Portal de Periódicos da CAPES

Colorectal Pretumor Progression Before and After Loss of DNA Mismatch Repair

2004; Elsevier BV; Volume: 164; Issue: 4 Linguagem: Inglês

10.1016/s0002-9440(10)63231-2

1525-2191

Peter Calabrese, Jen‐Lan Tsao, Yasushi Yatabe, Reijo Salovaara, Jukka‐Pekka Mecklin, Heikki Järvinen, Lauri A. Aaltonen, Simon Tavaré, Darryl Shibata,

Cancer Genomics and Diagnostics

A pretumor progression model predicts many oncogenic cancer mutations may first accumulate in normal appearing colon. Although direct observations of early pretumor mutations are impractical, it may be possible to retrospectively reconstruct tumor histories from contemporary cancer mutations. To infer when and in what order mutations occur during occult pretumor progression, we examined 14 cancers from individuals with heterozygous germline mutations in DNA mismatch repair (MMR) genes or hereditary nonpolyposis colorectal cancer (HNPCC). Somatic inactivation of the normal allele occurs sometime during a lifetime and results in loss of MMR, elevated mutation rates, and subsequent widespread somatic microsatellite mutations in HNPCC cancers. Patient ages at MMR loss can be estimated because intervals between MMR loss and cancer removal can be inferred from numbers of microsatellite tumor mutations. The relative order of MMR loss during pretumor progression may also be inferred from its collective ages of occurrence. Somatic MMR loss preceded cancer removal by an average of 6.1 years, occurred relatively late in life (average of 41.6 versus 47.7 years at cancer removal), and was a surprisingly late (fifth or sixth) step. Calculations indicate five or six oncogenic mutations could accumulate with relatively normal replication fidelity in normal appearing colon. HNPCC pretumor progression essentially begins from birth and ends with MMR loss, implying elevated mutation rates and tumorigenesis may be unnecessary for most progression. A pretumor progression model predicts many oncogenic cancer mutations may first accumulate in normal appearing colon. Although direct observations of early pretumor mutations are impractical, it may be possible to retrospectively reconstruct tumor histories from contemporary cancer mutations. To infer when and in what order mutations occur during occult pretumor progression, we examined 14 cancers from individuals with heterozygous germline mutations in DNA mismatch repair (MMR) genes or hereditary nonpolyposis colorectal cancer (HNPCC). Somatic inactivation of the normal allele occurs sometime during a lifetime and results in loss of MMR, elevated mutation rates, and subsequent widespread somatic microsatellite mutations in HNPCC cancers. Patient ages at MMR loss can be estimated because intervals between MMR loss and cancer removal can be inferred from numbers of microsatellite tumor mutations. The relative order of MMR loss during pretumor progression may also be inferred from its collective ages of occurrence. Somatic MMR loss preceded cancer removal by an average of 6.1 years, occurred relatively late in life (average of 41.6 versus 47.7 years at cancer removal), and was a surprisingly late (fifth or sixth) step. Calculations indicate five or six oncogenic mutations could accumulate with relatively normal replication fidelity in normal appearing colon. HNPCC pretumor progression essentially begins from birth and ends with MMR loss, implying elevated mutation rates and tumorigenesis may be unnecessary for most progression. Multiple mutations are present in colorectal cancers. Although it is uncertain when mutations occur, a pretumor progression model1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar predicts some cancer mutations are first acquired in normal appearing colons. Hereditary cancer syndromes identify mutations that may accumulate during pretumor progression. For example, individuals with familial adenomatous polyposis (FAP) or hereditary nonpolyposis colorectal cancer (HNPCC) inherit heterozygous mutations, respectively, in APC and DNA mismatch repair (MMR) genes, but are born with normal appearing colons.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar Although germline mutations precede tumor progression, the timing, number, and order of somatic mutations during pretumor progression are poorly characterized (Figure 1). Ideally the entire mutational history of a tumor could be documented from birth to removal. In HNPCC, there is a sequence of mutations leading to MMR-deficient cancers. Germline MMR mutations (usually MLH1 or MSH2) are followed by somatic inactivation of the normal allele, leading to functional MMR loss and elevated mutation rates, with microsatellite (MS) loci rates increased 100- to 1000-fold.3Shibata D Peinado MA Ionov Y Malkhosyan S Perucho M Genomic instability in repeated sequences is an early somatic event in colorectal tumorigenesis that persists after transformation.Nat Genet. 