Introduction

Parallel with improvements in diagnostics, e.g. increased awareness of hereditary colorectal cancer, application of mismatch-repair (MMR) protein immunostaining that pinpoints the affected gene, and refined mutation detection techniques, an increasing number of families with hereditary nonpolyposis colorectal cancer (HNPCC) have been identified. MMR gene mutations were linked to HNPCC, or Lynch syndrome, in the early 1990s [1–4]. Fifteen years later, about 600 unique variants have now been described in each of MLH1 and MSH2, and about 200 variants have been identified in MSH6 [5]. Most studies are limited in size and population-based data are scarce. Several MMR gene mutation databases have been established, among which the database currently run by the International Society of Hereditary Gastrointestinal Cancer, http://www.insight-group.org, was the first to which MMR gene alterations could be reported and made publicly available. Its second update in 2004 reported 448 pathogenic mutations, which affected MLH1 in 50%, MSH2 in 39%, and MSH6 in 7% [6]. The more recent Canadian database, http://www.med.mun.ca/MMRvariants, is based on published MMR gene alterations and reveals alterations in MLH1 in 39%, MSH2 in 40%, MSH6 in 16%, and PMS2 in 6% of the published cases [5].

The Danes are of Gothic-Germany ancestry and have inhabited Denmark since prehistoric times. The current population size is 5.5 million. A small German-speaking minority lives in southern Jutland, an Inuit population inhabits Greenland, and the Faroe Islands have a Nordic population, but these minority groups account for only about 120,000 individuals. Immigration is low with 97% of the population being Danish with minor contributions from Scandinavian, Inuit, Faroese, German, Turkish, Iranian, and Somali ethnic groups. We report the population-based Danish MMR gene mutation profile with several novel MMR gene alterations, a significant contribution from mutations in MSH6, and a large impact from recurrent mutations in MLH1 and MSH2.

Patients and methods

The Danish HNPCC-register, which was established in 1991, constitutes a national resource that registers individuals with suspected or verified hereditary colorectal cancer on a population basis. Family histories are obtained through a standardized questionnaire, and as far as possible all diagnoses are verified in clinical records, pathological reports or death certificates, which provides the basis for family classification, risk estimates, and recommended surveillance programmes. All families have been offered genetic counselling and, when relevant, genetic diagnostics. The families are classified according to the Amsterdam I/II criteria, as HNPCC likely (families fulfilling the Amsterdam criteria except for confirmed histology or lack of first degree relations, and families with two colorectal cancers with one individual diagnosed before age 50), as late onset families (with three verified colorectal cancers in two generations, and all individuals diagnosed ≥50 years of age), and as moderate risk families according to the Danish colorectal cancer group (families in which an individual has been diagnosed with colorectal cancer before age 50 and families with two first degree relatives with colorectal cancer irrespective of age) [7]. By October 31, 2007, 1,058 families in Denmark had, based on risk estimates, been recommended participation in colorectal cancer surveillance programmes with 228 families fulfilling the Amsterdam I/II criteria, 298 being HNPCC likely, 187 late onset, and 342 having weaker heredity according to the Danish colorectal cancer group.

Mutation analysis was performed at the Departments of Biochemistry in Aalborg, Århus, and Copenhagen. Genomic DNA was extracted from EDTA blood samples according to standard procedures. The mutation analyses were primarily based on full genomic sequencing using ABI 377 or ABI 3130 DNA sequencers (Applied Biosystems, Foster City, CA) followed by analysis for large intragenic deletions using multiplex ligation-dependent probe amplification (MRC-Holland, Amsterdam, the Netherlands) according to the manufacturers’ instructions. All deleterious mutations were classified as pathogenic as were also missense alterations that had been reported to segregate with disease and/or judged to be pathogenic based on loss of MMR function. If not previously recognized as polymorphisms, the remaining missense mutations were regarded as unclassified variants. Although some of the latter have in functional studies demonstrated reduced MMR function, there is in our opinion yet insufficient data to classify these variants as pathogenic.

Mutations that have been reported in at least four families from different research groups (http://www.med.mun.ca/MMRvariants) were considered recurrent. Among these, a smaller number of mutations have been defined as founder mutations through studies that have demonstrated shared haplotypes, determined the age of the mutation, and suggested a common geographical ancestry. Mutations that were identified in ≥4 apparently unrelated families in Denmark were considered recurrent in our series. Data from the Danish person registry and the Danish parish registers were in these families used to establish the geographical origin of the family. The place of birth for the first reported family member affected by HNPCC-associated cancer was used, and this knowledge did in most families date back to the nineteenth century.

