A growing number of disease-predisposing genes are identified and contribute to the complex hereditary colorectal cancer landscape (1). An identifiable cause of cancer predisposition can be demonstrated in 5% of colorectal cancer. Lynch syndrome is the most common hereditary colorectal cancer subtype with an estimated incidence of 1/1,200-1/660 (2). Germline mismatch-repair (MMR) gene mutations give rise to two phenotypic syndromes, i.e., the autosomal dominant, adult-onset Lynch syndrome and the recessive, childhood-onset constitutional mismatch repair deficiency (CMMRD) syndrome (3). Worldwide, more than 1,300 disease-predisposing MMR gene sequence variants have been reported (4). The estimated contribution from the different MMR genes to Lynch syndrome is ~50% MLH1, ~40% MSH2 (5), 7–20% MSH6 (5–8) and <5% PMS2 (9). Mutations in the EPCAM gene, located upstream of MSH2, represent an additional cause that is estimated to contribute to 1–3% of the disease-predisposing mutations (10–12). Founder effects, i.e., mutations that are overrepresented within a geographically or ethically isolated population, have been described in several populations, such as in the different Scandinavian populations (8,13,14) and in the Ashkenazi Jewish population (15–18).

Lynch syndrome is a multi-tumor syndrome and although the highest risks of cancer apply to colorectal, endometrial, ovarian and urinary tract cancer, a number of less common tumor types, such as cancer of the small bowel, brain tumors and skin tumors, have been linked to the syndrome (19). Different disease characteristics have been ascribed to mutations in the different MMR genes with a predominance of colorectal cancer in MLH1 and PMS2 mutation carriers, a high risk of extracolonic cancer in MSH2 mutation carriers and a high risk of gynecologic cancer in MSH6 mutation carriers. Compared to MLH1 and MSH2 mutation carriers, a later age at onset and a reduced penetrance has been described in MSH6 (20,21) as well as in PMS2 mutation carriers (9,22). The overall life-time risk of cancer at age 70 is estimated to be 70% (23). Age at onset is on average 20 years earlier than sporadic tumors, although the different tumor types show characteristic peak ages and phenotypes are highly variable, also within Lynch syndrome families. Identification of individuals and families with Lynch syndrome is challenging since family history has suboptimal sensitivity and the syndrome includes a broad tumor spectrum and variable penetrance and age at onset. However, reflex testing for MMR status is increasingly applied in colorectal cancer and is also discussed for endometrial cancer and will increase the likelihood of identifying individuals at increased risk in the future (24).

After 20 years of molecular diagnostics for Lynch syndrome, we compiled mutation data from the Swedish Lynch syndrome population with the aim to define the mutation spectrum, clarify the contribution from the different MMR genes, identify potential founder mutations and contribute to the world-wide data on Lynch syndrome mutations.

In Sweden, general guidelines for referral of cases with suspected hereditary colorectal cancer to genetic counseling include families/individuals with three or more cases of colorectal cancer or other Lynch syndrome-associated tumors with one family member diagnosed before the age of 50 (in line with the Amsterdam criteria except for the requirement of two first-degree relatives) or a single case of colorectal cancer diagnosed before the age of 50. In addition, clinicians have referred families suspected of Lynch syndrome based on the development of Lynch syndrome-associated tumor types. Reflex testing for MMR defects in colorectal cancer has not been implemented in Swedish pathology laboratories. Targeted analysis for MMR status, typically using four-protein immunohistochemical MMR staining and/or analysis for microsatellite instability (MSI) were applied for pre-screening in most cases.

All individuals/families genetically tested and found to carry MMR gene alterations classified as disease-predisposing genetic variants or a variant of uncertain significance between January 1994 and December 2014 were eligible for the study. Mutation data were collected from the six laboratories and/or oncogenetic clinics at the University hospitals in Umeå, Uppsala, Stockholm, Linköping, Gothenburg and Lund, responsible for genetic diagnostics. The Ethics Committee at Karolinska Institutet approved the study, which followed the tenets of the Declaration of Helsinki. All patients provided oral or written informed consent for genetic diagnostics.

Genetic screening of the proband/affected family member was performed using mainly Sanger DNA sequencing or massive parallel sequencing and the analyses were combined with multiplex ligation-dependent probe amplification (MLPA, P003 and P072; MRC-Holland, Amsterdam, The Netherlands) for the detection of large deletions or duplications.