1994; 6: 273-281Crossref PubMed Scopus (457) Google Scholar, 4Bhattacharyya NP Skandalis A Ganesh A Groden J Meuth M Mutator phenotypes in human colorectal carcinoma cell lines.Proc Natl Acad Sci USA. 1994; 91: 6319-6323Crossref PubMed Scopus (405) Google Scholar MS mutations subsequently accumulate every day, and HNPCC colorectal tumors characteristically contain ubiquitous somatic MS mutations.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar Recent studies illustrate that intervals between somatic MMR loss and tumor removal can be estimated retrospectively from numbers of MS tumor mutations.5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar Essentially the longer the intervals between somatic MMR loss and tumor removal, the greater the numbers of accumulated MS tumor mutations. A patient's age at somatic MMR loss can be estimated by subtracting the interval between MMR loss and tumor removal from age at tumor removal. Somatic MMR loss may occur during pretumor progression because congenital MMR deficiencies are compatible with normal appearing colons.7Parsons R Li GM Longley M Modrich P Liu B Berk T Hamilton SR Kinzler KW Vogelstein B Mismatch repair deficiency in phenotypically normal human cells.Science. 1995; 268: 738-740Crossref PubMed Scopus (297) Google Scholar, 8Ricciardone MD Ozcelik T Cevher B Ozdag H Tuncer M Gurgey A Uzunalimoglu O Cetinkaya H Tanyeli A Erken E Ozturk M Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1.Cancer Res. 1999; 59: 290-293PubMed Google Scholar, 9Wang Q Lasset C Desseigne F Frappaz D Bergeron C Navarro C Ruano E Puisieux A Neurofibromatosis and early onset of cancers in hMLH1-deficient children.Cancer Res. 1999; 59: 294-297PubMed Google Scholar, 10Vilkki S Tsao JL Loukola A Poyhonen M Vierimaa O Herva R Aaltonen LA Shibata D Extensive somatic microsatellite mutations in normal human tissue.Cancer Res. 2001; 61: 4541-4544PubMed Google Scholar In addition, for HNPCC tumors that appear shortly after negative clinical examinations, somatic MMR loss appears to precede normal examinations by many years.5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar, 11Kim KM Salovaara R Mecklin JP Jarvinen HJ Aaltonen LA Shibata D PolyA deletions in hereditary nonpolyposis colorectal cancer: mutations before a gatekeeper.Am J Pathol. 2002; 160: 1503-1506Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar Here we use methodologies of previous studies1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar, 5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar to illustrate how the timing, number, and order of occult pretumor somatic mutations may be inferred (Figure 1). Another group's modeling work suggests chromosomal instability may precede tumorigenesis;12Nowak MA Komarova NL Sengupta A Jallepalli PV Shih IEM Vogelstein B Lengauer C The role of chromosomal instability in tumor initiation.Proc Natl Acad Sci USA. 2002; 99: 16226-16231Crossref PubMed Scopus (393) Google Scholar we present experimental evidence that MS instability and the majority of oncogenic cancer mutations also precede HNPCC tumorigenesis. DNA was isolated from 14 formalin-fixed, paraffin-embedded colorectal cancers5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar from HNPCC patients with germline MLH1 or MSH2 mutations confirmed by sequencing. Somatic MS mutations were detected after dilution of the DNA to essentially single molecules, followed by polymerase chain reaction and acrylamide gel analysis.6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar Intervals between loss of MMR and cancer removal, and since the start of final clonal expansion were estimated based on MS mutation patterns or the drift of MS alleles from germline sizes.6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar Histories of 10 cancers were previously reported with an analysis that assumes one division per day and stepwise MS mutation with a rate of 0.005 per division.5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar Patient age at MMR loss was estimated by subtracting the time between MMR loss and tumor removal from the age of the patient at surgery. Clinical data were obtained from patient charts. In a previous study,1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar we have presented a pretumor progression model. Given the ages of patients when they were diagnosed with cancer, we developed a methodology to infer the number of mutations necessary for cancer and their mutation rates. In the present study we use the same methodology to infer the number of mutations necessary for MMR loss and their mutation rates, given the ages of patients at MMR loss (Table 1). Our approach is Bayesian: we assume a uniform prior on the number of mutations, and we calculate the posterior probabilities of the number of mutations (Table 2). In contrast to our