Results

Totally 138 alterations (previously described polymorphisms were not included) were identified, of which 58 affected MSH2, 39 MLH1, and 41 MSH6 (Tables 1, 2) [8, 9]. The alterations differed between the genes with exon deletions and splice site defects most commonly (17% and 14% of the alterations) found in MSH2, and missense mutations most frequently found in MLH1 (36%) and MSH6 (49%) (Fig. 1). Of the 138 alterations, 50 were regarded as unclassified variants and were found in 76 families. Totally 88 mutations were considered pathogenic, of which 43 affected MSH2, 26 MLH1 and 19 MSH6 (Table 1). Small insertions/deletions predominated and caused 47% of the alterations, followed by nonsense mutations, splice site alterations, and exon deletions, which contributed about equally (14–17% of the alterations), whereas missense mutations had the lowest (7%) contribution. The pathogenic mutations were found in 164 families; with mutations in MSH2 in 79 (48%) families, MLH1 in 58 (35%), and MSH6 mutations in 27 (16%) families. Clinically, 92 families fulfilled the Amsterdam I/II criteria, 69 were classified as HNPCC likely, one was a late onset family, and the remaining two families showed weaker degrees of heredity.

Table 1 Pathogenetic mutations in the Danish Lynch syndrome cohort
Full size table
Table 2 Alterations of unknown clinical significance
Full size table
Fig. 1
figure 1

The distribution of the different types of genetic alterations in MSH2, MLH1, and MSH6

Full size image

Of the 88 disease-predisposing mutations, 69 (78%) were found in a single Danish family. When mutations previously identified in at least four other families internationally were considered, six mutations in MLH1 and 17 in MSH2 accounted for 75/137 (55%) MLH1/MSH2 mutant families. Mutations that were recurrently observed (in 4–15 families) in the Danish population included four mutations in MSH2, three in MLH1, and two in MSH6. These mutations were traced back to the geographical area of the first known affected family member (Fig. 2). In MLH1, the frequently occurring mutations c.350C>T and c.1852_1854delAAG were found in nine and six geographically spread families, respectively. The MLH1 Danish founder mutation c.1667+2_1667_+8TAAATCAdelinsATTT, has, to the best of our knowledge, not been described in other populations, it was identified in 14 families from central Sealand (Fig. 2) and thus accounts for 14/58 (24%) families with mutations in MLH1. In MSH2, four recurrent mutations, a deletion of exons 1–2 (c.1-?_366?del), c.942+3A>T, c.1165C>T, and c.1786_1788delAAT, were identified and have previously been described in European Lynch syndrome families. In Denmark the former two mutations were geographically spread, whereas the latter two were found in families from Jutland (Fig. 2). In MSH6, two recurrent mutations, c.1444C>T and the splice site mutation c.4001+2T>C, were found in four families each.

Fig. 2
figure 2

Geographical origin of Lynch syndrome families with mutations identified in at least four apparently unrelated families in Denmark

Full size image

Discussion

This population-based update of the Danish Lynch syndrome population represents one of the largest reports from one of the smallest European countries. Compared to compiled data from e.g. the InSiGHT database (http://www.insight-group.org) and the database by Woods et al. (http://www.med.mun.ca/mmrvariants) the Danish population reveals a larger contribution from mutations in MSH6, which constitutes 22% of the pathogenic mutations and 44% of the variants of unknown pathogenicity. Current estimates suggest that MSH6 causes 5–7% of HNPCC, which is considerably lower than the 16% of the MSH6 mutant families in the Danish cohort [6, 10]. This high frequency may still represent an underestimate since MSH6 is associated with a less distinctive phenotype with higher age at onset, miscrosatellite-low or microsatellite-stable tumours, and reduced penetrance and may therefore escape recognition using current guidelines [7, 11]. The mutations linked to HNPCC are widely dispersed within the MMR genes, as was also observed in the Danish cohort. Mutation clustering has been suggested in exons 15–16 of MLH1 and 6/26 (23%) MLH1 mutations occurred in this region [5, 12, 13]. Insertions/deletions predominated and caused 47% of the disease-predisposing mutations. Nonsense mutations contributed about equally (8–10%) in all three genes. Exon deletions are estimated to cause 17–18% of the pathogenic germline mutations and caused 16% of the mutations in the Danish cohort with the highest frequency in MSH2 and the lowest in MSH6 (Fig. 1) [14–16].

