- © 2001 by American Society of Clinical Oncology
Molecular Screening for Hereditary Nonpolyposis Colorectal Cancer: A Prospective, Population-Based Study
- Antonio Percesepe,
- Francesca Borghi,
- Mirco Menigatti,
- Lorena Losi,
- Moira Foroni,
- Carmela Di Gregorio,
- Giuseppina Rossi,
- Monica Pedroni,
- Elisa Sala,
- Fabiana Vaccina,
- Luca Roncucci,
- Piero Benatti,
- Alessandra Viel,
- Maurizio Genuardi,
- Giancarlo Marra,
- Paula Kristo,
- Paivi Peltomäki and
- Maurizio Ponz de Leon
- From the Departments of Internal Medicine and Pathology, University of Modena, Modena; Division of Pathology, Ospedale di Carpi, Carpi; Division of Experimental Oncology I, Centro Riferimento Oncologico, Aviano; and Department of Medical Genetics, Catholic University of Rome, Rome, Italy; Institute for Medical Radiobiology, University of Zurich, Switzerland; Department of Medical Genetics, Haartman Institute, University of Helsinki, Finland; and Division of Human Cancer Genetics, Comprehensive Cancer Center, Ohio State University, Columbus, OH.
- Address reprint requests to Antonio Percesepe, MD, PhD, Department of Internal Medicine, University of Modena, Via del Pozzo 71, 41100 Modena, Italy; email: percesepe.antonio{at}unimo.it
Abstract
PURPOSE: Germline mutations in mismatch repair genes predispose to hereditary nonpolyposis colorectal cancer (HNPCC). To address effective screening programs, the true incidence of the disease must be known. Previous clinical investigations reported estimates ranging between 0.5% and 13% of all the colorectal cancer (CRC) cases, whereas biomolecular studies in Finland found an incidence of 2% to 2.7% of mutation carriers for the disease. The aim of the present report is to establish the frequency of the disease in a high-incidence area for colon cancer.
PATIENTS AND METHODS: Through the data of the local CRC registry, we prospectively collected all cases of CRC from January 1, 1996, through December 31, 1997 (N = 391). Three hundred thirty-six CRC cases (85.9% of the incident cases) were screened for microsatellite instability (MSI) with six to 12 mono- and dinucleotide markers. MSI cases were subjected to MSH2 and MLH1 germline mutation analysis and immunohistochemistry; the methylation of the promoter region was studied for MLH1.
RESULTS: Twenty-eight cases (8.3% of the total) showed MSI. MSI cases differed significantly from microsatellite-stable (MSS) cases for their proximal location (P < .01), high mucinous component (P < .01), and poor differentiation (P = .002). Of MSI cases studied (n = 12), only one with a family history compatible with HNPCC had a germline mutation (in MSH2). Five other patients with a family history of HNPCC (two with MSI and three with MSS tumors) did not show germline mutations.
CONCLUSION: We conclude that the incidence of molecularly confirmed HNPCC (one [0.3%] of 336) in a high-incidence area for CRC is lower than in previous biomolecular and clinical estimates.
THE MAIN THRUST TO the progress toward the control of and cure for colorectal cancer (CRC) is the continuous improvement of the prevention and treatment options for the disease. Ideal preventive strategies address screening of the general population by defining a risk profile for each subject, based on the information on individual lifestyle and genetic background. A paradigm for this strategy is represented by the pure genetic forms of CRC, familial adenomatous polyposis (FAP), and hereditary nonpolyposis CRC (HNPCC).1 In both syndromes, recognition of the disease and a close follow-up have resulted in a better survival and cure.2 Knowledge of the incidence of the disease is of primary importance to devise resources for those who will benefit most, ie, those harboring a predisposing germline mutation in one of the responsible gene(s). Although the typical phenotype of hundreds of polyps has traditionally aided the diagnosis of FAP, in HNPCC the absence of a characteristic feature of the tumor or of the patient has made the diagnosis more obscure. Therefore, relying on clinical grounds only, the estimates of the incidence of HNPCC have been quite variable, ranging from 0.5% to 13% of the total CRC burden.3-5 The identification of germline mutations that predispose to HNPCC and result in inadequate repair of mismatches during DNA replication, thus determining a unique phenotype (microsatellite instability [MSI]), has finally led to an unequivocal diagnosis of the disease in most cases.6 Moreover, the presence of a typical instability in tumor DNA has provided a marker for HNPCC, that has been used as a primary screening method for selecting those patients who will benefit from mutational analysis. In 1998, molecular screening of prospective CRC patients in Finland addressed the incidence of germline mutations of mismatch repair (MMR) genes in the southeastern regions of the country. Incident CRC cases were screened for MSI, and positive cases were submitted to mutational analysis of the mismatch repair genes MSH2 and MLH1.7 This method has been recently validated through a continuation study by the same authors.8 Within the same collaborative project, and with a broadly similar approach, we aimed at determining the frequency of germline mutations in MSH2 and MLH1 through the data of a CRC registry in a high-incidence area for colon cancer in Italy.
