Modal variety of microsatellite instability in human endometrial carcinomas

Somatic instability of repetitive DNA sequences comprising minimal reiterative motifs, i.e. microsatellite instability (MSI), has initially been reported in tumours arising in Lynch syndrome [LS, alias hereditary non-polyposis colorectal cancer (HNPCC)] patients (Aaltonen et al. 1993), in which germline mutations in the genes functioning in DNA mismatch repair (MMR) are often found. MMR is an important cellular system that counteracts replication errors caused by DNA polymerases and, consequently, guarantees the high fidelity of DNA replication on the genome. Repetitive sequences such as microsatellites are particularly prone to replication errors, because DNA polymerases often slip on the repetitive sequences, and strand misalignment is formed. These replication errors, if uncorrected, are fixed during subsequent replication as addition or deletion of one or more repeat units. Thus, the phenomenon of MSI has been considered to reflect MMR deficiency in tumour cells. As MSI is frequently observed in various human neoplasms (Arzimanoglou et al. 1998), analyses of MSI have been prevalent in the fields of oncology or pathology. Numerous studies have been done on a wide variety of human malignancies and addressed the characteristics of MSI⁺ tumours.

The reported frequency for MSI⁺ tumours in each malignancy, however, differs widely in the literature. In order to manage the confusion raised in the field, the National Cancer Institute (NCI) sponsored the workshop, ‘Microsatellite Instability and RER Phenotypes in Cancer Detection and Familial Predisposition’ (Boland et al. 1998) in 1997, which concluded that the variety of microsatellites used was a major cause of discrepancies among data from various laboratories and, consequently, recommended a panel of five microsatellites as a ‘working reference panel’. In addition, the NCI workshop recommended that the MSI⁺ phenotype should be classified into two different grades, i.e. MSI-H (high) and MSI-L (low), according to the frequencies of changes in a defined set of microsatellite markers. However, the diversity of data in the literature has not improved since then. Analysis of MSI is now commonplace, but several methodological problems have in fact been pointed out in the conventional assay techniques (Maehara et al. 2001) and may also account for the variability in results. We previously established a unique fluorescent technique designated as high-resolution fluorescent microsatellite analysis (HRFMA), in which products of polymerase chain reaction (PCR) are precisely and quantitatively resolved (Oda et al. 1997). In this technique, (a) electrophoretic profiles of microsatellites PCR products are simplified enzymatically or by primer sequence modifications, (b) each DNA fragment is detected quantitatively by use of an automated DNA sequencer, and (c) two differently labelled PCR products derived from tumour (red) and the corresponding normal tissues (green) are co-electrophoresed (see Fig. 1), in order to exclude migration errors. Application of this technique has revealed a number of previously unrecognised aspects of MSI in human cancer. In particular, we found two qualitatively distinct patterns of microsatellite alterations, i.e. Type A and Type B (Oda et al. 2005). Although this distinction has not widely been discussed, our previous data suggest that different molecular abnormalities may underlie these two modes of MSI (Oda et al. 2005). It is widely known that tumours exhibiting the MSI-H phenotype form a distinct entity with unique clinicopathological and molecular characteristics, particularly in colorectal cancer (Jass et al. 2002), and accordingly, two mutually exclusive pathways are hypothesised in colorectal tumourigenesis (Lengauer et al. 1998). Several reports, however, suggest that this distinction might be an oversimplification (Goel et al. 2003; Hawkins et al. 2001; Jass et al. 2002; Young et al. 2001). Attention has been drawn to the classification of the MSI phenotypes in human cancer.