previous study,1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar we consider a much smaller data set; as a consequence several of these probabilities are non-zero. These probabilities can be used to form a credibility interval for the numbers of mutations.Table 1Tumor HistoriesTumorStageAge at removalAge at MMR loss*Age (in years) at cancer removal minus tumor age.No. loci testedTumor age since loss of MMR (CI†95% confidence intervals.)Age of clonal expansionH1D4744243.4 (1.0 to 4.8)0.82H2D3531284.1 (1.6 to 5.7)0.77H3C4338294.6 (2.0 to 6.6)0.52H4A5853235.1 (2.0 to 7.6)0.27H5D4034215.5 (2.0 to 7.7)2.0H6B4640205.6 (1.8 to 8.5)0.93H7B4438305.7 (2.2 to 7.9)0.96H8B4337295.8 (2.5 to 8.2)0.25H9B4236285.9 (2.4 to 8.3)0.71H10A5448226.0 (2.1 to 7.7)1.2H11D3832216.1 (2.2 to 8.6)1.8H12B5751246.3 (2.7 to 9.2)0.52H13B6455268.8 (4.1 to 13)0.77H14B57442413 (5.3 to 19)0.96Average47.741.624.96.10.9* Age (in years) at cancer removal minus tumor age.† 95% confidence intervals. Open table in a new tab Table 2Oncogenic Mutations to MMR Loss or CancerMMR loss234*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.5*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.6*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.7*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.8*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.910Probability0.000.020.110.290.320.190.070.020.00Cancer234*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.5*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.6*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.7*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.8*Results from a Bayesian analysis1 using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer.910Probability0.000.020.100.260.310.200.080.020.00* Results from a Bayesian analysis1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar using the data in Table 1. We assumed a uniform prior on the number of mutations being between 2 and 10. The 95% credible interval is 4 to 8, with 5 or 6 the most likely numbers of oncogenic mutations to MMR loss or cancer. Open table in a new tab Mutations at 21 to 30 different CA-repeat MS loci were analyzed from 14 HNPCC cancers (Table 1). Intervals between MMR loss and tumor removal (Figure 1) were estimated from repeat unit differences between germline and tumor sizes.6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar Cancer histories were variable but MMR loss always preceded final clonal expansions (Figure 2). The average interval between MMR loss and cancer removal was 6.1 years, with an average final clonal expansion age of 1 year. Ages at MMR loss can be estimated by subtracting intervals between tumor removal and MMR loss from the age at cancer removal for each patient. MMR loss occurred relatively late during life at an average age of 41.6 versus 47.7 years at cancer removal. Mutations are associated with incremental changes in phenotype during tumor progression.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar Therefore, clinically more advanced cancers might have longer intervals between MMR loss and cancer removal, and greater numbers of MS mutations. However, there was no correlation between tumor age and clinical stage (Figure 3A). Higher stage cancers tended to have older clonal expansions (Figure 3B), but overall final clonal expansion intervals were short (average of 1 year). Of interest, intervals between MMR loss and cancer removal correlated with patient age at cancer removal (Figure 3C). Average tumor age was significantly greater for patients older than 50 years old at the time of surgery (8.0 versus 5.2 years, P = 0.022, two-tailed t-test). A relationship between tumor and patient ages is consistent with multistep models13Armitage P Doll R The age distribution of cancer and multistage theory of carcinogenesis.Br J Cancer. 1954; 1: 1-12Crossref Scopus (1247) Google Scholar, 14Cook PJ Doll R Fellingham SA A mathematical model for the age distribution of cancer in man.Int J Cancer. 1969; 4: 93-112Crossref PubMed Scopus (246) Google Scholar, 15Peto R Epidemiology, multistage models, and short-term mutagenicity tests.in: Hiatt H Watson JD Winsten JD Origins of Human Cancer. Cold Spring Harbor Laboratory, Cold Spring Harbor1977: 1403-1428Google Scholar that postulate that cancers have the same number of oncogenic mutations regardless of patient age at cancer. For example, cancers arising in a 42-year-old patient and a 60-year-old patient would have the same number of oncogenic mutations. Simplistically, the difference between these cancers is the time to acquire their mutations. If cancers require six oncogenic mutations, then on average each mutation requires ∼7 years in the 42-year-old patient and 10 years in the 60-year-old patient because mutations accumulate sequentially (Figure 4). The process starts at birth and the first mutations stochastically