|Home | About | Journals | Submit | Contact Us | Français|
Direct germline analysis could be used to screen high-risk patients for DNA mismatch repair gene mutations associated with Lynch Syndrome. To further evaluate this potential strategy, we examined the prevalence of MLH1, MSH2 and MSH6 mutations in a population-based sample of young-onset (age < 50 years) CRC cases.
Young-onset CRC cases were randomly selected from three Colon Cancer Family Registry sites (Cancer Care Ontario, Mayo Clinic, University of Southern California). Extracted DNA from peripheral blood leukocytes was shipped to Myriad Genetic Laboratories (Salt Lake City, UT) for MLH1, MSH2 and MSH6 sequencing, and duplication/deletion analyses for MLH1 and MSH2. Results were reported as: deleterious/suspected deleterious, likely neutral, variant of uncertain significance, or no alteration detected. Germline data were compared to Amsterdam II criteria (ACII) and immunohistochemistry testing (IHC) in secondary analyses.
In 195 subjects, 11 had deleterious/suspected deleterious mutations (5.6%; 95% CI, 2.8%–9.9%), 12 had likely neutral alterations (6.2%; 3.2%–10.5%), 14 had variants of uncertain significance (7.2%; 4.0%–11.8%), 2 had both a likely neutral alteration and a variant of uncertain significance (1.0%; 0.1%–3.7%) and 156 had no alteration detected (80.0%; 73.7%–85.4%). Sensitivity, specificity, positive and negative predictive values for detecting deleterious/suspected deleterious mutations by ACII were 36.4% (4/11), 96.7% (178/184), 40.0% (4/10), and 96.2% (178/185) and by IHC testing were 85.7% (6/7), 91.9% (136/148), 33.3% (6/18) and 99.3% (136/137).
In this population-based sample of young-onset CRC cases, germline MLH1, MSH2 and/or MSH6 mutations were more prevalent than previously reported for CRC patients overall. Yet, since only about 1 in 20 young-onset CRC cases had confirmed deleterious/suspected deleterious mutations, further comparative effectiveness research is needed to determine the preferred Lynch Syndrome screening strategy for this high-risk group.
In the United States, approximately 6.5% of all incident colorectal cancers (CRCs) are diagnosed prior to age 50 years (1). Unfortunately, these “young-onset” CRC cases often present with advanced stage tumors (2) and contribute disproportionately to the estimated 800,000 person-years of life lost annually to CRC (3). Recent data from the Surveillance, Epidemiology and End Results (SEER) program demonstrate that CRC incidence rates are gradually increasing in younger adults (4). Improved clinical management strategies are promptly needed to address the effects of this discouraging trend.
Lynch Syndrome is caused by germline defects in one or more DNA mismatch repair (MMR) genes and characterized by young-onset cancer(s) arising in the colorectum and other target organs (5). Most Lynch Syndrome-associated CRCs are reportedly due to MLH1, MSH2 or MSH6 mutations (6–8). Several clinical algorithms, molecular tests and statistical models have been developed to predict the presence or absence of Lynch Syndrome (7, 10–15). Yet, despite these readily accessible tools, Lynch Syndrome patients continue to be widely under-recognized in clinical practice (16–18).
Direct germline mutation analysis, i.e., without clinical or molecular triage, represents a straightforward, potentially applicable strategy for Lynch Syndrome screening in young-onset CRC patients. However, the feasibility of this approach depends, in part, on the prevalence of MMR gene mutations in this defined patient group. To our knowledge, no prior studies have examined the prevalence of MLH1, MSH2, and MSH6 mutations using an unselected, population-based sample of young-onset CRC cases identified at multiple North American centers. In the present study, we utilized data and tissue resources from the Colon Cancer Family Registry (Colon CFR) (9) to address this knowledge gap.
