Introduction

Lynch syndrome, also known as hereditary nonpolyposis colorectal cancer (HNPCC), is an autosomal dominant disorder associated with a germline mutation in one of the DNA mismatch repair (MMR) genes, most commonly MLH1 and MSH2 and less frequently MSH6 and PMS2. Lynch syndrome is characterized by a strongly increased risk of developing colorectal cancer and several extra-colonic malignancies including carcinomas of the endometrium, ovary, ureter, stomach and small intestine. Tumours develop at a relatively young age. Although the risks for some common types of cancer, for example breast cancer [1–3], or rarer tumour types, for example malignant fibrous histiocytoma (MFH) [4], do not appear to be significantly increased in Lynch syndrome, MMR deficiency in the presence of a corresponding germline defect has been demonstrated in incidental cases of these tumours.

Here we report a 44-year-old woman from a Lynch syndrome, Amsterdam positive family who was referred for DNA testing. She had a recent history of a colorectal adenoma and an undifferentiated carcinoma of her thyroid and was shown to carry the truncating MSH2 mutation that was known to segregate in her family. Traditionally, thyroid cancer is not considered to be part of the Lynch syndrome tumour spectrum. Our findings, however, suggest that this tumour was not coincidental, but likely developed in association with the underlying germline defect in the MSH2 gene. We reviewed the literature on unusual manifestations of inherited mismatch repair gene mutations.

Methods

After genetic counselling, DNA analysis of the MSH2 gene was performed in this 44-year-old woman by extracting DNA from lymphocytes, followed by a PCR amplification of exon 11 of the MSH2 gene. The PCR product was analyzed by denaturing gradient gel electrophoresis (DGGE) and compared with DNA from a family member carrying the mutation [5].

Immunohistochemical staining for MLH1, PMS2, MSH2 and MSH6 protein expression was performed on formalin-fixed, paraffin-embedded sections of tumour as described previously [6].

DNA was extracted from both tumour and normal tissue. Microsatellite instability analysis was performed on formalin-fixed, paraffin-embedded sections of tumour and corresponding normal tissue. Following DNA amplification using fluorescent labelled primers, a panel of five microsatellites recommended by the NCI [7] and consisting of BAT25, BAT26, D2S123, D5S346 and D17S250 was analyzed for allelic shift. The amplified PCR products were analyzed on an ABI Genetic Analyzer.

We searched the English literature through Entrez PubMed http://www.ncbi.nlm.nih.gov/sites/entrez using sets of keywords to identify publications on tumours reported in patients with germline mismatch repair gene mutations. The reference lists of publications found through this approach were searched for additional relevant papers.

Results

Genetic analysis of the MSH2 gene in the patient revealed the c.1704_1705delAG mutation, already known to segregate in her family. Her undifferentiated thyroid carcinoma showed complete loss of immunohi-stochemical expression of the MSH2 and MSH6 protein in the presence of normal positive internal controls, and no loss of the MLH1 and PMS2 protein. Of the five microsatellite markers tested, BAT26 showed instability. Therefore the thyroid tumour was classified as MSI-low.

The acknowledged Lynch syndrome tumour spectrum is shown in Table 1. The cumulative risks and average ages of diagnosis shown in this table were retrieved from papers reporting these data in proven mutation carriers [8–17], rather than from papers that had included untested patients and/or first-degree relatives in their analyses. The reports on 'unusual' tumours in Lynch syndrome patients are presented in Table 2. For comparison, we list the tumour spectrum associated with bi-allelic MMR gene germline mutations in Table 3.

Table 1 Lynch syndrome tumour spectrum. Cumulative risks and average ages at diagnosis
Full size table
Table 2 Unusual tumours in patients with Lynch syndrome
Full size table
Table 3 Tumours observed in patients with bi-allelic MMR gene germline mutations
Full size table

