There are different levels of evidence for characterizing genetic etiologies of diseases. The first recognition is simply case reports of single families with a striking aggregation of these diseases. The next level of evidence is often cross sectional or case control studies examining increased aggregation within family members. One of the stronger forms of clinical evidence are studies comparing concordance rates of disease between monozygotic and dizygotic twins. The advantage to this study design is that it may control age as well as environment but these studies are not always feasible. The toughest investigation, and one that is almost impossible in complex genetic heterogeneous diseases, is a segregation analysis examining different models of inheritance in a large collection of pedigrees. And finally, the most convincing proof is a genetic linkage analysis actually identifying evidence for a chromosomal locus or the genetic mutation itself that confers susceptibility to a particular disease. We’ll discuss the levels of evidence for each of these in gastroesophageal reflux and in Barrett’s cancers.
Again, as I mentioned the initial suggestion for a genetic etiology comes from a number of case reports in the literature of families that show a striking aggregation of GERD, Barrett’s, and esophageal adenocarcinomas.
The next level of evidence comes from a cross sectional study in which we compared 58 Caucasians with Barrett’s and esophageal cancer with 106 controls with GERD. The Barretts’ cancer cases were of course older and more likely to be males. The striking finding was that the Barrett’s and esophageal cancer cases reported a positive family history for Barrett’s or esophageal cancer significantly more often than the reflux controls.
Our central hypothesis is that Barrett’s esophagus and its associated cancers are complex genetic diseases. We believe that combined genetic and environmental influences are responsible for the development of these diseases. An inherited susceptibility to develop Barrett’s is likely present in a subset of patients.
Study endoscopies identified Short segment BE in 2 relatives of index patients with sporadic disease but no long segment BE or cancers. Study endoscopy identified short segment BE in 5 family members, long segment BE in 5 family members, and esophageal adenocarcinoma in 1 family member of an FBE family. In total, Barrett’s epithelium was identified in 6% of sporadic relatives versus 41% of FBE relatives, a significant difference.
The real coup through this study has been the development of this large family. The symbols here are Barrett’s shaded in the lower left, esophageal cancer shaded on the upper right. GERD shaded on the upper left, other cancer shaded on the lower right, and short segment Barrett’s as a red dot in the center. We identified this family through this proband whose father died of esophageal cancer. Soon after we enrolled them this cousin developed meastatic esophageal adenocarcinoma at age 40; we then screened his uncles and found two affected with Barrett’s. His uncle was then talking with a cousin in Florida who said she had Barrett’s. Through this cousin we identified this large family also in Cleveland. We have identified Barrett’s in 3 members of this large family now. We have also screened a number of family members on this side and found short segment Barrett’s in two others. Clearly a pattern that is strongly suggestive of dominant inheritance.
In the whole sample, the overall prevalence of FBE is 7.3%.
SEGREG compares one susceptibility type models that is no genetic transmission; two susceptibility type models which are Mendelian dominant or recessive; and three susceptibility type models that is a co-dominant disease. It also allows one to add polygenic loci that have non-Mendelian transmission. Along with estimates of parameters at the maximum likelihood the output provides calculated Akaikes A information content. Lower AIC reflect a better fitting transmission model.
We analyzed 881 pedigrees accrued at several centers. The trait was defined as the confirmed presence of Barrett’s esophagus, esopahgeal adenocarcinoma, or esophagogastric junctional adenocarcinoma. The analysis assumed single ascertainment in a proband sampling frame. Basic models estimated were environmental only, polygenic only, pure dominant/recessive without a polygenic component, polygenic loci plus a dominant or recessive locus, pure additive only, and polygenic loci plus an additive locus
The one susceptibility sporadic or environmental disease model had an AIC of 1437.41, which decreased to 1405.25 when sex was added as a covariate. It decreased to 1392.43 when founder status was added as a covariate. And decreased to 1365 when sex and founder were both added as covariates into the model. Thus, gender and founder status were used as covariates in all subsequent analyses.