Subjects were recruited through the Colon CFR, an international collaboration of six participating centers (University of Hawaii, Honolulu, HI; Fred Hutchinson Cancer Research Center, Seattle, SA; Mayo Clinic, Rochester, MN; University of Southern California Consortium, Los Angeles, CA; Cancer Care Ontario, Toronto, Canada; and University of Melbourne, Melbourne, Australia) organized in 1997 to create a comprehensive resource for genetic epidemiology studies. As described elsewhere (9), recruitment strategies differed across centers with respect to family ascertainment (population-based versus clinic-based) and CRC subject enrollment (all incident cases versus over-sampling by family history or early age of onset). For the current investigation, a random sample of exclusively population-based, young-onset (diagnosed prior to age 50 years) CRC cases who (a) were recruited during phase I of the Colon CFR collaboration (1997–2002) through Cancer Care Ontario (n=67), Mayo Clinic Rochester (n=67) and University of Southern California (n=67) and (b) had extracted DNA available for germline MLH1, MSH2 and MSH6 mutation analyses was included. Specifically, there was no preselection based upon family history or availability/results of tumor testing. From the initial sample of 201 young-onset CRC cases, 6 subjects were excluded based on tumor location in the appendix (n=4) or anus (n=2), rather than the colorectum. Demographic data, venipuncture samples, tumor blocks and pathology reports were collected according to established protocols (available at http://epi.grants.cancer.gov/documents/CFR/CCFR_BiospecimenCoreSOP_040607.pdf). CRC anatomic subsites were ascertained from pathology reports. Proximal cancers were defined as tumors located in the cecum, ascending colon, hepatic flexure, transverse colon and splenic flexure; distal cancers were defined as tumors located in the descending colon, sigmoid colon, rectosigmoid colon, and rectum, respectively.
Extracted DNA samples (from peripheral blood leukocytes) were shipped to Myriad Genetic Laboratories (Salt Lake City, UT) for full mutation analyses of MLH1, MSH2 and MSH6. The testing center was blinded to all clinical data associated with the specimens. Assay methods were identical to those used for clinical testing. Upon receipt, samples were assigned a unique barcode for robotic specimen tracking. DNA was amplified by polymerase chain reaction (PCR) for each subject. The amplified products were each directly sequenced in forward and reverse directions, using fluorescent dye-labeled sequencing primers: MLH1 (approximately 2,300 base pairs comprising 19 exons and approximately 560 adjacent non-coding intronic base pairs), MSH2 (approximately 2,800 base pairs comprising 16 exons and approximately 470 adjacent non-coding intronic base pairs) and MSH6 (approximately 4,080 base pairs comprising 10 exons and approximately 290 adjacent non-coding intronic base pairs). The analyzed, non-coding intronic regions of MLH1, MSH2 and MSH6 do not extend more than 20 base pairs proximal to the 5’ end and 10 base pairs distal to the 3’ end of each exon. Chromatographic tracings of each amplicon were analyzed by a proprietary computer-based review, followed by visual inspection and confirmation. Genetic variants were detected by comparison with a consensus wild-type sequence constructed for each gene. Sequence information of the coding region was derived from RefSeq NM_000249.3 (MLH1), NM_000251.1 (MSH2) and NM_000179.2 (MSH6). Intronic nucleotide information was derived from genomic sequences from NCBI - AC011816.17 (MLH1), AC079775.6 (MSH2), AC006509.15 (MSH6). All potential genetic variants were independently confirmed by repeated PCR amplification of the indicated gene region(s) and sequence determination as above. Large rearrangement testing for MLH1 and MSH2 was performed by Southern Blot Analysis in conjunction with multiplex ligation-dependent probe amplification (MLPA reagents from MRC Holland). For Southern analysis of MLH1 and MSH2, three sets of restriction digests were performed for each gene using individual enzymes or a combination of enzymes to yield appropriate banding patterns. Digested DNA was electrophoresed on agarose gels, transferred to nylon membranes, and hybridized with a gene-specific probe radiolabeled with 32P. Autoradiographs and phosphorimages were analyzed for the presence of novel bands and for fragment dosage, to assess the presence of deletions or duplications in exon regions. Southern blot and MLPA data were subjected to dual reviews involving technical personnel and at least one laboratory director for confirmation. Germline alterations were categorized as deleterious/suspected deleterious, likely neutral, or variant of uncertain significance, with specific sequence alterations recorded for each gene.