PATIENTS AND METHODS
The Registry
Details of the general organization of the specialized CRC registry have been described previously.5 Health Care District 16 includes Modena (northern Italy) and 10 surrounding communities, with a total of 265,227 inhabitants at the census 1991, with a crude CRC incidence for the period 1984 to 1995 of 64.5 new cases/100,000/year in men, and 55.2 in women.9 In addition to the clinical and pathologic features, the registry records family history of the patients stepwise by constructing genealogic trees and verifying referred diagnosis of any cancer type, as previously described.5 In brief, patients or their first-degree relatives are personally interviewed and detailed first-degree pedigrees are drawn. Neoplastic mortality or morbidity are verified by clinical and pathologic reports and/or death certificates. Patients are stratified according to six clinical criteria, associated with HNPCC (vertical transmission of cancer of the HNPCC spectrum, diagnosis of tumors of the HNPCC spectrum in the proband’s siblings, onset of CRC before the age of 50 years, proximal localization of CRC, tumor multiplicity, and mucinous histology); when two or more of these characteristics are present, the pedigree is expanded to include second- and third degree relatives.
Patients
From January 1, 1996, through December 31, 1997, 391 cases of colorectal adenocarcinoma were recorded in the local registry. Among them, 336 (85.9% of the incident cases) were prospectively collected as fresh-frozen or paraffin-embedded tissue. Specimens were not available for 55 cases, mainly for poor general conditions that contraindicated surgery. Informed consent was obtained from each patient before molecular analyses were carried out. Surrounding normal mucosa was used as a source of normal tissue.
MSI Analysis
For the initial 183 samples (the first year of registration), MSI analysis was originally carried out with a fluorescent-based polymerase chain reaction (PCR) technique, using a panel of 10 dipentanucleotide markers: D8S254, NM23, D18S35, TP53 (dinucleotide), D5S346, TP53 (pentanucleotide), D2S123, D1S288, D3S1611, and D7S501.7 The markers chosen were short enough to be used with paraffin-embedded material and to be easily multiplexed in a single run on a 373 DNA sequencer (Applied Biosystems, Foster City, CA). The number of successful amplifications ranged between five and 10. While this study was in progress, the Bethesda guidelines10 were developed, and to follow the published reference marker panel, two mononucleotides, BAT26, which is a particularly sensitive for MSI,11-12 and BAT25 were added. In all 183 cases, the MSI results achieved with the panel of dinucleotide markers remained unchanged.
The following 153 samples (the second year of registration) were studied with a radioactive technique13 using four mononucleotide (BAT25, BAT26, BAT40, and transforming growth factor beta RII) and two dinucleotide markers (D2S123 and D5S346). To comply with the Bethesda guidelines for high-degree MSI, at least 30% of the markers had to be shifted compared with the normal mucosa for a tumor to be defined MSI.