Microsatellite alterations are most frequently observed in tumours occurring in LS kindred. More than 90 % of LS tumours are MSI⁺ (Liu et al. 1996). In the sporadic setting, endometrial carcinoma (EC) is one of the most microsatellite-unstable human neoplasms. EC indeed occurs in LS patients and is the second most frequent malignancy in LS. Numerous data have been collected on the microsatellite alterations observed in EC, and similarly to colorectal cancer, the MSI-H phenotype is now regarded as characteristic of a subset of ECs (Yeramian et al. 2013). However, also in EC, the data accumulated in the literature are diverse (see Discussion). In addition, the Type A/B classification has not been applied in assessing MSI in ECs. Using our unique dual-colour fluorescent technique (Oda et al. 1997), we analysed microsatellite alterations in detail in a relatively large panel of tumours from EC patients and found that the two modes of MSI, i.e. Type A and Type B, are indeed observed in this malignancy. More importantly, the modes more closely correlated with the clinicopathological and molecular backgrounds of the tumours than the MSI-H/L grades, which suggests that, in addition to the established classification according to the frequencies of microsatellite changes, the modal classification may also be of potential use. Here, we report the modal variety of MSI in human endometrial cancer and its biological significance.

Numerous studies have been done on the microsatellite alterations observed in human ECs to date. Supplementary Table provides a summary of the literature. Studies of various sizes have been conducted on various subjects of different histological subtypes and genetic backgrounds, and the reported frequency for MSI⁺ tumours varies from 10 to 100 %, even if confined to the studies about endometrioid-type tumours in the sporadic setting. This variability in results may not be explicable merely from the variety of the subjects (Supplementary Table). As mentioned above, the reported frequency of MSI⁺ tumours in each malignancy is indeed diverse in the literature, and this confusion in the field has continued ever since. The variety of targets for analysis, i.e. microsatellites, will also undoubtedly lead to the variability in results. Although the NCI workshop in 1997 (Boland et al. 1998) recommended the ‘working reference panel’ comprising two mononucleotide and three dinucleotide microsatellites, various microsatellites, including tetranucleotide microsatellites, have in fact been used in the field (see Supplementary Table). Moreover, the MSI frequency differs widely even within the studies using only mononucleotide and dinucleotide microsatellites (Supplementary Table). We believe that, in addition to selection of targets for analysis, methodological problems at least partly account for the variability in results. Microsatellite length changes are sometimes as small as one or two base pairs in mononucleotide or dinucleotide microsatellites. Such small sequence alterations are not detectable in electrophoresis with significant migration errors, or under the influence of TDT activity of thermostable DNA polymerases such as Taq, which adds one additional base to PCR products in a sequence-dependent manner. In addition, cell populations carrying microsatellite alterations are not always predominant in a given sample. However, it is difficult to detect less abundant PCR products in an assay system using autoradiography or silver staining, due to their nonlinear detection characteristics. We previously established a sensitive fluorescent technique in which all of these methodological problems are overcome, and, in the present study, this technique was applied to address microsatellite alterations in ECs. The overall frequency of MSI⁺ tumours was 40 %, and several previous studies in the literature indeed reported similar frequencies (Furlan et al. 2006; Hardisson et al. 2003; Ohwada et al. 2002; Risinger et al. 2005) (Supplementary Table).

