Hereditary and somatic DNA mismatch repair gene mutations in sporadic endometrial carcinoma
- Robert B Chadwicka,
- Robert E Pyattb,
- Theodore H Niemannb,
- Samuel K Richardsa,
- Cheryl K Johnsona,
- Michael W Stevensa,
- Julie E Meeka,
- Heather Hampela,
- Thomas W Priorb,
- Albert de la Chapellea
- aDivision of Human Cancer Genetics, The Ohio State University, Department of Molecular Virology, Immunology, and Medical Genetics, and the Comprehensive Cancer Center, The Ohio State University, 420 West 12th Avenue, Columbus, Ohio 43210, USA, bDepartment of Pathology, The Ohio State University Medical Center, 121 Hamilton Hall, 1645 Neil Avenue, Columbus, Ohio 43210, USA
- Dr de la Chapelle,
Editor—Endometrial cancer (EC) is the second most common malignancy in the hereditary non-polyposis colorectal cancer (HNPCC) syndrome.1 In a recent large study, cumulative cancer incidences by the age of 70 in HNPCC mutation carriers were: colorectal 82%, endometrial 60%, gastric 13%, and ovarian 12%.2 Interestingly, in female mutation carriers the incidence of endometrial cancer (60%) exceeded that of colorectal cancer (CRC) (54%), as had been suggested earlier.2 3
Predisposition to HNPCC is the result of germline mutations in the mismatch repair genes.4 Detectable mutations in the two major genes, MLH1 andMSH2, account for some 3% of all colorectal cancers.5 One might therefore assume that a similar proportion of all endometrial cancer patients would have such mutations; however, in a number of studies addressing this question, extremely few germline mutations have been found. Summarising the studies by Katabuchi et al,6Kobayashi et al,7 Limet al,8 Gurinet al,9 and Kowalskiet al,10 only one germline mutation (in MLH1) was found in a total of 352 EC patients (0.3%). In these studies, mutations were sought in all patients whose tumours were microsatellite instability (MSI) positive.
Recent reports have suggested that MSH6might account for many endometrial cancers and that families with these mutations show atypical features of HNPCC with endometrial and ovarian cancers outnumbering colorectal cancers.11 12Additionally, MSI, a hallmark of HNPCC, was low in most tumours associated with MSH6 mutations or was preferentially shown by mononucleotide repeats rather than dinucleotide repeats.12-14 Previous studies have reported that 9-25% of sporadic endometrial cancers display microsatellite instability .7 9 15-18 In the majority of cases, this instability arises through hypermethylation of the MLH1promoter region.9 19-21 This epigenetic change results in reduced (or no) expression of theMLH1 transcript.22
This study was undertaken to revisit the issue of microsatellite instability and mismatch repair gene mutations in sporadic endometrial cancer. By initially studying tumour tissue, both germline and somatic mutations were evaluated in the MSH2,MLH1, and MSH6genes in a retrospective series of microsatellite stable and microsatellite unstable endometrial cancers.
Material and methods
All patients diagnosed with endometrial adenocarcinoma between October 1996 and February 1998 at the Ohio State University Hospital were considered retrospectively. Among these 85 patients, archival tissue was available from 74.
After appropriate investigational review board (IRB) approval, these 74 charts were reviewed and the tissue blocks recovered. Histological sections were made, stained with haematoxylin and eosin, and the histological diagnosis critically re-evaluated. Sections 50 μm thick were cut from regions of the tumour containing as high a proportion of tumour cells as possible (typically >50%). To obtain non-malignant tissue, sections were obtained either from tissue emanating from other organs, primarily lymph nodes, that were histologically cancer free, or alternatively sections were made from parts of the endometrial tissue that had no cancer cells. All materials were unlinked from their identifiers before being subjected to DNA extraction and genetic analyses.
Tissue sections were deparaffinised with two xylene washes. Rehydration was accomplished through 20 minute incubations in decreasing concentrations of alcohol (100%, 80%, 50%) at room temperature followed by an overnight incubation in double distilled water at 4°C. DNA was extracted by lysis of the tissue for 18 hours at 55°C with 1 mg/ml proteinase K in 400 μl of buffer (10 mmol/l Tris, 400 mmol/l NaCl, 2 mmol/l EDTA, and 0.7% sodium dodecyl sulphate, pH 8.2). Degraded proteins were precipitated with 2.5 volumes of saturated NaCl after centrifugation. DNA was precipitated with 2.5 volumes of 100% ethanol at –20°C, washed in 70% cold ethanol, then dissolved in 50 μl of TE buffer (10 mmol/l Tris and 1 mmol/l EDTA, pH 8.0).
