Background

Colorectal cancer represents 12% of the cancer cases in Sweden with about 5% of the patients being below 50 years of age [1]. Rectal cancer constitutes about 1/3 of the large bowel cancers and 5% of the patients are younger than 50 years of age at diagnosis. Tumor development at young age may be an indicator of hereditary cancer, such as hereditary nonpolyposis colorectal cancer (HNPCC), which is characterized by development of colorectal cancer at a mean age of 43 years and an increased risk also for cancer of the endometrium, ovaries, urothelium and the small intestine [2]. HNPCC has in population-based studies been demonstrated to account for about 2% of colorectal cancer, but has been estimated to cause about 1/3 of the colorectal cancer cases occurring at young age [3, 4]. HNPCC is caused by germline mutations in one of several mismatch-repair (MMR) genes, most commonly in the genes MLH1 and MSH2. Studies of colorectal cancers from young patients have revealed MSI frequencies between 17 and 58% (compared to about 15% in sporadic colorectal cancer), with the higher frequencies found in patients below age 35 at diagnosis [5–10].

Colorectal cancer cells evade growth control through a series of genetic changes, which have been recognized to affect distinctive pathways. The chromosomal instability (CIN) pathway is involved in about 70–80% of the tumors and is characterized by aneuploidy, multiple chromosomal rearrangements, and an accumulation of somatic mutations in oncogenes such as KRAS and β-catenin and tumor suppressor genes such as TP53 and the adenomatous polyposis coli (APC) gene [11]. The microsatellite instability (MSI) pathway characterizes about 15% of the colorectal cancers and is associated with diploid or near-diploid tumors and a defective MMR machinery, which is reflected as MSI in the tumor tissue. MSI tumors are more likely to be right-sided with mucinous histology, poor differentiation and tumor-infiltrating lymphocytes [12]. MSI is only infrequently caused by germline HNPCC-causing MMR gene mutations, but is more commonly due to somatic hypermethylation of the MLH1 promoter [13].

Differences in aetiology, clinical behaviour and histopathology between colon cancer and rectal cancer suggest different preferred tumorigenic pathways in these tumor types. The few molecular genetic studies that have separately analyzed these clinical entities demonstrate a lower frequency of MSI and KRAS mutations and a higher frequency of aneuploid tumors, p53 mutation/overexpression and β-catenin overexpression in rectal cancer. These findings are likely to reflect that CIN and the TP53 pathways contribute more than MSI to rectal cancer development [3, 14–18]. However, the importance of MSI versus CIN in rectal cancer development in young patients has to our knowledge not been characterized. We have therefore studied 30 rectal cancers from patients below 50 years of age, an age limit chosen in accordance with the Amsterdam criteria for the classification of HNPCC [2] with respect to flow cytometric DNA content, MSI, KRAS mutations and immunohistochemical expression of the MMR proteins MLH1 and MSH2 and the tumor suppressor proteins p53 and β-catenin. The aim was to clarify possible differences between colon cancer and rectal cancer in young patients as regards the contribution of the MSI and the CIN pathways to tumor formation and to estimate the proportion of cases attributable to defective MMR.

Methods

Patients

Thirty patients diagnosed with histologically verified adenocarcinomas of the rectum before age 50 were recruited for the study (Table 1). Tumor stage was Dukes' A in 6 patients, B in 6, C in 12 and D in 6 patients. Tumor location was low (<5 cm from the anal verge) in 5 patients, middle rectum (5–10 cm) in 13, and high (10–15 cm) in 12 patients. Tumor histopathology was reviewed by a gastrointestinal pathologist (B.H.) and differentiation was classified as moderate in 23 tumors and mucinous /signet ring cell in 5 tumors, as well differentiated in 1 tumor and as poor in 1 tumor. The patients had a mean age of 42 (range 25–49) years and the material included 9 women and 21 men. Family histories of cancer were available from 28 of the 30 patients; 20 had no history of familial colorectal and/or endometrial cancer, 6 had one or more relatives with colorectal and/or endometrial cancer at any age and 2 patients fulfilled the Amsterdam II criteria for the classification of HNPCC, whereas data on family history was not available from 2 patients (Table 1) [2]. Adjuvant treatment with 5-fluorouracil (5-FU) and leucovorin (in one case combined with oxaliplatin) was given to 16 patients. Median follow-up for the survivors was 46 (range 19–85) months. Six patients had metastasis at diagnosis and 9 patients died from tumor median 21 (range 3–40) months after diagnosis (Table 1). Paraffin embedded tumor specimens were retrieved, and were sectioned and stained with Hematoxylin & Eosin to verify that representative tumor tissue was present in the sample. Three 10 μM paraffin sections were used for DNA extraction (as described below) and 3 μM sections were used for immunohistochemical staining. Permission for the study was granted by the Lund university ethics committee.