Discussion

Undifferentiated thyroid carcinoma is not commonly associated with Lynch syndrome. In our patient the immunohistochemical loss of expression for the MSH2 and MSH6 protein suggested that this tumour was not coincidental, but due to the underlying mutation in the MSH2 gene. Loss of MSH6 expression in tumours is often observed in case of germline MSH2 mutations and can be explained by loss of its stabilizing partner MSH2. Broaddus et al. [18] contended that for both an adrenal and a thyroid carcinoma an MSH2 gene mutation was causally linked because the tumour showed loss of MSH2 protein with immunohistochemical staining, but retained expression of MLH1. This staining pattern was similar to that seen in the more common Lynch syndrome related malignancies in these families. Although both adrenal and thyroid carcinoma showed loss of MSH2 immunohistochemical expression, neither tumour was microsatellite instable (MSI-high). Loss of protein expression in the absence of MSI has been observed before in Lynch syndrome, most notably in patients with MSH6 mutations [6, 19].

In the past, the Lynch syndrome tumour spectrum has primarily been defined through an epidemiological and statistical approach. From a clinical point of view this approach is of course still very valid as many clinicians will be primarily interested in tumours that have a significantly increased risk of developing in their patients. Cumulative cancer risks for Lynch syndrome were usually based on retrospective cohort analysis of families meeting the Amsterdam criteria, often including families without proven mutations and untested first-degree relatives. More recently studies have focused on proven mutation carriers only. The risk figures listed in Table 1 are based on the latter type of studies [8–17]. Interestingly, the risk for gastric, ovarian, ureter/renal pyelum and brain tumours appears to be higher for carriers of MSH2 mutations than for carriers of MLH1 mutations. In addition to the statistical approach, the tumour spectrum can be broadened through analysis of tumours occurring in MMR gene mutation carriers. Again, patients with atypical Lynch syndrome tumours as listed in Table 2 more often have been reported to carry an MSH2 than an MLH1 mutation. Also a wider range of tumours is observed for MSH2 than for MLH1 in these patients. At this point we can only speculate on the reason for these differences. MLH1 and MSH2 each create a heterodimer with different partners and have different roles in the detection and repair of DNA mismatches. For each of these protein complexes, deficiency might have a different impact on types and quantity of mismatches left unrepaired and the effect deficiency has on different target genes. The absence of MSH6 and PMS2 mutations in Table 2 might simply be caused by the fact that these mutations have been less frequently observed in Lynch syndrome in general. Ascertainment bias, however, cannot be excluded as laboratories did not test MSH6 and PMS2 in their analyses of Lynch syndrome suspected patients until fairly recently. Nevertheless, the absence of MSH6 and PMS2 from the listed reports might also reflect a true difference in associated tumour spectrum.

The tumours listed in Table 2 are not known to develop significantly more frequently in MMR gene mutation carriers than in the general population. Loss of MMR function may or may not have contributed significantly to tumour development in these particular cases. Generally, in these organs loss of the wild type allele in MMR gene mutation carriers and/or subsequently the accumulation of clinically important unrepaired mutations in cancer-associated target genes are apparently relatively rare. It is interesting to look at the types of cancer that develop in patients who have inherited bi-allelic MMR gene mutations (Tabele 3.). These patients are born with a mismatch repair deficiency and can present with tumours that rarely occur in carriers of single allele MMR gene mutations who need to lose their WT allele in their tissues first. Several studies have demonstrated that these bi-allelic mutations can lead to a phenotypically distinct recessive syndrome with predominantly childhood onset brain tumours, leukaemia and lymphoma, bowel tumours and endometrial carcinoma [20–44]. A striking feature of these patients is that nearly all of them display some features, spotty hyperpigmentation of the skin and Lisch nodules of the irides, usually observed in neurofibromatosis type I. Some of the reported tumour types, sarcoma, NHL and early-onset breast cancer match the types incidentally reported in patients with single allele MMR gene mutations, which further supports the notion that these tumour types could be causally linked to inherited MMR gene mutations.

Whether or not MMR deficiency contributed significantly to development of the types of cancer occasionally seen in Lynch syndrome patients remains to be determined. From a practical point of view, we conclude that unusual tumours in Lynch syndrome can show loss of immunohistochemical staining that corresponds to the MMR germline mutation. Therefore these tumours, especially of those types that rarely occur in the general population, could be useful when trying to predict MMR gene mutations in Lynch syndrome suspected families for mutation analysis [6, 19] if the typical Lynch syndrome-associated tumours are unavailable.