Mutation Analysis
At the beginning of the study, germline mutations in the MLH1 and MSH2 genes were sought by direct exon-by-exon sequencing in all cases that were found to be MSI-positive. Amplification products were generated with primers located in the flanking introns approximately 50 base pairs from the respective intron/exon borders to detect all possible splice junction mutations. The sequences were determined on an Applied Biosystems 373 DNA sequencer using fluorescently labeled primers and protocols supplied by the manufacturer. Altogether, eight patients were studied by this method. After an analogous mutation screen in Finland showed that only patients with the hallmarks of hereditary cancer (positive family history or early age of cancer onset or tumor multiplicity) harbored an MMR gene mutation,7 we limited our mutation analysis to the MSI cases having at least one of those features (four patients). Those four cases as well as three additional cases with clinical HNPCC according to the Amsterdam criteria I14 (and with microsatellite stable [MSS] tumors) were studied by direct sequencing (two cases) or by single-strand conformation polymorphism (SSCP; five patients) according to the methods previously described.15 Samples showing an altered SSCP mobility pattern were sequenced by means of the Sequenase PCR product sequencing Kit (Amersham Life Science, Buckinghamshire, United Kingdom) or by the above-mentioned fluorescent method.
Southern blotting analysis of genomic DNA was used to address the possibility of major rearrangements in two selected cases (see Results). Analysis of genomic digests of HindIII and EcoRI was performed with full length MSH2 and MLH1 cDNA as probes.
DNA Methylation Analysis
A PCR-based assay was used, relying on the modification in the base sequence induced by chemical treatment with bisulfite, efficiently converting unmethylated cytosines to uracil while leaving the methylated ones unchanged. Two allele-specific PCR primers were designed for the MLH1 promoter region to be complementary to the modified or unmodified sequence according to the published protocols.16 Briefly, 2 μg of genomic DNA was denatured by 0.2 mol/L of NaOH for 10 minutes at 37°C. Thirty microliters of 10 mmol/L hydroquinone (Sigma-Aldrich, St. Louis, MO) and 520 μL of 3 mol/L of sodium bisulfite (Sigma) at pH 5.0 were added and incubated at 50°C for 16 to 18 hours. Purification through the PCR-purification system (Promega Italia, Milano, Italy), precipitation with ethanol, and resuspension in 20 μL of water followed. One microliter of template was subjected to 35 PCR cycles, with an annealing temperature of 53°C for the unmethylated and 61°C for the methylated reaction. We used published primers,16 with the only exception for the reverse primer of the methylated reaction: 5′ TCC GAA AAA CGA TAA AAC CCT ATA C 3′.
Immunohistochemistry
Sections were deparaffinized with xylene and dehydrated by using graded ethanols. Sections were immersed in 0.01 mol/L of sodium citrate buffer (pH 6.0) and subjected to heat-induced antigen retrieval in a microwave oven (350 W for 30 minutes). Immunoperoxidase staining using diaminobenzidine as chromogen was performed with the NEXES Automatic Staining System (Ventana, Strasbourg, France). Mouse antihuman MLH1 antibody (Clone 168-15, Pharmingen, San Diego, CA) and mouse antihuman MSH2 antibody (Clone G219-1129, Pharmingen) were used at 1/40 dilution.
RESULTS
Of the 336 CRC cases subjected to MSI analysis, 28 (8.3%) showed MSI. The 95% confidence interval (CI) is 5.4% to 11.3%, calculated as a normal approximation to the binomial distribution.17 In 12 patients with MSI tumors (including eight randomly chosen cases and four that were selected on the basis of their clinical features) (Table 1), normal-tissue DNA (either normal mucosa or blood) was screened for MSH2 and MLH1 mutations. Only one germline mutation was found (exon 16, del A at 2647 of the MSH2 gene; Fig 1), that occurred in an Amsterdam-positive family. Extrapolation to the entire series yields a rate of one (0.3%) of 336 for germline MMR gene mutations among all patients with CRC (95% CI, 0% to 0.9%). In the course of the above-mentioned analyses, two base substitutions regarded as polymorphisms were detected: MLH1 exon 17, codon 653 G → T (silent), and MLH1 intron 13, codon 1558+14 g → a.18
Clinicopathologic Characteristics of 28 MSI-Positive Patients and Results of MMR Mutation Analysis, Immunohistochemistry for MSH2 and MLH1, and Study of the MLH1 Promoter Methylation Status
Fig 1. MSH2 mutation. (a) SSCP analysis of normal individuals (C), and 4 patients are shown for exon 16. The mobility shift in patient no. 30 is indicated by the arrow. (b) DNA sequencing of exon 16 from the same patient evidencing a heterozygous A deletion (arrow).
