Main

Although the small intestine represents the largest segment of the gastrointestinal tract and is located between two areas of high cancer risk, the stomach and the colon, adenocarcinoma development is exceedingly rare in this location (incidence 1/100 000, SEER Cancer Statistics 2003–20071). Small bowel adenocarcinoma and colorectal cancer show overlaps in morphology and hereditary and non-hereditary risk factors, including the hereditary non-polyposis colorectal cancer syndrome, familial adenomatous polyposis, and Crohn's disease.2, 3, 4 Recent studies, however, provide evidence for a somewhat different molecular pathogenesis of the majority of sporadic carcinomas in both locations.5, 6, 7, 8, 9

Microsatellite and chromosomal instability (MSI and CIN) are two divergent types of genomic instability found in adenocarcinomas throughout the gastrointestinal tract, including the small intestine.6, 7, 9, 10, 11, 12 MSI is caused by inactivation of the DNA mismatch repair system that may either result from promoter hypermethylation of the DNA mismatch repair gene MLH1 in sporadic carcinogenesis,13 or following germline mutations in DNA mismatch repair system genes in hereditary nonpolyposis colorectal cancer.14 CIN involves gains or losses of whole chromosomes and/or chromosomal fractions and is considered to result from defects in chromosomal segregation, telomere stability, and DNA damage repair.15, 16 In ∼80% of colon cancers, CIN and/or MSI are found. A share of 20% of colorectal cancers, however, is microsatellite and chromosomally stable (MACS), and the molecular pathogenesis of these colorectal cancers is not completely understood.17, 18

The CpG island methylation phenotype (CIMP) is an epigenetic phenomenon found in a variety of cancers.19, 20 In colorectal and gastric cancers, CIMP frequently associates with MLH1 methylation, loss of MLH1 transcription, and MSI-H.19, 21, 22 An association of CIMP with BRAF mutations has been found in colorectal cancer23 but, interestingly, mutations in BRAF do not have a significant role in either CIMP-positive or -negative gastric cancers.24

Analyses of genomic and epigenetic interrelations in human neoplasms have added largely to our understanding of tumor development and have led to a novel molecular classification of colorectal cancer.25 As the pathogenesis of small bowel adenocarcinoma is not completely understood and interrelations of epigenetic and genetic alterations have not been investigated in this tumor type, we analyzed 37 primary small bowel adenocarcinomas stratified by MSI and KRAS status for chromosomal imbalances, CIMP, and the presence of BRAF mutations. In this study we show that the majority of small bowel adenocarcinomas display genomic instability (either CIN and/or MSI) and that a subset of 24% of tumors is microsatellite and chromosomally stable. We further demonstrate similar epigenetic alterations and relations of BRAF/KRAS mutations in MSI-H and MACS but not in CIN cancers, indicating MSI-H and MACS carcinogenesis to follow similar initializing or promoting epigenetic and genetic events.

Materials and methods

Tumor Samples

A total of 37 unselected primary adenocarcinomas of the small intestine, including 15 duodenal, 7 jejunal, and 7 ileal tumors, as well as 8 adenocarcinomas in small intestinal segment resection without specification of jejunal or ileal origin, were analyzed. Familial adenomatous polyposis coli, juvenile polyposis, and Peutz–Jeghers-syndrome were excluded using pathologic examination. One carcinoma (884) was associated with celiac disease and one (271) with Crohn's disease. The duodenal cancer (320) was diagnosed in a patient with hereditary nonpolyposis colorectal cancer. Hereditary nonpolyposis colorectal cancer was considered in two other patients with MSI-H carcinomas (884, 235). In one of these patients, germline analysis of the MLH1 gene (884) revealed no mutation. In the second patient, no germline analysis has been performed. Patient age ranged from 29 to 84 years (median age 64). The clinical data are given in Table 1. The study was approved by the institutional ethics committee (application no. 206/05) at the Medical Faculty at Heidelberg University.

Table 1 Clinical data, genetic, and epigenetic alterations of the 37 investigated cases
Full size table

DNA Isolation

For the isolation of DNA from tumor samples for methylation-specific PCR, as well as for mutational analysis of BRAF, tissue slides of 5-μm thickness were prepared. Tumor tissue was isolated under microscopic control of unstained slides by manual microdissection. For comparative genomic hybridization (CGH), areas of high tumor cell concentration (>70%) were marked on the slide and the corresponding area was removed from the tumor block with the help of a 0.2 mm punch needle. For CGH and PCR analyses, DNA was prepared using the DNeasy tissue kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions.