All two susceptibility models and three susceptibility models had lower AICs and fit the data clearly better than the one susceptibility envrionmental model. The best two susceptibility model without a polygenic component was a recessive model, which had an AIC of 1311. However, the models improved with the addition of up to three polygenic loci. Including three polygenic loci the model with a strikingly low AIC of 1,032 was the model that included an autosomally dominant Mendelian trait. The three susceptibility or additive models had similar AICs when they included 3 polygenic loci. However, homogeneity across generations was not supported by these additive models and the additive models were no better than the dominant model.
In conclusion our segregation analysis found that a random environmental model is grossly insufficient to explain Familial aggregation of Barrett’s esophagus, esophageal adenocarcinoma, and esophagogastric junctional adenocarcinoma; An incompletely dominant model that includes covariates for sex and founder status along with 3 polygenic locis provides the best fit to the observed data. In this model, the penetrance of the dominant allele is incomplete equaling 0.1005; the sporadic rate of the allele in this model is estimated at 0.0012. The calculated relative risk of this dominant allele is 82.5, Therefore, these results support a linkage analysis to search for rare dominant incomplete Mendelian alleles responsible for Familial Barrett’s esophagus in selected families.
Using only a portion of the previous study results
Peaks are around 10-20 cM wide- at least for these 2 chromosomes
Endoscopy in Gastrointestinal Oncology - Slide 2
Familial Barrett’s Esophagus & Esophageal Adenocarcinoma Amitabh Chak, MD Professor of Medicine & Oncology Case Western Reserve University
Pohl, H. et al. J Natl Cancer Inst 2005;97:142-146 EAC melanoma prostate breast/lung colon
Why Postulate A Genetic Basis? <ul><li>Sex (male) and race (white) predominance </li></ul><ul><li>BE restricted to a subset of GERD pts </li></ul><ul><li>Diseases with undefined etiologies, especially cancers, are likely “ Complex” (genetic + environmental) </li></ul>
Evidence for Genetic Etiology <ul><li>Case Reports of Familial Clustering </li></ul><ul><li>Cross Sectional or Case Control Studies of Familial Aggregation </li></ul><ul><li>Monozygotic vs. Dizygotic Twin Studies </li></ul><ul><li>Segregation Analysis </li></ul><ul><li>Linkage Analysis </li></ul><ul><li>Mapping of Genetic Mutation </li></ul>
Case Reports of Familial Clusters Reference Proband GERD/BE/EAC Crabb et al., Ann Int Med, 1985 80 BE 3 generations, 7/12 GER, 4/7 BE, multiple malignancies. Prior et al., Hepatogastro, 1986 66 BE Two sisters Jochem et al., Gastro, 1992 67 EAC 3 generations, 5 GER, 6 BE, 3 EAC Eng et al., Cancer Epi Bio Prev, 1993 EAC 3 generations, many GER(2 complicated GER), 7 BE, 2 EAC Fahmy et al., Am Jl Gastro, 1993 BE 4 families, multiple members with GER and BE
Pilot Cross Sectional Study – Familial Aggregation of BE/EAC (Chak et al., Gut) p<0.005 p<0.001
CENTRAL HYPOTHESIS <ul><li>BE and associated cancers are complex genetic diseases. </li></ul><ul><li>Combined genetic and environmental influences are responsible. </li></ul><ul><li>An inherited susceptibility to develop BE is present in a subset of affected patients. </li></ul>
Recruitment Of Pedigrees <ul><li>FBE Consortium </li></ul><ul><ul><li>University Hospitals Case Medical Center </li></ul></ul><ul><ul><li>Cleveland Clinic/HUP </li></ul></ul><ul><ul><li>Fred Hutch/UW </li></ul></ul><ul><ul><li>Creighton </li></ul></ul><ul><ul><li>Johns Hopkins </li></ul></ul><ul><ul><li>Mayo </li></ul></ul><ul><ul><li>U of Arizona </li></ul></ul><ul><ul><li>University of North Carolina </li></ul></ul>