Twenty (71.4%) of the 28 MSI CRCs showed a lack of expression of the MLH1 protein (Fig 2A) and 14 of those latter cases had hypermethylation of the MLH1 promoter region (Fig 2B; Table 1). MSH2 protein expression was normal in all tumors. Also the case showing a MSH2 germline mutation expressed MSH2 protein, which is explained by the fact that the mutation affected the very COOH terminus of the protein and therefore was predicted to result in a stable product. To clarify any doubt about the pathogenicity of the mutation, the pattern of segregation of the mutation in the family was studied: it occurred in three affected members and was absent in one unaffected relative.
Fig 2. (a) MLH1 protein immunohistochemistry. Abbreviations: N, an area of normal mucosa; T, tumoral tissue with absent expression of the MLH1 protein. (b) Bisulphite PCR analysis of the MLH1 promoter, showing examples of the results of the methylated (lanes M) and unmethylated (lanes U) reactions.
Site distribution of MSI cases was uniformly proximal (P < .01 v MSS tumors), grading was significantly (P = .002) shifted towards less differentiated features, and mucinous component more than 50% of the tumor was present in 46% of the cases (compared with 16% of the MSS tumors, P < .01). No difference was seen in sex, age of onset, multiplicity of tumors, staging, presence of synchronous adenomas, and presence of one affected first-degree family member with tumors of the HNPCC spectrum (Table 2).
Clinical and Pathologic Characteristics of the MSI-Positive and MSI-Negative Colorectal Tumors
Altogether, six cases were from families fulfilling the Amsterdam criteria14: three of those were MSI, and a germline mutation was found in one (see above). In the remaining two MSI cases (two brothers), direct sequencing provided no evidence of mutation. MLH1 protein expression was lost in both cases, and the promoter region of the MLH1 gene was methylated in one (Table 1, cases 36/96 and 9/97). Importantly, none of the two exhibited any apparent rearrangements in MSH2 or MLH1 by Southern blotting. The other three HNPCC cases according to the Amsterdam criteria I were MSS; they were analyzed for germline mutations in MSH2 and MLH1 with no evidence of mutation.
DISCUSSION
By a population-based approach to the HNPCC molecular diagnosis, the incidence of the disease in a high-incidence Western area for CRC was 0.3%, lower than any previous estimate, and the overall frequency of MSI was also as low as 8.3%. These two results will be discussed separately below.
The figure of 0.3% for the incidence of HNPCC represents a minimum estimate for various reasons. Technical limitations constitute one possible source for the low incidence figures; for example, SSCP, which was used for screening half of the patients, has a detection rate of only approximately 80%. Higher detection rates can be achieved by direct sequencing, which was used for the remaining half of our patients; however, because even this method misses certain types of mutations (especially large deletions),19 novel approaches, like the haploid conversion, have been developed to improve the effectiveness of mutation analysis.20 To overcome these limitations, we also examined the expression patterns of MSH2 and MLH1: although MSH2 was invariantly expressed, MLH1 protein was absent in 71% of the MSI cancers and the vast majority (70%) of the latter cases were associated with promoter hypermethylation, a recently discovered mechanism for epigenetic inactivation.16 MLH1 promoter hypermethylation is a property of true sporadic MSI tumors and is significantly less frequent in HNPCC,21 although the methylation mechanism might be associated with the inactivation of the wild-type allele in some hereditary cases as well.
A previous population-based biomolecular analysis performed on the Finnish population7 reported an overall incidence of germline mutations of 2% of the total CRC burden. By increasing the number of subjects studied, the same authors recently arrived at a higher estimate of the incidence of germline mutations, up to 2.7% of the total CRC cases.8 Notably, however, the presence of two founder mutations, which account for approximately half of the HNPCC-associated MMR mutations present in the Finnish population22 and are extremely rare in the rest of the world, may have contributed to the higher figures in the Finnish study. Moreover, the different overall mutation profiles among Finnish and Italian HNPCC kindreds emphasize population-specific differences that could account for the observed differences in the incidence of the disease. Additionally, in our area of northern Italy, the incidence of CRC in the general population is high,9 and this could result in the dilution of the hereditary cases in a background of environmentally related CRC.
