In a recent report published in the European Journal of Human Genetics, Pastrello et al1 elegantly demonstrated that in individuals with Lynch syndrome, large germline deletions of MSH2, which span exon 5 are likely to somatically delete the commonly used microsatellite instability (MSI) marker BAT26 from the remaining wild-type allele. As a consequence, BAT26 appears to be stable when assessed in tumour tissue. This work has highlighted a limitation of using BAT26 alone for the detection of MMR deficiency, a limitation of considerable magnitude given that large deletions of MSH2, of which up to one-half span exon 5,2, 3 comprise 15% of all known deleterious MMR mutations.
Several panels of MSI markers have been proposed, and have demonstrated excellent clinical utility.4, 5, 6 Tumours showing instability in 30–40% or more of these microsatellite markers are designated MSI-H (high-level MSI) and this phenotype is tightly associated with Lynch syndrome in early onset cancers. The microsatellite marker, BAT26, has shown particularly high sensitivity and specificity in the detection of an MSI phenotype in colorectal tumours, and it has been suggested that this marker alone could be used to screen for Lynch syndrome.4, 7, 8, 9 BAT26 is located immediately downstream of exon 5 in the MSH2 gene, and consists of a poly-A repeat of an invariant number of nucleotides, which is particularly susceptible to deletion-type replication errors in states of MMR deficiency. Its apparent, though not total, lack of polymorphism10 also makes it an especially useful MSI marker as normal reference DNA need not be tested alongside that from an individual tumour. Stability at BAT26, however, has been previously reported in Lynch syndrome tumours, occurring in three tumours with absent immunostaining for MSH2.11 As large intragenic deletions of MSH2 can account for up to one-third of its mutations,12 this finding may in part be owing to deletions that remove BAT26.
We investigated the stability of BAT26 in a series of 108 confirmed germline MMR mutation carriers from the Australasian Colorectal Cancer Family Study of which 6 (5%) were derived from a population-based subset, and the remainder from high-risk colorectal cancer family clinics. Of these 108 cases, 55 (51%) carried mutations in MSH2 and 53 (49%) in MLH1. All cases had undergone tumour MSI analysis and immunohistochemistry for MMR genes as described previously.13 Fifteen of 55 (27%) MSH2 mutation carriers were found to have large exonic deletions, seven of which spanned exon 5. Of these seven, four showed stability at BAT26 in their colorectal tumours. In the remaining 48 cases, where exon 5 was not deleted, only six were stable at BAT26 (Figure 1). Therefore, the rate of BAT26 stability in cases with large deletions spanning exon 5 (4/7 or 57%) was significantly higher than in all other types of MSH2 mutation (6/48 or 12.5%) (P=0.016, Fisher's exact test). These findings are consistent with the data of Pastrello et al,1 and suggest that a finding of BAT26 stability in a tumour that demonstrates no MSH2 protein is suggestive of a large genomic deletion encompassing exon 5. Assessment of the prevalence of BAT26 stability in MSH2-mutation carriers other than those who carried exon 5 deletions (6/48 or 12.5%) showed only borderline statistical significance (P=0.051, Fisher's exact test) when tested against that of MLH1-mutation carriers (1/53 or 1%). In our current data set, the average number of unstable markers per tumour (81.5%) in MLH1-deficient tumours from Lynch syndrome cases was significantly higher than that seen in MSH2-deficient tumours (69.9%) (P=0.0030; two-tailed t-test). Although this observation remains significant when cases with exon 5 deletions are removed (P=0.018), this does not remain so if all cases of BAT26 stability are removed (P=0.11), indicating that BAT26 stability in MSH2-deficient tumours is the likely reason for the difference in overall stability.