Immunohistochemistry (IHC) for MLH1, MSH2, and MSH6 protein expression was performed as previously described (10, 11) at Mayo Clinic Rochester (for MCR and USC cases) and at Cancer Care Ontario (for CCO cases), according to established protocols for clinical and research evaluation In brief, 4–6 µm tissue sections were cut from formalin-fixed, paraffin-embedded tissue blocks and stained using the avidin-biotin complex method of Ventana Medical Systems (BioTek Solutions buffer kit and DAB detection kit) and the Tech Mate 500 (Ventana) automated immunohistochemical stainer. Antibodies to MLH1 (clone G168-728, 1/250; Pharmingen, San Diego, Calif), MSH2 (clone FE11, 1/50; Oncogene Research Products, Cambridge, Mass) and MSH6 (clone 44, 1/500; Transductions Laboratories, Lexington, Ky) were employed. MMR protein expression was reported as present, absent, or inconclusive for each immunostain. Since tumor phenotype was not considered a primary study endpoint, complete IHC data were only available for 155/195 (79%) subjects.
All genotyping results were provided to the Colon CFR participating sites by Myriad Genetics. Data analyses were conducted by the Colon CFR investigators. Summary statistics were reported as mean (SD) or frequency (percent), as appropriate. Baseline demographic characteristics were compared across the three Colon CFR participating sites using the Student’s t test or chi-square test, as appropriate. Prevalence estimates for germline MMR gene alterations, in aggregate (i.e., finding present in one or more genes) and for MLH1, MSH2 and MSH6 separately, were defined a priori as the primary study endpoint, with accompanying 95% exact binomial confidence intervals (CI) reported. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for detecting deleterious/suspected deleterious MMR gene mutations were also estimated based on the Amsterdam II criteria (ACII ; fulfilled vs. not fulfilled) and IHC testing for MLH1, MSH2 and MSH6 protein expression (any absent vs. all present), as secondary analyses in subjects for whom relevant data were available. All statistical tests were performed two-sided (alpha level 0.05) using SAS version 9.0 software.
One hundred ninety-five subjects were included in the final analysis cohort, with 91 (47%) men and 104 (53%) women. The mean (standard deviation) age at CRC diagnosis was 42.9 (+/− 6.1) years. Subject characteristics were similarly distributed across the three participating Colon CFR sites, with the exception of age at CRC diagnosis (Table 1).
Germline MLH1, MSH2 and MSH6 sequencing data were obtained for 195 (100%), 195 (100%) and 189 (97%) subjects, respectively. Specific MMR gene sequence alteration data are provided in Table 2. Results were interpreted as deleterious/suspected deleterious mutations in 11 (5.6%; 95% CI, 2.8–9.9%%) subjects, likely neutral alterations in 12 (6.2%; 95% CI, 3.2%–10.5%) subjects, variants of uncertain significance in 14 (7.2%; 95% CI, 4.0–11.8%) subjects, both likely neutral alteration and variant of uncertain significance in 2 (1.0%; 95% CI, 0.1%–3.7%) subjects, and no alteration detected in 156 (80.0%; 95% CI: 73.7%–85.4%) subjects. The prevalence of deleterious/suspected deleterious mutations was not significantly associated with Colon CFR participating site: 3/65 (4.6%; 95% CI, 1.0%–12.9%) subjects from CCO, 3/65 (4.6%; 95% CI, 1.0%–12.9%) subjects from MCR and 5/65 (7.7%; 95% CI, 2.5%–17.0%) subjects from USC (p=0.79). Subjects with deleterious/suspected deleterious mutations were of comparable age to subjects with other variant or normal sequencing results (mean [SD] of 38.8 [9.6] versus 43.2 [5.8] years, respectively; p=0.19). The prevalence of deleterious/suspected deleterious mutations was also similar in men (8.8% [8/91]; 95% CI, 3.9–16.6%) and women (2.9% [3/104]; 95% CI, 0.6%–8.2%) (p=0.07). Conversely, among subjects for whom CRC anatomic subsite was known, deleterious/suspected deleterious mutations were more commonly observed (p<0.001) in those with proximal tumors (18.2% [8/44]; 95% CI: 8.2%–32.7%) than distal tumors (2.3% [3/130]; 95% CI: 0.5–6.6%).