MSI Analysis

Tumor samples had previously been typed for MSI as described.26 In brief, DNA isolated from the tumors and non-neoplastic normal mucosa was amplified. MSI analysis was performed using the standard NCI/ICG-hereditary nonpolyposis colorectal cancer marker panel27 and CAT25.28 PCR products were separated on an ABI3700 automated sequencer (Applied Biosystems, Warrington, UK). Tumors were classified as MSI-H if at least two of the markers displayed a shift in product length compared with the product of non-neoplastic mucosa.

CGH Analysis

CGH was performed as previously described in detail.29 In brief, tumor DNA was biotinylated and normal control DNA was extracted from healthy human placenta tissue and labeled with digoxigenin. After hybridization of the labeled DNA on metaphase spreads (CGH target slides; Abbott Molecular, Abbott Park, IL, USA) at 37 °C for 48 h, the slides were incubated with fluorescence-labeled anti-digoxigenin and anti-biotin antibodies. Metaphase spreads from each specimen were analyzed and photodocumented using an Axiovert S100 microscope (Zeiss, Jena, Germany), a CCD camera, and hardware and software as supplied by Metasystems (Altlussheim, Germany). Fluorescence ratios were determined using both fixed cutoff values (0.8 for losses and 1.25 for gains) and cutoff values with a twofold s.d. Chromosomal abnormalities detected with both approaches were scored. Centromeric and telomeric chromosomal areas as well as chromosome 19 were excluded from scoring. CGH data of 21 of the 37 carcinomas have previously been reported.7, 8

Methylation-Specific PCR

Methylation-specific PCR was performed for five CpG islands known to be methylated in colorectal cancer: MINT1, MINT2, MGMT, MLH1, and p16 (CDKN2A). Primers and PCR conditions were used as described previously (MINT1 and MINT2;30 MGMT;31 p16;32 MLH133). DNA was bisulfite converted using the EZ DNA Methylation-Gold Kit (Zymo Research, Orange, CA, USA) according to the manufacturer's instructions. One-fifth of a conversion reaction was used as template.

PCR products were visualized on 2% agarose TAE gel or 6% nondenaturing polyacrylamide gels (PAA). Methylation status was assessed based on the presence or absence of amplified product in the methylation-specific PCR. CIMP-high (CIMP-H), CIMP-low (CIMP-L), and CIMP-negative (CIMP-neg) were scored when >50, 20–50%, and 0% of the investigated loci were positive for methylation-specific PCR, respectively (Figure 1). Methylation-specific PCR data of 21 carcinomas were the subject of a previous study on the pathogenesis of celiac disease-associated small bowel adenocarcinomas.34

Figure 1
figure 1

(a) Representative results of methylation-specific PCR in six tumors showing two examples each of CIMP-H, CIMP-L, and CIMP-neg. m, methylated; um, unmethylated. (b) Graphical summary of genetic and epigenetic findings in small bowel cancer stratified by the presence of MSI-H, CIN, and MACS.

Full size image

BRAF Mutations

For analysis of BRAF mutations, the following primers were used: BRAF forward 5′-CCTAAACTCTTCATAATGCTTGCTC-3′, BRAF reverse 5′-CCACAAAATGGATCCAGACA-3′. PCR conditions were as follows: denaturing 92 °C (45 s), annealing at 60 °C (1 min), extension 72 °C (1 min); 35 cycles. PCR products were analyzed by single-strand conformation polymorphism (SSCP) method. Tumors with aberrantly shifted bands were subjected to sequencing. The PCR fragments of these tumors were purified using the high Pure PCR Purification kit (Roche Diagnostics, Mannheim, Germany) as recommended by the manufacturer, subjected to direct cycle sequencing using a bigDye termination kit (Applied Biosystems), and automated sequencing by ABIPrism 377 (Applied Biosystems).

Results

Stratification of Small Bowel Adenocarcinomas According to Chromosomal and Microsatellite Instability Identifies Four Genomic Subgroups

Of 37 carcinomas, 22 (19 microsatellite stable, 3 MSI-H) displayed chromosomal imbalances. According to the microsatellite status and the presence of chromosomal imbalances, four groups of tumors could be distinguished.