accumulate at early ages. The older ages at MMR loss for cancers removed from older patients generally fit a multistep model (Figure 4). The late ages at MMR loss relative to cancer removal allow for earlier mutations. Numbers of mutations can be inferred from ages of occurrence of any measurable event. For example, cancer frequencies increase exponentially with age and this age at cancer information can be used to infer total numbers of oncogenic cancer mutations. Similarly, numbers of oncogenic mutations preceding MMR loss can be estimated if ages at MMR loss are known. The most likely number of mutations can be estimated with a Bayesian approach1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar for small samples. From the ages at MMR loss (Table 1), MMR loss most likely represented the fifth or sixth oncogenic mutation (Table 2). A similar analysis using ages at cancer removal (Table 1) estimated HNPCC cancers most likely require five or six oncogenic mutations (Table 2). Both estimates are consistent with HNPCC cancers requiring multiple oncogenic mutations for transformation, with MMR loss representing a late oncogenic mutation. The past appearance of a tumor progenitor may be inferred when a negative clinical examination precedes tumor removal. Unless a tumor was missed, a negative clinical examination suggests the absence of a visible tumor phenotype at that age. Negative examinations (surgery or colonoscopy) preceded removal of four HNPCC cancers (Figure 2). Loss of MMR was estimated to precede these negative examinations, indicating MMR loss and earlier oncogenic mutations occur during pretumor progression. To test the feasibility of this conclusion, mutation rates required to accumulate five or six oncogenic mutations by the estimated ages of MMR loss were calculated as in Calabrese and colleagues.1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar For our baseline model (15 million crypts, 64 stem cells per crypt, one stem cell division per day) and neutral mutations, rates of ∼8.8 × 10−7 to 4.4 × 10−6 per locus are needed to collect five or six oncogenic mutations in a single HNPCC cancer progenitor stem cell by the times of MMR loss indicated in Table 1. If every mutation is followed by a sweep that results in very rapid fixation, lower rates (3.1 × 10−8 to 6.8 × 10−8) are needed to collect five or six mutations. Rates would be between the two scenarios if early progression combines selective and neutral mutations. These estimated mutation rates are within the realm of normal replication fidelity,16Loeb LA Mutator phenotype may be required for multistage carcinogenesis.Cancer Res. 1991; 51: 3075-3079PubMed Google Scholar, 17Tomlinson IP Novelli MR Bodmer WF The mutation rate and cancer.Proc Natl Acad Sci USA. 1996; 93: 14800-14803Crossref PubMed Scopus (354) Google Scholar indicating it is feasible to accumulate multiple oncogenic somatic mutations before MMR loss during pretumor progression. Ideally one could identify all oncogenic cancer mutations, their relative order, and when during a lifetime they occurred. Here we outline an approach (Figure 1) that can characterize progression from birth to tumor removal based on a pretumor progression model.1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar In HNPCC, somatic loss of MMR occurs between birth and tumor removal because normal cells are MMR proficient (MLH1+/−) but cancers are MMR-deficient (MLH1−/−). MMR loss results in greatly elevated mutation rates and large numbers of somatic MS mutations are present in HNPCC cancers.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar The longer the intervals after MMR loss, the greater the numbers of MS tumor mutations. A quantitative analysis of MS tumor mutation patterns5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar can estimate time since MMR loss and time since the start of the final clonal expansion (Figure 2). With this information, patient age at the time of somatic MMR loss can be estimated (Figure 1). If cancer is a multistep process (one mutation after another), ages at cancer and ages at intermediate mutations should also follow a multistep pattern. Cancer ages of occurrence are consistent with five to seven oncogenic mutations or stages.13Armitage P Doll R The age distribution of cancer and multistage theory of carcinogenesis.Br J Cancer. 1954; 1: 1-12Crossref Scopus (1247) Google Scholar, 14Cook PJ Doll R Fellingham SA A mathematical model for the age distribution of cancer in man.Int J Cancer. 