Taking advantage of the previous work by Aaltonen et al,7 which recommended mutation screening only in MSI cases having at least one of the HNPCC features (positive family history, young age of onset, or cancer multiplicity), we modified the initial approach and instead of screening all MSI cases for mutations focused only on those having at least one of the above-mentioned hallmarks. Importantly, at the beginning of our study and before the disclosure of the results achieved in Finland, eight patients with MSI tumors were screened for MMR mutations irrespective of those clinical features, and no germline mutations were found, implying that the modification of the criteria was not responsible for the low incidence of mutations. On the other hand, three patients, whose genealogic trees fulfilled the clinical Amsterdam criteria for HNPCC14 but whose tumors did not have MSI, failed to show any MSH2 and MLH1 mutations, suggesting that the selection criteria proposed by Aaltonen et al7 are both sensitive and specific, at least as far as MSH2 and MLH1-linked HNPCC is concerned. Sensitivity of the whole strategy for mutation screening, however, is not 100%, as shown during the recent validation of the criteria performed by the Finnish authors, because one of their 18 HNPCC cases, a 61-year-old woman apparently without any of the above-mentioned clinical features in the nuclear pedigree, would have escaped detection if the clinical/MSI-based selection alone was strictly applied.8
A possibility remains that inherited mutations in another MMR gene, MSH6, recently found mutated in atypical HNPCC families with colorectal and endometrial cancers, could account for an additional share of the incident cases.23-25 However, the absence of a definite tumor phenotype for MSH6 (MSS or MSI, high or low) and of typical clinical features of the patients, suggests that a further clarification of these issues is necessary before a population-based mutation analysis for this gene should be started.
The low rate of germline HNPCC mutations in the general population parallels the findings from other genetic diseases; for example, in hereditary breast cancer, the initial estimates for BRCA1-associated cases that were based on striking familial aggregates and families already positive for linkage have dropped from 45%26 to 7% to 16%27-28 of the familial and early-onset cases in the majority of Western populations.
The presently observed relatively low overall frequency of MSI cases may reflect the fact that we mainly relied on the BAT26 marker in combination with a large overall marker panel. BAT26 has been shown to be selectively mutated in CRCs with high-level instability.11,12 Many previous analyses were limited to dinucleotide markers instead, some of which turned out to be too sensitive to detect MSI29 in that low-degree instability tumors without any evidence for the involvement of MMR genes were included.30 Through an increasing knowledge of the phenomenon, and by using more accurate markers, we report an overall frequency of MSI lower than in many previous studies31-34 but similar to a recent study using an equivalent strategy.35 This reduction in the rate of unstable tumors is consistent with a similar observation in other cancers, including those of the breast and prostate36: after the initial reports, only few cases of high-degree MSI in organs not belonging to the HNPCC spectrum have been described. Otherwise, we confirm the previously proposed features for sporadic MSI colorectal tumors: late-onset, right-sided, poorly differentiated, and mucinous and no relation to family history of tumors.
Finally, we ascertained all of our cases through CRC and therefore would miss potential germline mutation carriers presenting with other cancers, eg, endometrial cancer, which is the second most common malignancy in HNPCC. Kowalski et al37 recently screened 125 primary endometrial adenocarcinomas for MSI followed by mutation analysis of MSH2 and MLH1 by SSCP. One germline mutation was found (one [0.8%] of 125), suggesting that the inclusion of endometrial cancers might increase the incidence figure for HNPCC somewhat, but not significantly.
In summary, the incidence of HNPCC associated with mutations in MSH2 and MLH1 genes in a Western area was calculated: the low rate of the disease, 0.3% of all CRC cases, though representing a minimum estimate, suggests that prevention of the disease in mutation-carriers is a feasible and unavoidable, high-priority task.
Acknowledgments
Supported by grants from the European Commission (contract no. BMH4-CT96-0772) and the Italian Ministry of University and Technological Research.
ACKNOWLEDGMENT
We thank Albert de la Chapelle, MD, for his helpful advice throughout the study period.
- Received July 27, 2000.
- Accepted June 1, 2001.