Microsatellite sequences were amplified using the Bethesda panel.13 Owing to limited availability of normal tissue, tumour DNA was used for MSI determination without its corresponding normal DNA pair. Amplifications were done in 15 μl PCR reaction volumes using 1 μl of each 8 μmol/l primer (the 5′ primer is fluorescently labelled), 10 ng of genomic DNA, and 8 μl of Qiagen's HotStarTaq Master Mix. The thermal cycling profile was one cycle at 95°C for 12 minutes, followed by 45 cycles at 95°C for 10 seconds, 55°C for 15 seconds, and 72°C for 30 seconds, followed by one cycle at 72°C for 30 minutes, followed by a soak at 4°C. Respective PCR reactions for each marker were pooled together and loaded on to the PE3700 automated sequencer. Allele sizing and calling was done using Genotyper software (Applied Biosystems). For polymorphic markers D2S123, D5S346, and D17S250 samples were scored as MSI positive if more than two alleles were present. Since normal DNA was not available for comparison from all samples, it is possible that instances of MSI where the normal sample was homozygous and the tumour sample had only two alleles were missed. However, this limitation does not apply to the two mononucleotide repeat markers, suggesting that this did not lead to any serious underestimate of MSI, as these markers are more sensitive to MSI than the dinucleotide repeats.23 24 For homozygous markers BAT25 and BAT26, samples were scored as MSI positive if the pattern deviated from the normal homozygous pattern. The BAT markers are polymorphic in African Americans.25 26However, the ethnicity of the subjects in this study was known and all three African American subjects (Nos 31, 65, and 70) were screened for mutations in the mismatch repair genes. Out of the 74 tumours tested, 17 (or 23%) were found to be MSI positive, 14 MSI high and 3 MSI low (table 1).
All exons of the MSH2,MLH1, and MSH6genes were screened by direct sequencing of genomic PCR products. In order to facilitate direct sequencing of PCR products for mutational analysis, all 5′ and 3′ PCR primers were tailed with M13 forward (TGTAAAACGACGGCCAGT) and M13 reverse (CAGGAAACAGCTATGACC) sequences (table 2 and supplemental data).27 PCR reactions were done in 25 μl volumes with 100 nmol/l of each of the respective PCR primers, 25 ng of genomic DNA, 100 μmol/l of each dNTP, 1.0 U AmpliTaq Gold DNA polymerase (Perkin-Elmer), 10 mmol/l pH 8.3 Tris-HCl, 50 mmol/l KCl, and 2 mmol/l MgCl2. PCR fragments were purified using the Exonuclease I/Shrimp Alkaline Phosphatase PCR Product Presequencing Kit (USB). After purification according to the manufacturer's protocol, 2 μl of the PCR products were sequenced using the BigDye Terminator AmpliTaq FS Cycle Sequencing Kit (Applied Biosystems). Determination of somatic or hereditary mutation status for all mutations was done by comparing tumour chromatograms to normal DNA chromatograms amplified from archival lymph node tissue DNA from the respective patients.
Out of the 17 MSI positive endometrial tumours and an additional 25 that were MSI negative, only two germline changes were found (table 1 and supplemental data). These were a GAG(Glu) to GGG(Gly) mutation at codon 578 of the MLH1 gene in patient 3 and a GGC(Gly) to GAC(Asp) at codon 322 of theMSH2 gene in patient 52. TheMLH1 mutation has been reported to be pathogenic previously and the tumour from this person was MSI high.28 Additionally, a functional assay indicates that this mutation is pathogenic.29 Also, MLH1 immunohistochemistry showed virtually no expression of MLH1 in this patient's tumour (supplemental data). The patient is a 74 year old white female with a history of hypertension, diabetes mellitus type II since the age of 54, and chronic renal failure. It was noted that her family history was positive for diabetes mellitus and hypertension, but there was no mention of a family history of cancer on chart review.
The GGC(Gly) to GAC(Asp) at codon 322 mutation ofMSH2 found in patient 52 is listed in the ICG-HNPCC database (http://www.nfdht.nl/) as both a pathogenic mutation and a polymorphism.30-32 In the first paper describing this change it was considered a clinically insignificant polymorphism because it was found in one out of 30 unrelated controls.30 Moreover, in another study the same change was reported not to segregate with the cancer predisposition.33 To investigate its incidence further, we tested 50 grandparents from the families collected by the Centre d'Etude du Polymorphisme Humain and found it in one person. The glycine is conserved among human, mouse, rat, and yeast. MSH2 immunohistochemistry of this tumour showed reduced expression of MSH2 suggesting that this amino acid change may potentially contribute to pathogenicity (supplemental data). Notably, this tumour was MSI high and had an additional somatic truncating mutation in exon 4 ofMSH2 (fig 1). The MSH2 antibody is specific for the carboxy terminus of MSH2 and thus less protein would be expected to be detected in a tumour with a truncating mutation in the amino terminus. Additionally, the tumour had a somatic mutation of GAT(Asp) to AAT(Asn) at codon 203 ofMLH1. Immunohistochemistry using anti-human MLH1 antibody also showed reduced expression of MLH1 protein (supplemental data). The patient is a 49 year old white woman with a history of ulcerative colitis since the age of 20. The family history is significant for her mother with breast cancer alive at the age of 71, her father with bladder cancer who died at the age of 79, and her paternal grandfather who died from colon cancer in his late 60s. Thus, in summary, the evidence regarding the MSH2 germline amino acid substitution is inconclusive in that it may be either an innocuous polymorphism or a low penetrant pathogenic mutation. For these reasons we do not count it as a pathogenic germline mutation in this study. No pathogenic germline mutations were found in theMSH6 gene in the 42 endometrial cancers studied.