1969; 4: 93-112Crossref PubMed Scopus (246) Google Scholar, 15Peto R Epidemiology, multistage models, and short-term mutagenicity tests.in: Hiatt H Watson JD Winsten JD Origins of Human Cancer. Cold Spring Harbor Laboratory, Cold Spring Harbor1977: 1403-1428Google Scholar Ages at cancer from the current patients were consistent with HNPCC colorectal cancers requiring five or six oncogenic mutations for transformation. Similarly, ages at MMR loss from multiple HNPCC cancers (Table 1) can be used to estimate numbers of oncogenic mutations preceding MMR loss. Potentially MMR loss may occur anytime and in any order during HNPCC progression. However, instead of a random order for MMR loss, its ages of occurrence were consistent with MMR loss late in life as a fifth or sixth oncogenic mutation (Figure 1). A final uncertainty is the separation between pretumor and tumor progression. Although MMR loss is compatible with normal phenotypes,7Parsons R Li GM Longley M Modrich P Liu B Berk T Hamilton SR Kinzler KW Vogelstein B Mismatch repair deficiency in phenotypically normal human cells.Science. 1995; 268: 738-740Crossref PubMed Scopus (297) Google Scholar, 8Ricciardone MD Ozcelik T Cevher B Ozdag H Tuncer M Gurgey A Uzunalimoglu O Cetinkaya H Tanyeli A Erken E Ozturk M Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1.Cancer Res. 1999; 59: 290-293PubMed Google Scholar, 9Wang Q Lasset C Desseigne F Frappaz D Bergeron C Navarro C Ruano E Puisieux A Neurofibromatosis and early onset of cancers in hMLH1-deficient children.Cancer Res. 1999; 59: 294-297PubMed Google Scholar, 10Vilkki S Tsao JL Loukola A Poyhonen M Vierimaa O Herva R Aaltonen LA Shibata D Extensive somatic microsatellite mutations in normal human tissue.Cancer Res. 2001; 61: 4541-4544PubMed Google Scholar MMR loss may also occur in tumor cells. Interval tumors or cancers that appear shortly after negative clinical examinations provide insights into past phenotypes of tumor progenitors. Although false-negative examinations are possible, a negative examination indicates an age at which a tumor progenitor still retains a normal phenotype. In conjunction with an estimate for the start of final clonal expansion, a negative clinical examination marks the minimal age for the end of pretumor progression (Figure 1). These approaches describe HNPCC progression as a stepwise series of mutations leading to a MMR-deficient cancer. Most oncogenic mutations including somatic MMR loss initially fail to confer visible changes in phenotype and accumulate in stem cells during pretumor progression. Somatic MMR loss is preceded by a number of other somatic oncogenic mutations. Visible tumor progression is restricted to the terminal few years (Figure 1). Our model assumes mutation rates are normal until both MMR alleles are lost. We note that cells with heterozygous MMR mutations are repair proficient18Parsons R Li GM Longley MJ Fang WH Papadopoulos N Jen J de la Chapelle A Kinzler KW Vogelstein B Modrich P Hypermutability and mismatch repair deficiency in RER+ tumor cells.Cell. 1993; 75: 1227-1236Abstract Full Text PDF PubMed Scopus (972) Google Scholar and mice with heterozygous MMR mutations do not exhibit increased mutation frequencies.19Narayanan L Fritzell JA Baker SM Liskay RM Glazer PM Elevated levels of mutation in multiple tissues of mice deficient in the DNA mismatch repair gene Pms2.Proc Natl Acad Sci USA. 1997; 94: 3122-3127Crossref PubMed Scopus (132) Google Scholar However, subtle increases in mutation rates because of haploinsufficiency or dominant-negative interactions are also consistent with our analysis because inferred mutation rates during pretumor progression were slightly higher for HNPCC cancers compared to sporadic cancers (8.8 × 10−7 to 4.4 × 10−6 for HNPCC cancers versus 3.2 × 10−7 to 7.4 × 10−7 per locus for sporadic cancers and the neutral model1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar). Of note, yeast with heterozygous MMR mutations exhibit a threefold increase in mutations compared to wild-type yeast.20Drotschmann K Clark AB Tran HT Resnick MA Gordenin DA Kunkel TA Mutator phenotypes of yeast strains heterozygous for mutations in the MSH2 gene.Proc Natl Acad Sci USA. 1999; 96: 2970-2975Crossref PubMed Scopus (78) Google Scholar A number of assumptions underlie our pretumor progression model and its conclusions.1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar Therefore, rather than providing a final description, the current analysis demonstrates how modeling can produce a coherent and internally consistent description of progression throughout a lifetime. To test the model for consistency with HNPCC biology we consider two observations, congenital MMR deficiency and surveillance attempts to prevent HNPCC cancers. A major uncertainty is whether MMR loss occurs early or late during progression. Progression to cancer could be rapid after MMR loss because mutation rates are elevated.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar However, early MMR loss alone does not appear sufficient for rapid or efficient progression because colorectal cancers are infrequent in individuals with homozygous germline repair mutations,7Parsons R Li GM Longley M Modrich P Liu B Berk T Hamilton SR Kinzler KW Vogelstein B Mismatch repair deficiency in phenotypically normal human cells.Science. 