All variants reported were classified at the nucleotide and protein levels according to the Human Genome Variation Society (HGVS) nomenclature (25). As reference sequences NM_000249, NM_000251, NM_000179 and NM_000535 were used. All sequence variants were then adjusted to the classification used in the InSiGHT database (http://insight-group.org/variants/database/). Variants previously not described in the InSiGHT database were, whenever possible, classified according to the InSiGHT VIC rules (4). Frequency data for certain variants were obtained from the ExAc database using the Alamut software (Alamut Visual, v. 2.7, Interactive Biosoftware, Rouen, France). Variants with a classification of 1 (benign) or 2 (likely benign) are not included (4).

In Sweden, the Lynch syndrome cohort consisted of 369 families with disease-predisposing mutations. These families were found to carry mutations in MLH1 (n=149), MSH2 (n=132, including one family with a deletion of the EPCAM gene), MSH6 (n=67) and PMS2 (n=21) (Table I). The contributions from the different MMR genes were MLH1 40%, MSH2 36%, MSH6 18% and PMS2 6% (Fig. 1A). In total, 201 unique alterations were identified, including 48 missense sequence variants, 31 nonsense variants, 43 insertions/deletions, 35 splice site variants and 36 whole exon/exons deletions/duplications. Splice site alterations were the most common mutation type in MLH1, frameshift mutations predominated in MSH2 and MSH6 and missense variants were most frequent in PMS2 (Fig. 1B). Copy number variations, i.e., deletions or duplications of whole exon/exons, constituted 21% of the mutations in MLH1, 22% in MSH2 including EPCAM, 4% in MSH6 and 22% in PMS2 (Fig. 1B).

The Swedish Lynch syndrome sequence variant spectrum is broad with 133 of the 201 (66%) alterations being private, i.e., observed in a single family, 26% observed in 2–3 families, and 18 variants observed in ≥4 families (Table II). In relation to the different genes, private mutations accounted for 46/71 MLH1 variants, 49/76 MSH2 variants (including the EPCAM deletion), 31/45 MSH6 variants and 6/9 PMS2 variants. Of the 201 unique variants, 137 were present in the InSiGHT LOVD with a classification made by an expert panel for 136 of these variants (4) (http://insight-group.org/variants/database/). For the remaining 64 sequence variants, 31 could, based on the predicted protein consequence from the sequence alteration, be classified as class 3–5 according to the five tier system (4).

Alterations observed in 4 or more families (Table II), i.e., recurrent alterations, included the MLH1 sequence variations c.62C>T, c.131C>T, deletion of exon 6 (c.454-?_545+?del), c.546-2A>G, deletion of exon 11 (c.885-?_1038+?del), deletion of exon 16 (c.1732-?_1896+?del) and the c.2059C>T variation. These variants have previously been recognized in Lynch syndrome families and are classified as disease-predisposing. In MSH2, recurrent alterations included deletion of exons 1–6 (c.1-?_1076+?del), deletion of exons 1–8 (c.1-?_1386+?del), c.942+3A>T, deletion of exons 7–10 (c.1077-?_1661+?del) and the c.1786_1788del. These variants have previously been reported in Lynch syndrome families from different countries and are classified as disease-predisposing. The deletion of MSH2 exons 1–6 was the most common recurrent variant identified in a total of 10 families. MSH6 had a high number of private mutations with only the c.1346T>C, c.2780_ 81insA and the c.2931>G pathological variants identified in ≥4 families. In PMS2, the sequence variant of unknown significance c.2113G>A was the most common variant found in 7/21 families. The deletions of exons 2–9 (c.24-?_988+?del) and c.736_741delins11 were both identified in 4 families. All of the recurrent sequence variants in MHS6 and PMS2 have previously been reported.

No recurrent mutation suggestive of a Swedish founder mutation was identified. We did, however, recognize a contribution from other Scandinavian founder mutations in the Swedish population. The Finnish founder mutation MLH1 c.1732-?_1896+?del was found in 6% of the Swedish Lynch syndrome families and constituted 15% of the MLH1 families. The Danish founder mutation MLH1 c.1667+2_1667+8delinsATTT was observed in two families.

