Patients and methods

PATIENTS

Eighty-three patients with early onset colorectal cancer (age <50 years) were analysed for tumour RERs (microsatellite instability (MSI)) using a combination of mono- and dinucleotide microsatellite markers. Patients were ascertained randomly through a population based survey of colorectal cancer in eastern England and the results of MSI for dinucleotide markers on the first 33 patients aged 45 years or less have been reported previously.34

MICROSATELLITE INSTABILITY ANALYSIS OF COLORECTAL CANCERS

For each case (n=83), paraffin embedded tumour and normal tissue sections were obtained and analysed for microsatellite instability with CA dinucleotide microsatellite markers as described previously.35 In addition, MSI was assessed with two mononucleotide repeat markers chosen from BAT25, BAT26, and BAT40. Tumours were considered as showing conventional dinucleotide repeat instability (d-MSI) if two or more of five dinucleotide repeats tested showed instability and were considered as showing mononucleotide repeat instability also if one or more of two mononucleotide repeats showed instability. Tumours showing mononucleotide but not dinucleotide microsatellite instability were designated m-MSI+. We have used these two definitions of microsatellite instability to test the hypothesis that there might exist a class of tumours showing only mononucleotide repeat instability. Because all of the tumours designated d-MSI+, but only some of the tumours designated m-MSI+, would fall under the recently published definition of MSI-H,35 we have generally not used the MSI-H designation in this study to avoid confusion.

DETECTION OF MUTATION IN MSH6

Mutations in hMSH6 were detected by amplifying individual exons and flanking intron sequences by PCR using the primers devised from the sequence of thehMSH6 genomic locus (Genbank accession number U73732-U73737). Primer details are available on request. PCR set up was performed with a Tecan Genesis 100 Robotic Workstation and PCR was performed on Perkin Elmer 9600 and 9700 PCR instruments. The PCR products were sequenced on an ABI 377 sequencer (Perkin Elmer/Applied Biosystems) with standard M13 forward and reverse primers and Big Dye terminator chemistry using reagents obtained from Perkin Elmer/Applied Biosystems. In some cases it was necessary to resequence PCR products using primers that prime on internal regions of a PCR product. Analysis of the sequence chromatograms was carried out using Sequencher (Gene Codes Corp, Ann Arbor, MI) as well as by visually examining printed chromatograms to detect sequence changes and heterozygous nucleotides.

Results

YOUNG ONSET COLORECTAL CANCER: FREQUENCY AND CHARACTERISTICS OF TUMOUR MICROSATELLITE INSTABILITY

Eighty-three patients with colorectal cancer diagnosed before the age of 50 years were ascertained from a population based study and analysed for tumour MSI with dinucleotide microsatellite markers. An Amsterdam criteria family history of HNPCC was present in two patients. Seven of 83 (8.4%) patients had d-MSI+ tumours (defined as instability at two or more of five dinucleotide markers). This group with classical MSI comprised the two cases with HNPCC (Amsterdam criteria), one patient with a non-HNPCC family history, and four isolated cases.MLH1 and MSH2mutation analysis was performed in the seven cases with d-MSI+ tumours and germline MLH1 mutations were identified in the two HNPCC cases only.

Sixty-one of 76 cancers without d-MSI were analysed for mononucleotide repeat instability at two mononucleotide markers chosen from BAT25, BAT26, and BAT40. Microsatellite instability was detected in at least one of two mononucleotide markers (m-MSI+) in seven of 61 cases (11%). Four of these m-MSI+ cases were sporadic cases, while the remaining three came from patients whose families contained two first degree relatives with colorectal cancer but did not satisfy the Amsterdam criteria for HNPCC.

The clinicopathological features of d-MSI+, m-MSI+ only, and MSI- tumours were compared. There were no differences in mean (SD) tumour size (4.8 cm (1.6), 5.5 cm (2.3), and 4.2 cm (1.6) respectively), histological differentiation, or lymph node involvement. However, d-MSI tumours were significantly more frequently right sided in location than those without d-MSI+ (5/7 v 15/60). Tumours with m-MSI+ phenotype differed from d-MSI+ tumours by being mainly left sided (6/7 v 2/7). We postulated that m-MSI+ tumours might have MSH6 mutations.