1995; 268: 738-740Crossref PubMed Scopus (297) Google Scholar, 8Ricciardone MD Ozcelik T Cevher B Ozdag H Tuncer M Gurgey A Uzunalimoglu O Cetinkaya H Tanyeli A Erken E Ozturk M Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1.Cancer Res. 1999; 59: 290-293PubMed Google Scholar, 9Wang Q Lasset C Desseigne F Frappaz D Bergeron C Navarro C Ruano E Puisieux A Neurofibromatosis and early onset of cancers in hMLH1-deficient children.Cancer Res. 1999; 59: 294-297PubMed Google Scholar, 10Vilkki S Tsao JL Loukola A Poyhonen M Vierimaa O Herva R Aaltonen LA Shibata D Extensive somatic microsatellite mutations in normal human tissue.Cancer Res. 2001; 61: 4541-4544PubMed Google Scholar and before 30 years of age in HNPCC families.21Aarnio M Mecklin JP Aaltonen LA Nystrom-Lahti M Jarvinen HJ Life-time risk of different cancers in hereditary non-polyposis colorectal cancer (HNPCC) syndrome.Int J Cancer. 1995; 64: 430-433Crossref PubMed Scopus (522) Google Scholar With congenital MMR deficiencies, cancers are rare even though every colon stem cell lacks MMR. The lack of colorectal cancers with congenital MMR loss is consistent with our model, in which a number of oncogenic mutations usually precede MMR loss. Progression to cancer is efficient once MMR loss occurs, with an average interval estimated here at 6.1 years, but a number of other mutations must first precede MMR loss. Most individuals with congenital MMR deficiencies die within the first decade of life7Parsons R Li GM Longley M Modrich P Liu B Berk T Hamilton SR Kinzler KW Vogelstein B Mismatch repair deficiency in phenotypically normal human cells.Science. 1995; 268: 738-740Crossref PubMed Scopus (297) Google Scholar, 8Ricciardone MD Ozcelik T Cevher B Ozdag H Tuncer M Gurgey A Uzunalimoglu O Cetinkaya H Tanyeli A Erken E Ozturk M Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1.Cancer Res. 1999; 59: 290-293PubMed Google Scholar, 9Wang Q Lasset C Desseigne F Frappaz D Bergeron C Navarro C Ruano E Puisieux A Neurofibromatosis and early onset of cancers in hMLH1-deficient children.Cancer Res. 1999; 59: 294-297PubMed Google Scholar, 10Vilkki S Tsao JL Loukola A Poyhonen M Vierimaa O Herva R Aaltonen LA Shibata D Extensive somatic microsatellite mutations in normal human tissue.Cancer Res. 2001; 61: 4541-4544PubMed Google Scholar and may fail to develop colorectal cancers because they seldom live long enough to acquire these other stages. Potentially these mutations may be difficult to acquire after MMR loss because our model suggests they are usually acquired in MMR-proficient cells (Figure 1). Although our pretumor model does not explain why a number of oncogenic mutations usually precede somatic MMR loss, it is consistent with the observed rarity of colorectal cancers with congenital MMR deficiencies. Cancers should be preventable if most mutations accumulate during tumor progression. A cancer should be preceded by an adenoma, and periodic surveillance and polypectomy should prevent cancers. However, HNPCC cancers frequently appear relatively soon after screening surveillance examinations.22Jarvinen HJ Aarnio M Mustonen H Aktan-Collan K Aaltonen LA Peltomaki P De La Chapelle A Mecklin JP Controlled 15-year trial on screening for colorectal cancer in families with hereditary nonpolyposis colorectal cancer.Gastroenterology. 2000; 118: 829-834Abstract Full Text Full Text PDF PubMed Scopus (1135) Google Scholar, 23de Vos tot Nederveen Cappel WH Nagengast FM Griffioen G Menko FH Taal BG Kleibeuker JH Vasen HF Surveillance for hereditary nonpolyposis colorectal cancer: a long-term study on 114 families.Dis Colon Rectum. 2002; 45: 1588-1594Crossref PubMed Scopus (193) Google Scholar, 24Renkonen-Sinisalo L Aarnio M Mecklin JP Jarvinen HJ Surveillance improves survival of colorectal cancer in patients with hereditary nonpolyposis colorectal cancer.Cancer Detect Prev. 2000; 24: 137-142PubMed Google Scholar Potentially these interval cancers develop from missed adenomas. In addition, tumor progression may be extremely rapid because of the increased mutation rate after MMR loss.2Kinzler KW Vogelstein B Lessons from hereditary colorectal cancer.Cell. 1996; 87: 159-170Abstract Full Text Full Text PDF PubMed Scopus (4321) Google Scholar However, this last possibility contradicts observations that colorectal cancers are rare with congenital MMR deficiencies.7Parsons R Li GM Longley M Modrich P Liu B Berk T Hamilton SR Kinzler KW Vogelstein B Mismatch repair deficiency in phenotypically normal human cells.Science. 