Table 1 Summary of clinicopathological data
Full size table

DNA-extraction, PCR amplification and DNA sequencing

DNA was extracted from three 10-μM sections of paraffin embedded tumor tissue using proteinase-K incubation for 2–12 hours at 65°C. After incubation, the samples were boiled for 10 min to inactivate the enzyme, quickly centrifuged and 1 μL was used for the PCR amplifications. PCR was performed using AmpliTaq Gould (Perkin Elmer – Roche, Branchburg, NJ, USA) and the reagents supplied according to the PCR programme: initial denaturation (94°C, 7 min) followed by 35 cycles of denaturation (94°C, 15 sec), annealing (45–50°C, 15 sec) and extension (72°C, 15 sec) whereafter a final elongation step at 72°C for 10 min was added followed by cooling at 4°C. All primer sequences and exact PCR conditions are available from the authors upon request. DNA sequencing was performed on an Applied Biosystems 373 sequencer using cycle sequencing with AmpliTaq polymerase FS and the reagents supplied by Applied Biosystems (Foster City, CA, USA).

MSI-analysis

MSI was determined using the mononucleotide markers BAT25, BAT26, BAT34C4, BAT40 and the dinucleotide markers D2S123 and D17S250. All these MSI markers are among those recommended in the NCI reference panel for the classification of MSI [19]. All primer sequences are available from the authors upon request. Fluorescently labelled forward primers were used for the PCR reactions: Tet (green) for BAT25, Fam (blue) for BAT26, BAT34C4 and D2S123 and Hex (yellow) for BAT40 and D17S250. The PCR products were mixed with formamide and GeneScan 500 TAMRA size standard (Applied Biosystems), denatured and separated in Performance Optimized Polymer-4 (POP-4™) on a ABI PRISM™ 310 Genetic Analyzer (Applied Biosystems). The tumors were classified as MSI-high if at least 2/5 markers displayed MSI, as MSI-low if instability was present in 1/5 markers and as MSS if none of the 5 markers showed instability.

CTNNB1mutations

Exon 3 of the CTNNB1 gene, which codes for β-catenin, was analysed using DNA sequencing and a PCR-based assay designed to detect larger deletions of this region. Exon 3 was sequenced using the primers 5'-TGTAAAACGACGGCCAGTCGTATTTATAGCTGATTTGA-3' (forward) and 5'-TACTCTTACCAGCTACTTGT-3' (reverse). Genomic deletions involving exon 3 were analyzed by PCR amplifiction using the primers 5'-CCAGCGTGGACAATGGCTAC-3' (forward) and 5'-TGAGCTCGAGTCATTGCATAC-3' (reverse), followed by size separation in acrylamide gels.

KRASmutations

KRAS was analyzed for point mutations in codons 12, 13 and 61 by sequencing. PCR amplification was performed in two fragments using the primer pairs K1 (covering codons 12–13) 5'-GGCCTGCTGAAAATGACTGA-3' (forward) and 5'-GTCCTGCACCAGTAATATGC-3' (reverse) and K2 (covering codon 61) 5'-TTCCTACAGGAAGCAAGTAG-3' (forward) and 5'-CACAAAGAAAGCCCTCCCCA-3' (reverse).