Self-reported family history data were available for all cases, and 10/195 (5.1%) fulfilled ACII. Sensitivity, specificity, PPV and NPV estimates for detecting subjects with deleterious/suspected deleterious mutations by ACII were 36.4% (4/11), 96.7% (178/184), 40.0% (4/10) and 96.2% (178/185). Complete IHC data were available for 155 (79.5%) cases, and absent expression of at least one MMR protein was detected in 18/155 (12%) subjects, distributed as: 10 MLH1 deficient only (6.5%), 5 MSH2 and MSH6 deficient (3.2%) and 3 MSH6 only deficient (1.9%). Sensitivity, specificity, PPV and NPV estimates for detecting subjects with deleterious/suspected deleterious mutations by IHC testing were 85.7% (6/7), 91.9% (136/148), 33.3% (6/18) and 99.3% (136/137).
In this multi-center study of young-onset CRC cases, direct germline analyses of MLH1, MSH2 and MSH6 yielded a prevalence of 5.6% for deleterious/suspected deleterious mutations. To our knowledge, this study represents the first description of non-triaged MMR gene testing in a North American, population-based sample of young-onset CRC cases. Encouragingly, our findings are consistent with results from a prior, population-based study in Scotland that reported an MMR gene mutation prevalence of 4.4% for CRC cases diagnosed prior to age 55 years (13). Conversely, the reported prevalence of MMR gene mutation carriers among CRC patients overall (i.e., irrespective of age) is 2.2%, even after molecular pre-screening (14). Together, these data emphasize the need for heightened awareness of possible Lynch Syndrome in the context of young-onset CRC.
Aggressive CRC surveillance programs appear to reduce morbidity and mortality for MMR gene mutation carriers (15–17). Yet, widespread application of existing Lynch Syndrome screening tools has been compromised by inadequate provider knowledge, inattention to family history, inaccessible laboratory expertise, and other factors. In addition, the screening performance characteristics of clinically-based algorithms, such as the Amsterdam criteria (I or II) or Bethesda Guidelines (original or revised) are suboptimal (18,19). Even within our relatively homogeneous cohort of young-onset CRC cases, the ACII failed to detect a majority of subjects (6/11; 54.5%) with deleterious/suspected deleterious MMR gene mutations. To overcome the shortfalls associated with these clinically-based algorithms, Hampel and colleagues have proposed that all CRC patients, regardless of age, undergo primary IHC testing for MMR protein expression, with subsequent gene mutation and/or methylation analyses determined by the IHC results (23). In our study, IHC testing yielded 1 false negative (IVS14+2T>C alteration in MLH1) and 13 false positive results in secondary analyses of 155 young-onset CRC cases for whom complete protein expression data were available. Possible contributing factors to the observed specificity estimate (91.9%) include epigenetic modification (i.e., MLH1 promoter hypermethylation), recently described mutations not detectable by the applied methodology (e.g., deletion in TACSTD1 ) or mutations in other non-analyzed genes that interact with the MMR proteins tested. By extrapolation, IHC triage would have permitted detection of the large majority of MMR gene mutation carriers in our subject sample, with a marked reduction in the proportion of young-onset CRC cases (18/155; 11.6%) for whom targeted or complete germline analyses were indicated.
Several statistical models have also been developed to assist in selecting high-risk CRC patients for MMR gene mutation analyses. Notable examples include the Leiden Model, PREMM1,2, MMRpredict and MMRpro (6, 13, 21–23). Using a consecutive series of 725 CRC cases (< 75 years of age at diagnosis; 18 confirmed MMR gene mutation carriers) identified through the Newfoundland Colorectal Cancer Registry, Green, et al. evaluated and directly compared the discriminatory capacity of the Leiden Model, PREMM1,2, MMRpredict and MMRpro (24). At a defined sensitivity threshold of 94%, these predictive models exhibited greater specificity (range 61–88%) than the revised Bethesda Guidelines (51%), with the MMRpredict model demonstrating the most favorable performance characteristics. Interestingly, development of the MMRpredict model was based on data from CRC cases diagnosed at age < 55 years, implying that this model may be particularly beneficial in stratifying Lynch Syndrome risk for young-onset CRC cases. Unfortunately, limited availability of some key input parameters (i.e., family history of endometrial cancer) did not allow us to fully assess these statistical models in our study.