We further investigated the association between BAT26 stability and MSH2 immunodeficiency in a series of colorectal cancer cases referred to the Mayo Clinic Molecular Genetics Laboratory over a 5-year period (2001–2005), for MMR testing. These cases generally represented a moderate to high-risk group for having HNPCC, owing to young age of onset, presence of a family history of colon cancer or HNPCC-related malignancy, or the presence of a tumour's histology suggestive of defective MMR. Both MSI and IHC were used to define the presence of defective DNA MMR in tumour specimens, following established methodology.13 Of the 1724 colon cancer cases with unequivocal test results, 469 (26%) had a tumour phenotype of MSI-H. Of these, 242 showed a loss of MLH1, 122 MSH2/MSH6, 27 MSH6 alone, 15 PMS2 alone and 14 showed normal expression for all four of these proteins (Table 1). The remaining 49 cases had either missing or equivocal IHC data, or different and less frequent combinations of protein loss. The frequency of stability with the mononucleotide repeats BAT26 varied considerably across these five different groups. Stability at this marker was detected in less than 1% of the MLH1-deficient tumours. On the other hand, BAT26 stability was detected in 6.6% of the MSH2-deficient cases (P=0.0031, Fisher's exact test). Although it was not possible to screen this series of tumours for germline mutations in MMR genes owing to the unavailability of corresponding blood samples, the results suggest that BAT26 stability occurs more commonly in the carriers of germline MSH2 gene mutations, at least a proportion of which are likely to be large deletions that span the BAT26 repeat.
These findings suggest that although BAT26 may be a good marker of MSI in MLH1 mutation carriers, it is not so for those with MSH2 mutation. It is also not surprising that BAT26 has been shown in prior studies to be a good marker for MSI with very few false negative results, as the majority of sporadic cases of colon cancer with defective MMR result from promoter hypermethylation of MLH1. When screening higher risk patient groups, however, the abnormal gene distribution is quite different with nearly half representing other MMR gene combinations, and approximately 70% of the non-MLH1 cases are associated with MSH2 immunodeficiency. As a result, it is clear that BAT26 cannot be used alone to identify defective MMR (or the MSI-H tumour phenotype) in high-risk patient groups.
In this report, we have confirmed the findings of Pastrello et al, who reported an association between BAT26 stability and MSH2 exon 5 deletion. We have also demonstrated additional instances where BAT26 stability occurs in MSI-H colorectal tumours outside the setting of MSH2 exon 5 deletion. In all but one case where mutation testing had been completed, BAT26 stability was associated with a germline mutation in MSH2. The explanation for these findings is unclear. However, we have previously reported that the average number of markers that show instability is higher in MLH1 deficient tumours of sporadic origin11 than in those from Lynch syndrome families, when calculated over an entire panel of mononucleotides and higher-order repeats. Our current data set extends this finding to show that MLH1-deficient tumours from Lynch syndrome cases show greater instability than MSH2-deficient tumours, and that this difference is largely owing to the differences in frequency of BAT26 stability between the two groups. It is possible that with lower average levels of MSI, the likelihood that BAT26 will escape instability is slightly increased in MSH2-deficient tumours, although the reason for this is unknown. Alternatively, the lower level of instability in MSH2-deficient tumours may reflect differences in the developmental ‘age’ of the tumour at presentation. In summary, from a fully characterized series of 108 MMR mutation carriers, we demonstrate that 18% (10/55) of MSH2 mutation carriers showed stability at BAT26 in their tumours, a significant proportion of which were due to large deletions of MSH2 involving exon 5. Further, we have shown in an independent tumour series from a moderate- to high-risk patient subset that BAT26 stability was more likely to be seen in tumours that showed immunodeficiency of MSH2. Taken together, our findings support those of Pastrello et al (2006) and suggest that screening tumours for Lynch syndrome with BAT26 only may be ill advised.
We thank the many families who have participated in research programs, for their participation has facilitated this study. This work was supported by the National Cancer Institute, National Institutes of Health, under RFA CA-95-011 and through cooperative agreements with the members of the Colon Cancer Family Registry and principal investigators. Collaborating centers for this work include the Australasian Colorectal Cancer Family Registry ( UO1 CA097735), and Mayo Clinic Co-operative Family Registry for Colon Cancer Studies (UO1 CA074800).
About this article
British Journal of Dermatology (2007)