This study is the first compiled data on the Swedish Lynch syndrome cohort and demonstrates mutations in MLH1 in 40%, MSH2 in 36%, MSH6 in 18% and PMS2 in 6% of the families (Fig. 1A). The Swedish mutation spectrum is broad with a total of 201 different mutations of which 66% were private and 9% were classified as recurrent, i.e., found in ≥4 families (Table II). The contribution from the different MMR genes is in line with international reports, which are mainly based on Western populations (4). The predominant types of alterations in MSH2 and MSH6 were small insertions/deletions and in MLH1 splice site variants (Fig. 1B). Whole-exon deletions significantly contributed and accounted for 20–22% of the mutations in MLH1, MSH2 and PMS2, but were rare (4%) in MSH6 (Table I, Fig. 1B). Our data support evidence on a significant contribution from whole-exon deletions in MSH2 and PMS2, and demonstrate a higher rate of large deletions than previously reported in MLH1 (28,29). Of the 201 sequence variants reported, 137 are available in the InSiGHT database, whereas 64 have not previously been reported.

In Sweden, 80% of the population is of Swedish origin and 20% were either born in another country or born in Sweden by two parents from another country. Among non-Swedish ethnic groups, Finns represent the largest group and during recent decades Sweden has received immigrants from a large number of countries with particularly large contributions from Denmark, Norway, Germany, Chile, former Yugoslavia, Iran, Irak, Eritrea, Somalia and Syria. Strong founder effects have been reported in Finland where two MLH1 mutations account for 63% of the families with Lynch syndrome (13). The Finnish founder mutation MLH1 c.1732-?_1896+?del, which leads to deletion of MLH1 exon 16, was identified in 6% of our Lynch syndrome families and constituted 15% of the MLH1 families, which is in line with the Finnish ancestry in 5% of the Swedish population. Two families in Sweden carried the Danish founder mutation MLH1 c.1667+2_1667+8delinsATTT (8). Two of the most frequent mutations in the Swedish population, i.e., the MLH1 c.546-2A>G and MSH2 c.1-?_1076+?del (deletion of exons 1–6), have been described as founder mutations in the US (30,31). From the mid 1800's until the early 1920's, 1.5 million Swedes migrated to US and it is therefore plausible that this US founder mutation is of Swedish origin. Regarding the deletion of exons 1–6, the common haplotype found in the US was analyzed in two Swedish samples with the same mutation although the results cannot confirm a common ancestry (30). The MLH1 c.2059C>T pathogenic variant is also common in the Swedish population.

Several recurrent mutations identified in MSH2, e.g. the c.1-?_1076+?del, c.942+3A>T and c.1786_1788del have also been reported from Denmark and in Norway (8). Also several of the MSH6 mutations identified, such as c.1444C>T, c.1483C>T, c.2302_2304del, c.3647-2A>C, c.3991C>T and c.4001+2T>C have also been observed in several families from Norway and/or Denmark and these mutations may be of Scandinavian origin. In PMS2 the c.736_741delins11 mutations have been reported from Denmark and Norway and the c.2113G>A transition (class 3) has also been identified in families from Norway.

We did not detect any individuals with CMMRD in our cohort. Two families harbored more than one MMR gene variants. Both of these families did fulfill the Amsterdam criteria. One family of Arabic origin had a MLH1 c.1989G>A (class 4) variant that affects splicing and a concomitant MSH6 c.773T>C variant, which has not been reported in the ExAc database. Another family had a MSH6 c.1649del frameshift variant and a concomitant MSH2 c.1484C>T variant of unknown significance according to ClinVar. In these families, the MSH6 c.773T>C and the MSH2 c.1484C>T variants may represent benign variants.

Identification of individuals with Lynch syndrome is cost effective with significant positive effects on morbidity and mortality from colorectal cancer (32). In Sweden, Lynch syndrome diagnostics have traditionally been based on individual or physician suspicion of hereditary cancer in which case families have been referred for genetic counseling followed by genetic diagnostics. In total, 369 Lynch syndrome families have been identified. Assuming a carrier frequency in the lower range (1/1,200), at least 8,000 individuals would be estimated to be mutation carriers in the Swedish population of 9.8 million. Though the absolute number of mutation carriers in Sweden is not known, it can be estimated that no more than one-quarter of the mutation carriers have at present been identified. Comparison is also possible with our neighboring country Denmark where Lynch syndrome families are registered on a national basis. Denmark has, relative to the size of the population, identified an additional 60% of Lynch syndrome families (data not shown).

In summary, the Swedish Lynch syndrome cohort with 369 families carries 201 unique alterations, of which 64 have not been previously reported. The mutation spectrum shows the expected contribution from the different MMR genes, underscores the roles of MSH6 and PMS2, which caused 18% and 6% of the mutations in the families, respectively. The cohort reveals a higher contribution from large deletion in MLH1 than previously reported. An overlap with mutations identified in the other Nordic countries is identified and our data suggest that US founder mutations in MLH1 and MSH2 may be of Scandinavian origin.