1995; 268: 738-740Crossref PubMed Scopus (297) Google Scholar, 8Ricciardone MD Ozcelik T Cevher B Ozdag H Tuncer M Gurgey A Uzunalimoglu O Cetinkaya H Tanyeli A Erken E Ozturk M Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1.Cancer Res. 1999; 59: 290-293PubMed Google Scholar, 9Wang Q Lasset C Desseigne F Frappaz D Bergeron C Navarro C Ruano E Puisieux A Neurofibromatosis and early onset of cancers in hMLH1-deficient children.Cancer Res. 1999; 59: 294-297PubMed Google Scholar, 10Vilkki S Tsao JL Loukola A Poyhonen M Vierimaa O Herva R Aaltonen LA Shibata D Extensive somatic microsatellite mutations in normal human tissue.Cancer Res. 2001; 61: 4541-4544PubMed Google Scholar With our model, cancers that appear shortly after negative examinations represent neither missed adenomas nor extremely rapid tumor progression. Instead, our model predicts that it will be difficult to prevent all HNPCC cancers by surveillance because the majority of oncogenic mutations occur during pretumor progression (Figure 1). Interval and noninterval cancers contain essentially the same numbers of MS mutations,5Tsao JL Tavaré S Salovaara R Jass JR Aaltonen LA Shibata D Colorectal adenoma and cancer divergence: evidence of multi-lineage progression.Am J Pathol. 1999; 154: 815-824Abstract Full Text Full Text PDF Scopus (53) Google Scholar, 6Tsao JL Yatabe Y Salovaara R Jarvinen HJ Mecklin JP Aaltonen LA Tavaré S Shibata D Genetic reconstruction of individual colorectal tumor histories.Proc Natl Acad Sci USA. 2000; 97: 1236-1247Crossref PubMed Scopus (123) Google Scholar, 11Kim KM Salovaara R Mecklin JP Jarvinen HJ Aaltonen LA Shibata D PolyA deletions in hereditary nonpolyposis colorectal cancer: mutations before a gatekeeper.Am J Pathol. 2002; 160: 1503-1506Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar indicating a sudden visible emergence of an HNPCC cancer is not unusual. Interval cancers do not arise de novo because pretumor progression starts from birth and an occult stem cell tumor progenitor may accumulate nearly all its required mutations at the times of negative examinations. Although our model suggests it will be difficult to prevent all HNPCC cancers, MS tumor histories suggest surveillance may help prevent cancer death. After transformation, a single cell must physically generate sufficient progeny for invasion and metastasis. MS tumor histories indicate advanced cancers tend to represent older clonal expansions (Figure 3B), suggesting more time is needed for invasion and metastasis. Therefore, although a cancer may suddenly emerge after pretumor progression, early detection may prevent its spread. Consistent with our model, surveillance does not prevent all HNPCC cancers, but interval cancers tend to be of low clinical stage.22Jarvinen HJ Aarnio M Mustonen H Aktan-Collan K Aaltonen LA Peltomaki P De La Chapelle A Mecklin JP Controlled 15-year trial on screening for colorectal cancer in families with hereditary nonpolyposis colorectal cancer.Gastroenterology. 2000; 118: 829-834Abstract Full Text Full Text PDF PubMed Scopus (1135) Google Scholar, 23de Vos tot Nederveen Cappel WH Nagengast FM Griffioen G Menko FH Taal BG Kleibeuker JH Vasen HF Surveillance for hereditary nonpolyposis colorectal cancer: a long-term study on 114 families.Dis Colon Rectum. 2002; 45: 1588-1594Crossref PubMed Scopus (193) Google Scholar, 24Renkonen-Sinisalo L Aarnio M Mecklin JP Jarvinen HJ Surveillance improves survival of colorectal cancer in patients with hereditary nonpolyposis colorectal cancer.Cancer Detect Prev. 2000; 24: 137-142PubMed Google Scholar Frameshift mutations in short mononucleotide repeats of coding loci such as TGFBR2 and BAX are common in MMR-deficient cancers.25Markowitz S Wang J Myeroff L Parsons R Sun L Lutterbaugh J Fan RS Zborowska E Kinzler KW Vogelstein B Brattain M Willson JKV Inactivation of the type II TGF-beta receptor in colon cancer cells with microsatellite instability.Science. 1995; 268: 1336-1338Crossref PubMed Scopus (2157) Google Scholar, 26Rampino N Yamamoto H Ionov Y Li Y Sawai H Reed JC Perucho M Somatic frameshift mutations in the BAX gene in colon cancers of the microsatellite mutator phenotype.Science. 1997; 275: 967-969Crossref PubMed Scopus (1238) Google Scholar, 27Duval A Hamelin R Mutations at coding repeat sequences in mismatch repair-deficient human cancers: toward a new concept of target genes for instability.Cancer Res. 2002; 62: 2447-2454PubMed Google Scholar Such frameshift mutations likely occur after MMR loss and their high frequencies suggest substantial progression after MMR loss. However, frameshift mutations may be frequent or biologically important, but not rate limiting15Peto R Epidemiology, multistage models, and short-term mutagenicity tests.in: Hiatt H Watson JD Winsten JD Origins of Human Cancer. Cold Spring Harbor Laboratory, Cold Spring Harbor1977: 1403-1428Google Scholar, 28Moolgavkar SH Luebeck EG Multistage carcinogenesis: population-based model for colon cancer.J Natl Cancer Inst. 1992; 84: 610-618Crossref PubMed Scopus (123) Google Scholar, 29Knudson AG Two genetic hits (more or less) to cancer.Nat Rev Cancer. 