistochemistry and monomorphic microsatellite markers.

In an attempt to elucidate whether sporadic small bowel adenocarcinomas are molecularly different from celiac disease–related small bowel adenocarcinomas, we compared the molecular findings between 33 sporadic small bowel adenocarcinomas and 15 celiac disease–related small bowel adenocarcinomas. Interestingly, only 33% of the non–celiac disease–related small bowel adenocarcinomas were microsatellite unstable, whereas 67% of the celiac disease–related small bowel adenocarcinomas were microsatellite unstable. This observation is in concordance with the results reported by Potter et al. (14) It has previously been proposed that chronic vitamin deficiencies or chronic inflammation as seen in celiac disease patients who do not adhere strictly to the recommended gluten-free diet can affect the mismatch repair system (46, 47). In the current study, the small bowel mucosa of only one of the 15 celiac disease patients showed complete histologic recovery despite being on gluten-free diet. Current treatment of small bowel adenocarcinoma commonly includes surgical resection and adjuvant 5-fluorouracil as it is standard in colorectal cancer (1). It has been reported that microsatellite instability–positive colorectal cancer does not response to treatment with 5-fluorouracil as well as microsatellite stable colorectal cancer (48). Because our study and others showed that the frequency of microsatellite instability in celiac disease–related small bowel adenocarcinomas is much higher (43-73%; refs. 8, 14, 15, 19) than in colorectal cancer (15-20%; ref. 37) and non–celiac disease–related small bowel adenocarcinomas (10%), the assessment of the microsatellite status of the celiac disease–related small bowel adenocarcinomas may be worth considering before starting chemotherapy.

Our study showed no differences in DNA copy number changes between celiac disease–related and non–celiac disease–related small bowel adenocarcinomas, and we therefore concluded that both tumor types are characterized by the same chromosomal aberrations. However, it needs to be emphasized that the cohort of patients with celiac disease–related small bowel adenocarcinomas is relatively small and that a study in a second larger cohort is required to confirm these results.

Although not statistically significant, celiac disease–related small bowel adenocarcinomas showed more frequently a loss on 5q15-q23.1 compared with celiac disease–related small bowel adenocarcinomas. This prompted us to investigate the mutation and promoter methylation status of the tumor suppressor gene APC (5q21-q22). Mutations in the MRC region of the APC gene were five times more common in non–celiac disease–related small bowel adenocarcinomas, whereas APC promoter hypermethylation was found much more frequently in celiac disease–related small bowel adenocarcinomas, which is in concordance with previous studies (13, 18, 34, 49). The APC mutation and methylation status in small bowel adenocarcinoma are in contrast to the results reported in colorectal cancer in which 60% to 80% of the tumors show nonsense mutations of the MRC region and only 20% to 30% display promoter hypermethylation (50, 51). No nonsense APC mutations were found in small bowel adenocarcinomas in the current study. It has been proposed that small bowel adenocarcinoma develops through the adenoma-to-carcinoma sequence, as in colorectal cancer (12). Our results on the APC gene indicate that this gene may not play a major role in the pathogenesis of small bowel adenocarcinoma because nonsense mutations seem not to occur. Interestingly, missense mutations of the APC gene have also been described in familial adenomatous polyposis patients (49). These studies have shown that some of these missense mutations may confer increase susceptibility to colorectal cancer (52, 53). The implications of these type of mutations remain unknown; however, it has been hypothesized that missense mutations could have a dominant effect and affect splicing or the RNA stability of the gene (53, 54). In the present series, presence of missense or silent APC mutations, but not tumor stage, was associated with poor survival in small bowel adenocarcinoma patients. The lack of association between survival and tumor and lymph node stages can be explained by the fact that 44 of 48 tumors were stage III or IV, whereas only two tumors with an APC mutation were lymph node positive. The large proportion of late stage tumors is inherent to the clinical course that is typical for small bowel adenocarcinoma, that is, limited awareness of the entity and the, until recently, limited options to diagnose these tumors at early stages. Yet, although this is the largest series of small bowel adenocarcinomas thus far that has been subjected to genomewide DNA copy number analysis, the sample size and unequal stage distribution form a limitation of the current study.

In summary, our study indicates that gains on 7p22.3-7p22.1, 7p14.3, 7q22.1, 8q22.3-24.3, 13q11-q34.3, and 20p13-q13.33 are the most frequent DNA copy number changes in small bowel adenocarcinoma, and further studies are warranted to identify the oncogenes located in these regions. However, the exact differences among the frequency of DNA copy number changes in small bowel adenocarcinoma and other gastrointestinal malignancy needs to be elucidated. Celiac disease–related small bowel adenocarcinomas showed defects in the mismatch repair system in larger numbers than in colorectal cancer and previously reported series of sporadic small bowel adenocarcinomas; however, at the chromosomal level, there is no statistical difference between celiac disease–related and non–celiac disease–related small bowel adenocarcinomas. Furthermore, additional functional studies on the contribution of missense mutations of the APC gene in the pathogenesis of small bowel adenocarcinoma are required.

No potential conflicts of interest were disclosed.

We thank the members of the British Society of Gastroenterology for reporting their patients to the National Survey.

Grant Support: Netherlands Organization for Medical and Health Research (ZonMW; grant number 6120.0022), the Dutch Cancer Society (grant number KWF 2004-3051), the British Society of Gastroenterology for financial support for identifying the patients with small bowel adenocarcinoma through its Research Unit and the Pathologic Society of England and Ireland.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.