Group 1, MSS/CIN

In all, 19 microsatellite stable (MSS) carcinomas displayed chromosomal imbalances. The most consistent finding was a deletion at chromosome 18q found in 11 of 19 carcinomas in this group. This alteration was commonly found in combination with additional chromosomal imbalances. Deletions at a lower frequency were detected for chromosomal arms 4q (5/19), 6q (4/19), 8p (4/19), and 17p (4/19) whereas gains were common at chromosomes 20q (8/19) and chromosome 7 (5/19).

Group 2, MSI-H/CIN

In all, three MSI-H carcinomas displayed chromosomal alterations. Two tumors showed gains at chromosome 8. No other recurrent imbalances were found. All tumors lacked deletions at 18q.

Group 3, MSI-H/CIN negative

Of nine MSI-H tumors, six did not show chromosomal alterations by CGH analysis.

Group 4, MSS/CIN negative

Of 28 MSS tumors, 9 did not show chromosomal alterations by CGH analysis. The results of CGH are summarized in Table 1.

Stratification of Small Bowel Adenocarcinomas According to Methylation Status: CIMP-H Is Associated with MSI-H and MACS

Among the 37 investigated carcinomas, 24 tumors were found to be methylated for at least one investigated locus. A total of 11 carcinomas were scored CIMP-H (3 CIN/MSS, 4 MSI-H, and 4 MACS), 13 carcinomas CIMP-L (4 MSI-H, 6 CIN/MSS, and 3 MACS), and 13 CIMP-neg (10 CIN/MSS, 1 MSI-H, and 2 MACS). CIMP-H was found in 15% (3/19) of CIN/MSS carcinomas, in 44% (4/9) of MSI-H, and in 44% (4/9) of MACS tumors. MINT2 and MINT1 represented the most commonly methylated loci (17 and 16 cases, respectively), followed by methylation at MGMT (14 cases), p16 (8 cases), and MLH1 (7 MSI-H and 1 MSS case). Whereas methylations at single loci were found for MGMT, MINT1, and MINT2, methylation at MLH1 was only seen in combination with at least one additionally methylated locus, and methylation of p16 was only observed in CIMP-H carcinomas.

BRAF Mutations Are Associated with MSI-H, MACS, and CIMP

BRAF mutations were observed in five carcinomas. In four cases, a codon 600 mutation (GTG to GAG; V600E) was found, and one additional case displayed a previously published mutation consisting of a 3-bp deletion (c1869–187135) predicted to result in a loss of W604 on the protein level. All tumors with BRAF mutations were CIMP positive, including three CIMP-H and two CIMP-L carcinomas. BRAF mutations were observed in 2 of 9 MSI-H carcinomas, in 2 of 9 MACS carcinomas, and in 1 of 19 CIN carcinomas.

Discussion

In this study, we aimed at elucidating the association of genetics, epigenetics, and mutations in KRAS/BRAF in small bowel adenocarcinomas. As a first step, we therefore stratified 37 carcinomas by the presence of genomic instability. Among 9 MSI-H and 28 MSS carcinomas investigated, we identified 3 of the MSI-H (33%) and 19 of the MSS carcinomas (68%) to display CIN. We thus found MSI-H to be inversely associated with CIN. Moreover, compared with MSS cancers with CIN, the extents and chromosomal locations of CIN in MSI-H cancers were different (Table 1), indicating a selective growth advantage of divergent chromosomal alterations in MSI-H and MSS cancers. These data are comparable with the situation in colorectal cancer, in which generally lower numbers and different types of chromosomal alterations are detected in MSI-H compared with MSS carcinomas.36, 37

MSS small bowel carcinomas with CIN presented the largest genetic subgroup (51%). Among the alterations seen in this group, deletions at chromosome 18q were the most common shared feature (58% in the CIN group, 30% of all investigated carcinomas) and were commonly associated with gains at chromosomes 20q and 7 as well as with losses of chromosomes 4q, 8p, and 17p (Table 1). Our finding of 30% 18q deletions among all investigated small bowel adenocarcinomas is in contrast with a previous matrix CGH study by Diosdado et al38 reporting <10% of 18q deletions in 48 small intestinal adenocarcinoma, whereas the frequencies of deletions at 4q, 8p, 17p, and gains at chromosomes 7 and 20 were similar to those of our study. The reason for this discrepancy is not clear. It may be explained by sampling variation, and in part result from the high number of celiac disease-associated carcinomas (15/48) included in the study by Diosdado et al38 (1/37 in our study), which have been shown to follow a somewhat different genetic pathway than nonceliac disease-related sporadic carcinomas.38, 39