2001; 1: 157-162Crossref PubMed Scopus (821) Google Scholar because they arise quickly after MMR loss. Some frameshifts in MMR-deficient cancers may be passenger mutations because shifts in noncoding mononucleotide repeats are also frequent30Zhang L Yu J Willson JK Markowitz SD Kinzler KW Vogelstein B Short mononucleotide repeat sequence variability in mismatch repair-deficient cancers.Cancer Res. 2001; 61: 3801-3805PubMed Google Scholar and some TGFBR2 mutations lack selective value in cell culture assays.31Ilyas M Efstathiou JA Straub J Kim HC Bodmer WF Transforming growth factor beta stimulation of colorectal cancer cell lines: type II receptor bypass and changes in adhesion molecule expression.Proc Natl Acad Sci USA. 1999; 96: 3087-3091Crossref PubMed Scopus (45) Google Scholar Only certain mutations are likely critical for a tumor phenotype. MS mutations are neutral and in this analysis they merely serve as molecular odometers to measure time after MMR loss. If most oncogenic HNPCC tumor mutations accumulate after MMR loss during tumor progression, there could be a relationship between incremental changes in phenotype and time after MMR loss. However, there was no relationship between clinical cancer stage and the interval after MMR loss (Figure 3A). Instead, and consistent with our multistep model, MS histories indicate cancers in older individuals generally require more divisions to achieve the same number of oncogenic mutations before and after MMR loss (Figure 3, Figure 4). A pretumor progression model1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar illustrates how experimental mutation and epidemiological data may be integrated into a coherent plausible lifetime tumor history (Figure 1). Many parameters of the model are uncertain,1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar and therefore its conclusions represent a framework for analysis rather than a definitive description. Instead of a quiescent prelude before tumor progression, HNPCC tumor histories suggest MMR loss and most oncogenic mutations accumulate in stem cells during pretumor progression. Consistent with a conclusion that pretumor progression to cancer does not require elevated mutation rates or tumorigenesis,1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar most oncogenic HNPCC cancer mutations appear to accumulate before MMR loss in normal appearing colon. The identities of the mutations that precede MMR loss are unknown. However, pretumor progression predicts that mutations that confer stem cell survival advantages within niches will accumulate faster and more frequently. Interestingly, activating somatic β-catenin mutations are common in HNPCC tumors,32Miyaki M Iijima T Kimura J Yasuno M Mori T Hayashi Y Koike M Shitara N Iwama T Kuroki T Frequent mutation of beta-catenin and APC genes in primary colorectal tumors from patients with hereditary nonpolyposis colorectal cancer.Cancer Res. 1999; 59: 4506-4509PubMed Google Scholar but disruption of the mutant β-catenin allele in a MMR-deficient colorectal cancer cell line enhanced or had minimal effects on tumorigenicity,33Sekine S Shibata T Sakamoto M Hirohashi S Target disruption of the mutant beta-catenin gene in colon cancer cell line HCT116: preservation of its malignant phenotype.Oncogene. 2002; 21: 5906-5911Crossref PubMed Scopus (38) Google Scholar suggesting the mutation had a different role early in progression.34Chan TA Wang Z Dang LH Vogelstein B Kinzler KW Targeted inactivation of CTNNB1 reveals unexpected effects of beta-catenin mutation.Proc Natl Acad Sci USA. 2002; 99: 8265-8270Crossref PubMed Scopus (116) Google Scholar This β-catenin mutation (like others in HNPCC cancers32Miyaki M Iijima T Kimura J Yasuno M Mori T Hayashi Y Koike M Shitara N Iwama T Kuroki T Frequent mutation of beta-catenin and APC genes in primary colorectal tumors from patients with hereditary nonpolyposis colorectal cancer.Cancer Res. 1999; 59: 4506-4509PubMed Google Scholar), likely occurred before MMR loss because it was a nonrepeat sequence deletion rather than a mononucleotide repeat frameshift. β-Catenin and other members of the WNT signaling pathway appear to influence stem cell survival,1Calabrese P Tavaré S Shibata D Pre-tumor progression: clonal evolution of human stem cell populations.Am J Pathol. 2004; 164: 1337-1346Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar and the activating β-catenin mutation may have conferred its selective advantage during pretumor progression. In the next study35Kim KM Calabrese P Tavaré S Shibata D Enhanced stem cell survival in familial adenomatous polyposis.Am J Pathol. 2004; 164: 1369-1377Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar we illustrate heterozygous APC mutations that enhance niche stem cell survival in normal appearing FAP crypts. We thank Sylvia I. Lambrechts for her expert technical assistance.