To enable accurate mutation analysis in patients with m-MSI+ tumours, we developed a set of primer pairs for PCR amplification and sequencing of the 10 exons and flanking intron sequences ofhMSH6. Sequencing of the completeMSH6 gene coding region was performed in the five m-MSI+ cases for whom peripheral blood DNA was available (two familial and three sporadic). A number of putative polymorphisms were characterised (table 1). These included a number of silent changes, one that caused a Glu to Gly substitution at codon 39 (exon 1), and a number of substitutions, insertions, and deletions in introns. In cases where these changes were suspected to affect either the function or expression of MSH6, their allele frequency was determined by analysing DNA from 20 normal controls. In all cases, the detected sequence change was sufficiently common to be considered a polymorphism. However, in one m-MSI+ case, we detected a heterozygous germline nonsense mutation (C642G, Tyr214Ter) (fig 1). Translation of the mutant sequence would be predicted to result in the synthesis of a truncated protein containing only the 213 N-terminal amino acids of the 1260 amino acid long MSH6 protein. Functional studies inSaccharomyces cerevisiae have shown that this truncated protein would be inactive in mismatch repair (unpublished data). The proband presented at the age of 43 years with a left sided, moderately differentiated, 5 cm diameter adenocarcinoma of the large bowel. He was an only child with no family history of cancer. Attempts to determine if there was loss of the wild typeMSH6 allele in tumour tissues were uninformative owing to insufficient quantity and quality of tumour DNA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Detection of a germline MSH6 mutation by sequence analysis. The upper panel shows the sequencing chromatogram from the region of exon 4 containing codon 214 from a wild type control DNA. The DNA and predicted protein sequence are shown above the chromatogram. The lower panel shows the sequencing chromatogram from the same region of a patient DNA sample containing a heterozygous nucleotide at nucleotide 642 of the cDNA sequence. The mutant DNA sequence and predicted mutant protein sequence are indicated above the chromatogram.

Discussion

The analysis of tumours for MSI using standard microsatellite markers provides a screening strategy for targeting patients with possible germline MLH1 orMSH2 mutations. Thus, the overwhelming majority of tumours in HNPCC kindreds withMSH2 or MLH1mutations show MSI. However, the specificity of MSI analysis is reduced by phenocopy tumours in which both alleles ofMSH2 or MLH1 were inactivated by somatic events. Early age at cancer diagnosis is a characteristic feature of inherited cancer susceptibility and a high frequency of MSI and germlineMSH2/MLH1mutations has been described in patients with very early onset (<35 years) colorectal cancer.36 However, the incidence of MSI and germlineMSH2/MLH1mutations is low in moderately young cases (35-49 years)34 36 37 suggesting the possible involvement of other genes in these cases. Germline MSH6mutations have not been reported in classical HNPCC kindreds, but the presence of a MSH6 mutation in a Japanese HNPCC family which did not meet Amsterdam criteria led to a suggestion that the phenotype of germline MSH6mutations might differ from those of MSH2 orMLH1 mutations.33 Thus there were no cases of early onset (<50 years) colorectal cancer in the Japanese family and extraintestinal (endometrium, ovary, and pancreas) cancers were prominent. A further germlineMSH6 mutation was also reported from Japan in a 62 year old man with rectal cancer and two colonic adenomas, a weak non-HNPCC family history of gastrointestinal tumours, and MSI in the tumours.32 These results are similar to those obtained with mutant mice where MSH6 mutant mice developed cancer at later ages than eitherMSH2 or MLH1mutant mice.31 38 39

Although it seems unlikely that the MSH6truncating mutation we identified could not be pathogenic, this possibility should be considered. However, we note that expression of an N-terminal fragment of yeast MSH6 (that closely corresponds to what would be produced by the human mutant gene we describe) fails to complement the MSH6mismatch repair defect in in vitro studies (Das Gupta and Kolodner, unpublished data). Long term follow up of the proband and his children may provide further information on the pathogenicity and penetrance of this mutation. The finding of a germlineMSH6 mutation in an isolated case of young onset colorectal cancer further expands the phenotype ofMSH6 mutations and should prompt further studies of the role of these mutations in early onset colorectal cancer. Recently, Shin et al 40 analysed Korean HNPCC families and 45 sporadic early onset (<40 years) cases for mutations of the mononucleotide repeat in the MSH6gene. They reported frameshift alterations in two of 44 suspected HNPCC cases in whom MSH2 orMLH1 mutations had not been detected, suggesting that MSH6 mutations might be found in early onset cases. In this study we targeted these cases with m-MSI for MSH6 mutation. Although this approach is logical, further research is required to determine if germline MSH6 mutations can be associated with d-MSI or MSI negative tumours. The results ofMSH6 mice knockout studies would suggest that the latter possibility could occur. A germlineMSH6 mutation was only found in one of the five m-MSI+ tumours analysed. Although this finding clearly shows the potential significance of m-MSI, the four cases withoutMSH6 mutations could result from somatic inactivation of MSH6 or failure to detect a germline mutation (despite sequencing of the completeMSH6 coding region). Alternatively, germline or somatic mutations in other mismatch repair genes could be implicated in this group; in familial colorectal cancer we found a single m-MSI+ tumour in a patient with a germline MSH2mutation (unpublished observation). Our findings and a report of germline MSH6 mutations in Dutch familial colorectal cancer families41 indicate a need for further studies of tumour MSI and germline MSH6mutation analysis in familial and isolated colorectal cancers.