Chromosome 18q harbors the SMAD4 gene, a mediator of TGFβ signaling, which was previously described as a mutational target for a variety of cancers, including small bowel adenocarcinoma.7, 40 The importance of SMAD4 inactivation for small bowel carcinogenesis is underlined by the fact that deletions at the SMAD4 locus were found in two carcinomas with only minimal numbers of other chromosomal imbalances (T401, T970), and that somatic intragenic mutations of SMAD4 were identified in two tumors lacking 18q loss (T246, T502; data not shown).

Although the majority of the carcinomas investigated in this study could be assigned to one type of genomic instability, a significant share of 24% of tumors were found microsatellite and chromosomally stable. This finding is similar to the situation in colorectal cancer, where MACS is identified in 15–30% of tumors.17, 41, 42, 43 The pathogenesis of MACS tumors is not understood. It is noteworthy, however, that we have previously identified44 large-scale N-terminal deletion mutations of β-catenin in three small bowel adenocarcinomas characterized as MACS in the present study, indicating that inactivation of DNA repair mechanisms other than those implicated in mismatch or chromosomal repair could be involved in MACS tumorigenesis.

To investigate the interaction of genomic instability types with epigenetics, we analyzed the tumors for CIMP. We observed 11 carcinomas displaying high-level CIMP (CIMP-H). These 11 carcinomas were not evenly distributed among CIN, MSI-H, and MACS cancers. Similar to the situation in colorectal cancer, CIMP-H with MLH1 methylation was particularly frequent in MSI-H carcinomas, providing evidence for a link of CIMP-H and MSI-H in small bowel adenocarcinomas also. Interestingly, we found that CIMP-H is similarly frequent in MACS and is less frequently present in MSS tumors with CIN. These data are in line with a previous observation in colorectal cancer reporting an inverse correlation of aberrant methylation and CIN.45, 46 BRAF mutations have been reported to associate with CIMP-H in colorectal cancer. Our data show that the situation is similar in small bowel cancer. We found BRAF mutations in three CIMP-H and two CIMP-L carcinomas. The BRAF mutation frequency was identical in MSI-H and MACS carcinomas (2/9 cases each), whereas only 1 of 19 CIN carcinomas was BRAF mutated, indicating that not only CIMP-H but also BRAF mutations preferentially occur in the absence of CIN. In contrast, mutations in KRAS were commonly found in MSS/CIN carcinomas (58%) and were infrequent in MSI-H and MACS carcinomas.

Our comprehensive analyses of genetic and epigenetic alterations show that small bowel adenocarcinomas can be stratified into different molecular subgroups. Our findings reveal similar principles of molecular tumorigenesis than recently reported in two comparable studies on colorectal cancer45, 46 (Figure 2), and indicate that CIMP is related to MSI- and MACS-associated intestinal carcinogenesis, whereas it is inversely correlated with CIN. According to our analysis, 50% of small bowel adenocarcinomas are CIN, frequently KRAS mutated, and usually show no or a low level of methylation. Among the remaining 50% of chromosomally stable tumors, MSI-H and MACS cancers had an equal share. MSI-H and MACS small bowel carcinomas usually lack KRAS mutations and display similar frequencies of high-level methylation and BRAF mutations, indicating common mechanisms to trigger MSI-H and MACS. Future analysis may show if the molecular subtypes of small bowel adenocarcinoma correlate with tumor biology and prognosis.

Figure 2
figure 2

Comparison of the distribution of CIN, MSI, and MACS in small intestinal adenocarcinoma and colorectal cancer. The results of this study and the average values of two similar studies on colorectal cancer45, 46 are presented. The frequency of CIMP-H in the individual genetic subtypes highlights the predominant occurrence of CIMP-H in MSI and MACS carcinomas. *The frequency of CIMP-H in MSI/CIN colorectal carcinomas could only be calculated for one study45 (0/3 cases).

Full size image