Somatic truncating frameshift mutations were found in the coding repeat of seven adenosines in exon 4 of MSH2 (fig1). These 1 bp deletions lead to a predicted truncation at amino acid 245 of the MSH2 protein and both tumours were MSI high. This region ofMSH2 has not been reported previously to be hypermutable. Mutations in this repeat were confined to MSI positive cancers, indicating that MSH2 is also a target in the microsatellite instability model of carcinogenesis similar to the TGFBRII, BAX, IGF2R, MSH3, MSH6, TCF4, MBD4, and RIZgenes.34-41 We are unable to evaluate the clinical and pathogenic significance of mutations that arise in mismatch repair genes in tumours that are already mismatch repair deficient. A case in point is tumour 52 that showed immunohistochemical deficiency of both MLH1 and MSH2 protein, no MLH1 germline mutation, but two genetic changes in MSH2. Owing to the absence of high quality tissue in this retrospective study we were unable to determine the methylation status of the MLH1promoter and whether or not the two changes inMSH2 affected different alleles. Such studies need to be done in order to determine if somatic mismatch repair gene mutations in mismatch repair deficient cells confer additional functional properties. In this context it is important to note that coding mononucleotide tracts that are vulnerable to frameshifts exist not only in MSH6, but also in MSH2 as shown here.13 Somatic mutation of the mismatch repair genes (not including MLH1 promoter methylation) occurred at a rate of 5-10% in endometrial cancers, 2/42, 3/42, and 4/42 for MSH2,MLH1, and MSH6respectively (table 1).
In an unselected consecutive retrospective series of 74 endometrial cancers, microsatellite instability was found in 17 (23%). These 17 patients plus 25 other patients whose tumours were MSI negative were studied for MSH2,MLH1, and MSH6somatic and germline mutations. Germline mutation inMLH1 was found in one endometrial cancer patient. Thus, hereditary mutations in these genes contribute to sporadic endometrial cancer at a rate of at least 1.4% (1/74). If the germline missense change in MSH2 is causally related to the cancer in one patient then the contribution of hereditary mutations would be 2/74. This incidence is higher than in several previous studies.6-10
In all, the previous authors studied 352 EC tumours for microsatellite instability and found 78 that were MSI positive (22%). These 78 patients were studied by various methods for germline mutations inMLH1 and MSH2 and only one was found, corresponding to a frequency of 1/352 or 0.3%. Because of small numbers and varying methodologies, it is not possible to assess which of the two estimates (this study of 1.4% and that by previous authors of 0.3%) is most likely to be correct. Intuitively, the high incidence of EC in HNPCC families would seem to suggest that germline mutation must occur with appreciable frequency in “sporadic” EC. The large study of Wijnen et al 12 disclosed as many as 10 different truncating germline mutations of MSH6 among HNPCC or HNPCC-like families. These families had been ascertained clinically as being positive for the original or modified Amsterdam criteria. The families in which MSH6 was mutated characteristically had many patients with EC. It therefore seems that in addition to MLH1 andMSH2, one would expectMSH6 to be mutated in some “sporadic” EC patients. As can be seen in table 1, “sporadic” EC patients nevertheless often have some degree of a family history of cancer. This study is the first one that specifically searched for mutations inMSH6 in sporadic EC; the absence of mutations is somewhat surprising in view of the findings of Wijnenet al.12 It remains to be seen whether a larger series of patients might disclose mutations inMSH6. If not, one may need to consider whether truncating mutations of MSH6 are somehow enriched in the Dutch population and rare elsewhere.
To revisit the previously stated minimal role of hereditary nonpolyposis colorectal cancer (HNPCC) in “sporadic” endometrial cancer (EC), 74 unselected ECs were studied for microsatellite instability (MSI); 17/74 (23%) were MSI(+).
Mutational analyses were performed for MSH2, MLH1, and MSH6 in these 17 patients and an additional 25 MSI negative patients, most with a family history of cancer.
One definite germline mutation was found in MLH1. A missense change in MSH2 needs further study. Thus, the proportion of hereditary mutations was at least 1/74 (1.4%), but MSH6 did not contribute.
Frameshifts in a previously unreported hypermutable region of seven coding adenosines in exon 4 of MSH2 were discovered in two MSI positive tumours.
In conclusion, hereditable mismatch repair deficiency accounts for a small but definite proportion of sporadic EC.
We thank Dr Päivi Peltomäki for critical reading of this manuscript and Dr Wendy Frankel, Dr Gerard Nuovo, and Tina McKeegan for assistance with immunohistochemistry. This work was supported by NIH grants CA16058 and CA67941 and European Commission Grant BMH4CT960772.
website extra The supplementary data (figs 2–7) can be found on the JMG website www.jmedgenet.com