Only 7% of the pathogenic mutations were missense alterations, but this may be an underestimate since a multitude of unclassified variants await segregation analysis and functional studies for classification as disease-predisposing mutations or normal sequence variants. The missense mutations c.350C>T and c.1733A>G in MLH1, the Askenazi jewish founder mutation c.1906G>C in MSH2 and the c.2090G>T in MSH2 have been identified in multiple HNPCC families, have convincingly been shown to segregate with disease, and to lead to loss of or reduced MMR protein function [17–21]. The MLH1, c.2059C>T, has been found to segregate with disease in a Danish family (LOD score 1.5, data not shown) and to lead to loss of immunohistochemical MLH1 expression in tumours from two family members. The MLH1, c.1942C>T, has demonstrated reduced MMR capacity and has in a homozygous state been linked to HNPCC [22]. These alterations are thus considered disease-predisposing and are used for genetic testing. Among the three MMR genes, MSH6 showed the largest frequency (49%) of missense mutations, all of which are currently considered to represent unclassified variants.

A number of recurrent/founder MMR gene mutations have been reported in Lynch syndrome families from North America, Asia, and Europe [12, 23–27]. The current view suggests that genetic hot-spot mutations and founder mutations in the MMR genes have a smaller impact than e.g. within breast and ovarian cancer, where founder mutations that account for some 10–20% of the mutations in BRCA1 and BRCA2 have been identified. This notion may, however, change with an increasing number of MMR gene mutations identified, although population-based studies are needed to recognize their importance. The largest impact from founder mutations in Lynch syndrome has been described from Finland and from the Ashkenazi jewish population. Finland has a homogenous and genetically distinct population with influences from Eurasia and Siberia, in which two founder mutations in MLH1 account for more than half of the families [28]. In the Ashkenazi jewish population the MLH1 c.1906G>C, with an estimated origin in Eastern Europe between 1440 and 1715, accounts for about 1/3 of the Lynch syndrome families [17, 29]. Such strong founder effects are likely to be confined to homogenous populations, but our observation that 23 recurrent mutations cause 55% of the MLH1/MSH2 mutant families and that the Danish MLH1 founder mutation explains 24% of the MLH1 mutant families demonstrates that a limited number of recurrent alterations have a large impact in the Danish population.

The MSH2 c.942+3A>T, which has been found in numerous families world-wide, represents a founder mutation in Newfoundland, and contributes to a substantial fraction of the pathogenic MSH2 mutations in e.g. England and Germany [24, 25]. This mutation has, however, also been demonstrated to occur de novo at a high frequency, which is compatible with the observation of this mutation in 15 geographically spread families in Denmark (Fig. 2) [23]. The MSH2 c.1786_1788delAAT was identified in 11 families from western Jutland, the MSH2 c.1165C>T in four families from eastern Jutland, and a deletion of exons 1–2 in MSH2 was found in four geographically spread families (Fig. 2). Recurrent mutations in MLH1 included the missense mutation c.350C>T and the in-frame deletion c.1852_1854delAAG, which were found in nine and six geographically spread families, respectively, and the c.1732–2A>T in three families from Jutland. The Danish founder mutation MLH1 c.1667+2_1667_+8TAAATCAdelinsATTT has previously been reported in five Danish families, but has not been detected in other populations [30]. It has now been found in 14 families, all of which originate from central Sealand (Fig. 2). Several exon deletions have recurrently been identified with e.g. the deletions of exon 16 and exons 17–19 in MLH1 recognized in the UK and in Sweden and the deletion of exons 1–6 in MSH2 as an American founder mutation [16, 27, 31, 32]. A deletion of exons 1–2 in MSH2 was found in four families from northern Jutland and southern Sealand. In MSH6, the nonsense mutation c.1444C>T was found in four families from western Sealand and Fyn and has also been identified in a Dutch family [33] and the splice site mutation c.4001+2C>T was found three families from Sealand and Fyn and in one family from Sweden, where is has also been identified (Nilbert and Okkels, unpublished observations).

In summary, the Danish HNPCC/Lynch syndrome population currently consists of 164 individuals/families affected by 88 pathogenic MMR gene mutations and an additional 76 families carrying 50 unclassified MMR gene variants. Among the disease-predisposing mutations, MSH6 plays a significant role and accounts for 22% of the mutations and 16% of the Lynch syndrome families. In MLH1 and MSH2, mutations previously reported by several groups were found in more than half of the families, and the Danish MLH1 founder mutation caused 24% of the MLH1 mutant families. This suggests that directed analysis of recurrent/founder mutations may be considered when full mutation screening is not feasible, e.g. in cases where only archival tissue is available for genetic diagnostics.