Major strengths of our study include the population-based, multi-center design with randomly identified, clearly defined young-onset CRC cases, reducing the possibility of referral or selection bias; MLH1, MSH2 and MSH6 germline analyses using well-established laboratory techniques, decreasing the potential for misclassification of mutation status; and universal gene testing, which permitted a more accurate description of the mutation prevalence than would have been obtained through triaged germline analyses. As noted above, the germline mutation prevalence reported from our study is consistent with limited existing data (13), lending credence to the external validity of our findings. One acknowledged weakness of our study is the absence of more comprehensive molecular marker data (such as microsatellite instability phenotype, MLH1 promoter methylation status, BRAF mutation testing, etc.) to further complement the reported germline analyses. Practical and technological constraints at the time the present study was initiated did not permit routine availability of such data. Since our primary aim was to estimate the prevalence of MLH1, MSH2, and MSH6 mutation carriers in an unselected, population-based case sample, we did not want to introduce possible selection bias by requiring ancillary laboratory data for study inclusion. Of note, further molecular characterization of Colon CFR case subjects is ongoing and will be reported in future studies.
Other MMR genes such as PMS2 have been associated with Lynch Syndrome, but were not assessed in our study. Although not definitively established (5, 14, 25–29), MLH1, MSH2 and MSH6 reportedly account for around 95% of all mutations observed in Lynch Syndrome families, suggesting that additional germline analyses would have had limited effect on our mutation prevalence estimates. Duplication/deletion analyses for MSH6 were not conducted in the small number of young-onset CRC cases for whom deficient protein expression was detected by IHC, which could have contributed to a slight underestimation of the observed prevalence estimates as well. Importantly, likely neutral alterations, variants of uncertain significance, or both were detected in 6.2%, 7.2% and 1.0% of our young-onset CRC cases, respectively. Such sequence alterations can pose clinical and psychological challenges with respect to cancer risk assessment (30, 31), particularly if additional protein expression and/or other molecular data are not available. These issues would need to be carefully considered for any screening strategies that apply germline mutation analyses as the primary testing modality.
In summary, data from our study and others (13) suggest that only about 1 in 20 young-onset CRC cases are attributable to germline mutations in MLH1, MSH2, or MSH6. Yet, a more systematic approach to Lynch Syndrome screening seems appropriate in this patient population, since the prevalence of MMR gene mutations is considerably higher than reported for CRC patients overall, and early detection of Lynch Syndrome kindreds has been shown to provide distinct clinical benefits. Nonetheless, given the relatively low proportion of MMR gene mutation carriers observed in our study, further comparative effectiveness research is needed to determine the most appropriate strategy for Lynch Syndrome screening in this high-risk patient group.
The authors gratefully acknowledge funding for and performance of the reported gene mutation analyses by Myriad Genetic Laboratories (Salt Lake City, UT).
Grant Support: The Colon Cancer Family Registry is supported by NIH National Cancer Institute grants CA074783 (Ontario Registry for Studies of Familial Colorectal Cancer), U01 CA074800 (Mayo Clinic Cooperative Family Registry for Colon Cancer Studies) and U01 CA074799 (USC Familial Colorectal Neoplasia Collaborative Group), as well as U01 CA097735 (Australian Colorectal Cancer Family Registry), UO1 CA074794 (Seattle Colorectal Cancer Family Registry), and U01 CA074806 (University of Hawaii Colorectal Cancer Family Registry). The study sponsor had no role in the study design or collection, analysis, and interpretation of the data.
Funding for and conduct of the gene mutation analyses described in this report were provided by Myriad Genetic Laboratories, Salt Lake City, UT.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: Dr. Limburg served as a consultant for Genomic Health, Inc. from 8/12/08-4/19/10.
Author Roles: Study concept and design (SG, RWH, JAB, SNT, NML); acquisition of data (PJL, SG, RWH, JAB, SNT, NML); analysis and interpretation of data (PJL, WSH, HHC, SG, RWH, JAB, GC, MOW, SNT, NML); drafting of the manuscript (PJL, WSH, HHC, SG, RWH, JAB, GC, MOW, SNT, NML); critical revision of the manuscript for important intellectual content (PJL, WSH, HHC, SG, RWH, JAB, GC, MOW, SNT, NML); statistical analysis (PJL, WSH, HHC); obtained funding (SG, RWH, JAB, NML)
Transcript Profiling: Not applicable.
Writing Assistance: None.