Prevalence of FBE (Chak et al., CEBP 2006) <ul><li>30 (7.3%) of 413 probands studied prospectively have a confirmed affected first degree or second degree relative (BE or EAC): </li></ul><ul><ul><li>BE: 6.2% </li></ul></ul><ul><ul><li>EAC: 9.5% </li></ul></ul>
Age of Incidence Study <ul><li>Genetic diseases often have an earlier age of onset </li></ul><ul><li>Age of incidence of BE cannot be defined </li></ul><ul><li>Aim - To compare Age of Onset of Familial Barrett’s Cancer with “Sporadic” Cancers </li></ul>
Age of Cancers in Families with 3 or more affecteds Mean Age is 58.7 vs. 62.6 years, p < 0.05
Segregation Analysis <ul><li>A statistical test of Mendel’s first law </li></ul><ul><li>Tests whether the aggregation of disease in families is consistent with an environmental factor or transmission of a factor according to Mendelian genetic models </li></ul>
Models in SEGREG <ul><li>One susceptibility type – no genetic transmission </li></ul><ul><li>Two susceptibility types – Mendelian dominant or recessive </li></ul><ul><li>Three susceptibility types – Co-dominant </li></ul><ul><li>Can also add polygenic loci with non-Mendelian transmission </li></ul><ul><li>Lower Akaike’s A information content (AIC) = better fitting model </li></ul>
SEGREG – Preliminary Results <ul><li>881 pedigrees accrued at several centers </li></ul><ul><li>Trait defined as confirmed BE (> 1 cm), EAC, EGJAC </li></ul><ul><li>Analysis assumed single ascertainment </li></ul><ul><li>Basic models were environmental only, polygenic only, dominant/recessive no polygenic, polygenic plus dominant/recessive, additive only, polygenic plus additive </li></ul>
CONCLUSIONS <ul><li>Random environmental model insufficient to explain familial aggregation of BE, EAC, and EGJAC </li></ul><ul><li>Incompletely dominant model with covariates for sex and founder status plus 3 polygenic loci provide best fit </li></ul><ul><li>Penetrance of dominant allele = 0.1005, sporadic rate = 0.0012 </li></ul><ul><li>Relative risk of dominant allele = 82.5 </li></ul><ul><li>These results support a linkage analysis to search for rare dominant incompletely penetrant allele(s) responsible for FBE </li></ul>
Linkage <ul><li>Relates sharing of chromosomal regions to sharing of trait </li></ul><ul><li>Model based methods define explicit relationship between phenotype and genotype </li></ul><ul><li>Model free methods test for increased sharing among affected individuals </li></ul>
Romero et al. (DDW 2006) <ul><li>Genotyped 278 people in 31 families using microsatellite markers </li></ul><ul><li>Performed Model Based Linkage Analysis using various traits including ECA, BE, and GERD </li></ul><ul><li>Identified linkage with BE on Chr 2, lod score of just over 3 </li></ul>
FBEC linkage study <ul><li>-53 individuals typed in 15 pedigrees </li></ul><ul><ul><li>Average 3.5 per pedigree </li></ul></ul><ul><ul><li>All Caucasian pedigrees </li></ul></ul><ul><ul><li>68% male </li></ul></ul><ul><ul><li>39 affected (BE or EAC), 14 unaffected </li></ul></ul><ul><ul><li>33 sibling pairs: </li></ul></ul><ul><ul><ul><li>21 concordantly affected pairs </li></ul></ul></ul><ul><ul><ul><li>11 discordant pairs </li></ul></ul></ul><ul><ul><ul><li>1 concordantly unaffected pair </li></ul></ul></ul><ul><li>-100K Affymetrix Array </li></ul>
Summary of Family Studies <ul><li>BE and EAC proposed to be complex genetic diseases manifested as FBE. </li></ul><ul><li>FBE is restricted to a subset. The proportion of probands with confirmed FBE is less than 10%. </li></ul><ul><li>Pedigree analysis suggests FBE is caused by segregation of a transmissible (genetic) factor. </li></ul><ul><li>Linkage analysis has identified putative chromosomal regions linked to BE and EAC. </li></ul><ul><li>Sequencing is required to find the genetic variants that confer susceptibility to BE and EAC. </li></ul>