Immunohistochemistry for TP53, β-catenin, MLH1 and MSH2

Fresh 5-μM sections from paraffin embedded tumor blocks were transferred to glass slides (DAKO ChemMate capillary gap microscope slides, 75 μM, DAKO A/S, Denmark). Following deparaffinisation in xylene and rehydration through graded ethanol, the sections were microwave treated at 750 W in citrate or EDTA buffer (Table 2) for 15 min to achieve antigen retrieval. Staining was performed in an automated immunostainer (TechMate™ 500 Plus, DAKO A/S, Denmark) using the biotin-streptavidin-peroxidase procedure with diaminobenzidine as the chromogen. Endogenous peroxidase activity was blocked by incubation in H2O2. All antibodies used were commercially available mouse monoclonal IgG's (Table 2). After counterstaining with hematoxylin the slides were dehydrated in ascending concentrations of alcohol and mounted. The immunohistochemical staining (Fig. 1) was independently evaluated by two of the investigators (E.F. and M.N.). The staining for MLH1 and MSH2 was nuclear and was classified as present or absent in the tumor tissue, and retained expression was in all tumors identified in surrounding normal crypts and in the tumor stroma. Nuclear p53 staining was evaluated in 5 high power fields (HPFs) (x40) and was classified as positive if >5% of the tumor cell nuclei were stained. Immunohistochemical staining for β-catenin normally results in a membraneous and/or a weak cytoplasmic staining, whereas a mutation leading to β-catenin stabilization is reflected as an increased cytoplasmic or nuclear staining. We classified the predominant β-catenin staining pattern in 5 HPFs according to cellular localization as membraneous, cytoplasmic or nuclear.

Figure 1
figure 1

Immunohistochemical staining of tumor 10 a) normal nuclear staining for MLH1 in the tumor as well as in stromal tissue, b) loss of nuclear expression of MSH2 in tumor tissue but retained staining in stromal cells, c) nuclear staining for p53, and d) membraneous and intense cytoplasmic staining for β-catenin.

Full size image
Table 2 Antibodies and conditions used for immunohistochemistry
Full size table

DNA flow cytometry

50-μM sections from paraffin embedded tumor tissue were used for flow cytometric DNA analysis. The nuclear suspension was stained with propidium iodide for the DNA quantification in an Ortho-Cytoron Absolute flow cytometer (Ortho Diagnostics, Raritan, NJ, USA). Up to 20,000 nuclei were analyzed per sample, which created a histogram profile with 256 channel resolution. Chicken and trout red blood cells were used as internal controls. A cell population was defined as a G0/G1 peak with a corresponding G2 peak. Tumors with a single G0/G1 peak were classified as DNA diploid (DNA-index of 1.0). If more than one cell population was present the sample was classified as aneudiploid. The degree of DNA aneuploidy was expressed as the DNA index (DI) which represents the position of the aneuploid peak in relation to the diploid reference cells. ModFitLT 3.0 software from Verity Software House Inc. (Topsham, ME, USA) was used for the analysis of the DNA histograms.

Results

The results of the molecular genetic characterization of the tumors are presented in Table 3. In summary, aneuploidy was found in 21/30 tumors (7 of which were near-diploid with a DI of 1.1–1.2) whereas 9 tumors (including the 3 MSI tumors) were diploid. No tumor displayed any mutation involving exon 3 of CTNNB1, but immunohistochemical staining for β-catenin revealed a membraneous staining in 5 tumors, a cytoplasmic staining in 19 tumors and a nuclear staining for β-catenin in 6 tumors. p53 was immunohistochemically expressed in 18/30 tumors; strong wide-spread nuclear staining was found in 16 tumors and a weaker staining in >5% of the tumor cells in 2 tumors. KRAS mutations were identified in 6 patients (20%); all mutations were missense mutations, 5 mutations occurred in codon 12 and one in codon 13, whereas no mutation was detected in codon 61. All 6 tumors with KRAS mutations displayed immunohistochemical p53 protein expression. MSI was found in 3 tumors and the remaining 27 tumors were MSS. All 30 tumors displayed normal expression of MLH1, whereas loss of expression for MSH2 was detected in 3 tumors. A family history of colorectal cancer (in 2 patients in combination with endometrial cancer) was reported in 8/28 evaluable patients, 2 of which also fulfilled the Amsterdam II criteria for the diagnosis of HNPCC.