The MSI⁺ phenotype is now connected to specific clinicopathological characteristics of tumours. In colorectal cancer, MSI⁺ tumours more frequently occur in the proximal colon and often exhibit characteristic histopathological features such as poor/signet ring cell differentiation, mucin secretion and lymphocyte infiltration (Jass et al. 2002), which are currently known as ‘MSI-H histology’ (Umar et al. 2004). Patient outcomes are in general believed to be more favourable in MSI⁺ colorectal carcinomas (Popat et al. 2005). Accordingly, various clinicopathological features of tumours have been examined and are regarded as associated with the MSI⁺ phenotype in other human malignancies. In EC, tumour histology, grade, location, stage and patient survivals have been reported (Supplementary Table). These parameters did not correlate with MSI in this study (Tables 3, 4). The clinical and histopathological phenotypes of tumours are determined by highly complicated systems. Their relationships to MSI may therefore be more complex than hitherto suspected. MSI is now regarded as a molecular hallmark of LS. Indeed, more than 90% of LS tumours are MSI⁺ (Liu et al. 1996), which implies that MSI well reflects molecular abnormalities underlying LS tumourigenesis. One important finding of the present study is that Type B MSI did significantly correlate with family history of HNPCC-associated cancers, and that, on the other hand, MSI-H did not. This may be partly because the MSI-H phenotype sometimes includes Type A tumours (see Table 1) and inevitably tends to be heterogeneous in terms of the mode of microsatellite alterations. The tight connection between Type B MSI and LS tumourigenesis may suggest a biological importance of Type B alterations and, in addition, a potential advantage of the Type A/B classification. The MSI⁺ phenotype is also connected to various other genomic changes. Genetic instability observed in colorectal cancer has been regarded as deriving two mutually exclusive pathways, the chromosomal instability (CIN) pathway frequently associated with mutations in various oncogenes or tumour suppressors such as TP53 and the microsatellite instability (MIN) pathway, in which TP53 mutations are rare and, instead, mutations are found in genes harbouring mononucleotide repeats within their ORFs (Schwartz et al. 1999). The frequency for KRAS mutations in MSI⁺ tumours has been controversial. Mutations in the KRAS oncogene were initially regarded as infrequent in MSI⁺ tumours (Ionov et al. 1993; Salahshor et al. 1999; Samowitz et al. 2001). However, in our previous study, we have shown that KRAS mutations are indeed observed in microsatellite-unstable tumours and, more importantly, are relatively frequent in tumours exhibiting Type A MSI (Zhao et al. 2008). Also in this study, KRAS mutation was closely associated with Type A instability. We have previously demonstrated that Type A MSI is a direct consequence of defective MMR (Oda et al. 2005). A close association of Type A MSI with point mutations is highly consistent with the mutator phenotype in cells deficient in MMR (de Wind et al. 1995; Reitmair et al. 1997). These observations may thus suggest that the modal classification is biologically relevant.

Using HRFMA, we confirmed that endometrioid-type EC is the most microsatellite-unstable neoplasm among the human malignancies thus far tested. In general, the frequencies for MSI-H and MSI-L in colorectal cancer are 5–10 and 10–20 %, respectively. The MSI-H phenotype is very rare in the other human neoplasms, and the most frequently observed microsatellite-unstable phenotypes are generally MSI-L. Nevertheless, the MSI-H phenotype predominates in our panel of ECs (Table 1). The frequency for MSI-H exceeded 30%. The MSI-H phenotype is typical of LS (Liu et al. 1996). The high frequency of this phenotype in endometrioid-type EC may suggest that the majority of this subset of ECs arises from molecular backgrounds similar to those of LS or, in other words, via the MIN pathway. This hypothesis may be consistent with the consensus that mutations in TP53 are relatively infrequent in endometrioid-type EC (Yeramian et al. 2013), which are similarly regarded as rare in the MIN pathway of colorectal cancer (Ionov et al. 1993; Salahshor et al. 1999; Samowitz et al. 2001; Simms et al. 1998). Tumourigenesis in this pathway has not yet been well understood. Destabilisation of repetitive motifs may disrupt the genes harbouring repeats such as TGFBR2, IGF2R, BAX, CASP5 etc (Schwartz et al. 1999). On the other hand, mutations in established oncogenes and tumour suppressor genes and chromosomal instability leading to LOH in tumour suppressor loci are not hypothesised in the MIN pathway. Tumourigenesis without this classical model is still enigmatic. Type B MSI is observed in tumours exhibiting the MSI-H phenotype (Ikeda et al. 2001). This has been confirmed also in the present study (Table 2). Type B MSI indeed predominates in microsatellite-unstable ECs (Table 1). Our previous study suggests that defective MMR may be a promoting and, consequently, highly coincidental (as typical in Lynch syndrome patients), but insufficient factor for Type B changes, whereas being necessary and sufficient for Type A instability. In other words, whereas Type A MSI is a direct consequence of defective MMR, molecular abnormalities in addition to MMR deficiency may contribute to Type B alterations of microsatellites (Oda et al. 2005). Frequent Type B instability in endometrioid ECs suggests the possibility that this malignancy may provide a clue to Type B mechanisms and also serve as a good model for the MIN pathway tumourigenesis. Further analyses of the EC genome warrant particular attention.