Table 3 Summary of molecular genetic alterations
Full size table

Discussion

We characterized 30 rectal cancers from patients younger than 50 years at diagnosis with respect to tumor biologic phenomena associated with the MSI and the CIN pathways for colorectal tumor development. Rectal cancer has been associated with a higher frequency of aneuploidy, MSS tumors, mutations in KRAS and TP53 and nuclear accumulation of β-catenin compared to colon cancer, which probably reflects different preferred pathways in the development of these tumor types [3, 12, 14–18]. Our findings of high frequencies of MSS (27/30) tumors, aneuploidy (21/30), immunohistochemical p53 expression (18/30) and increased β-catenin expression (25/30) (Table 3) is in line with previous reports of genetic alterations in rectal cancer, but support such a model for the development of rectal cancer also in the majority of young patients.

Of the 30 rectal cancers investigated herein, 3 showed MSI, which is higher than the 1–2% MSI detected in larger series of rectal cancers, but lower than the 15% MSI tumors observed among sporadic colorectal cancers. This finding indicates that the frequency of MSI in young patients with rectal cancer is considerably lower than the 17–58% MSI tumors and 20–40% HNPCC-causing mutations reported in mixed series of colorectal cancers from young patients [5–10, 14, 18]. All three MSI tumors identified in our study displayed immunohistochemical loss of MSH2 expression. Whereas immunohistochemical loss of MLH1 may reflect either a germline mutation or somatic hypermethylation of the MLH1 promoter, loss of MSH2 expression is almost exclusively caused by a germline mutation of the MSH2 gene [12, 19, 20]. Thus, hypermethylation of MLH1 plays a limited role in rectal cancers from young patients, which is in accordance with an association between MLH1 hypermethylation and proximal tumors, older patients and female sex [13]. Consequently, the finding of MSI in a rectal cancer is, together with a family history of the disease, a strong indicator of HNPCC. Although HNPCC-associated tumors show a prediliction for the proximal colon about 40% of HNPCC-associated tumors will develop in the distal colon or within the rectum.

Mutations in codons 12 and 13 of KRAS are found in about 1/3 of colorectal tumors [21, 22]. These mutations develop early in colorectal tumor progression [23]. The prognostic value of KRAS mutations is somewhat unclear, but a recent study compiling data from over 3000 patients with colorectal cancer, did in a multivariate analysis demonstrate a significant impact on survival by the codon 12 glycine → alanine mutation, which is found in about 9% of the tumors [24]. We identified KRAS mutations in 6/30 rectal cancers (Table 3). All mutations were heterozygous missense mutations of types that have previously been described, and one patient displayed the codon 12 glycine → alanine change. A lower frequency of KRAS mutations has been found in distal tumors and an inverse relationship between MSI and alterations of KRAS and TP53 mutations has been described [22], a finding that is supported by our data with KRAS mutations in 6/30 tumors and nuclear expression of p53 in 18/30 rectal cancers.

β-catenin is involved in cell-cell adhesion as well as in signal transduction. Free β-catenin is bound to the adenomatosis polyposis coli (APC) protein and to glycogen synthase kinase 3B (GSK-3B), which phosphorylate serine/threonine residues in β-catenin and thereby destabilizes the protein, which is degraded by the ubiquitin-proteasome pathway. Mutations of APC or of the regulatory domain of CTNNB1 has been demonstrated in colorectal cancer and lead to stabilization and cytoplasmic accumulation of free β-catenin [25, 26]. Increased cytoplasmic levels of β-catenin allows the protein to enter the nucleus where it promotes transcriptional activation through the TCF/LEF transcription factors. We found no mutations in the regulatory domain (exon 3) of the CTNNB1 gene, but immunohistochemitry for β-catenin revealed an increased cytoplasmic expression or a nuclear accumulation in 25/30 tumors. Nuclear accumulation of β-catenin has been reported to occur at a higher frequency in rectal cancer than in colon cancer, whereas mutations affecting the CTNNB1 gene, have been reported to occur at a higher frequency in MMR-deficient tumors [26, 27].

Conclusions

We conclude that rectal cancer, also when developing in young patients, is frequently associated with DNA-aneuploidy, increased expression of p53 and β-catenin, whereas MSI and mutations in KRAS occur at a low frequency.

In this study only 10% of the rectal cancers from young patients were MSI-high, but all of these tumors showed loss of expression for MSH2, which suggests an underlying HNPCC-causing mutation. Thus, other mechanisms than defective MMR and HNPCC probably cause rectal cancer development in the majority of young patients.