Family Studies in Primary Biliary Cirrhosis (PBC)

This proposal is predicated on cumulative data revealing a major role for genetic factors in PBC and the recent positional cloning of genes for Crohn's disease, type 2 diabetes and rheumatoid as well as psoriatic arthritis data provide the "proof-of-principle" that susceptibility genes for multigenic diseases can be identified by linkage and association strategies. Our group has access to >600 patients with PBC and these individuals plus family members provide an excellent resource for PBC gene discovery. Moreover, an International PBC Consortium we have initiated will serve not only to validate our findings but also to serve as a source for additional patient accrual should we be searching for genes with very small effects. Thanks to the remarkable advances in genotyping technology and sequence data emanating from the Human Genome Project, patient populations such as ours can now be fully mined so as to identify susceptibility genes for common multigenic diseases. By combining the patient resources of our group and that of other Canadian centers with available knowledge of the full genome sequence and the extensive information on SNPs within this sequence, as well as our expertise in high throughput genotyping and rapid computational approaches of genetic data, we are very well positioned to begin delineating the susceptibility alleles for PBC. As such alleles are identified, the information will then be used to define the molecular pathophysiology of this disease, to determine whether genetic markers can be used to predict risk and/or stratify patients in relation to prognosis and drug responsiveness, and ultimately, to identify novel targets for improved therapeutic intervention.

For the purposes of this study only patients with a definite diagnosis of PBC will be recruited ie:- those that had an elevated serum alkaline phosphatase level and positive AMA by ELISA or immunofluorescence (>1/40 titer) at the time of their initial diagnosis and who have had a liver biopsy which confirmed the diagnosis of PBC. After signing the consent form the patients will complete a questionnaire. With the patient's permission the coordinator will contact the patient's parents and siblings that are either affected or unaffected by PBC to ask them to participate in the study. A total of five tubes of blood will be taken from each participant in the study. The blood samples will be used to study the liver biochemistries and to study the inherited genetic material (DNA) in order to find out what causes PBC. In addition we are going to review the medical records for all the patients with PBC from TWH recruited to this study using the records held in Misys to examine and verify their medical and surgical history, past physical exams, laboratory tests and treatments. Each patient will also be asked to recruit unrelated (usually spouses) household members or friends who live locally to act as controls. They will then be treated similarly with regard to the consent form, questionnaire and blood work.

The plan is to have approximately 600 patients with PBC along with approximately 1800 of their relatives and other unrelated healthy volunteers to take part in this study - this is a multicentre study - sites being Toronto, London and Halifax.

Statistical Analyses/Power Calculations: The use of association analysis to identify susceptibility alleles is highly dependent on linkage disequilibrium (LD) between the SNP markers and the disease loci. To demonstrate the extent to which LD can be used to delineate PBC loci, our collaborator Dr. C. Amos has carried out calculations of the power to detect significant effects (at the p ≤ 0.05 level) in a case-control analysis assuming different amount of LD. Dr. Amos directs a statistical genetics group at MD Anderson Cancer Centre and is a longstanding collaborator of K. Siminovitch. He and his group members are also the statistical advisors to several disease-mapping consortia, such as NARAC, a US-based consortium for discovery of rheumatoid arthritis genes. To estimate the power of an association study to identify PBC susceptibility alleles, Dr. Amos used data from Talwalkar and Lindor, estimating PBC prevalence at 0.05% among adult females and a 15 fold increased risk for PBC among first-degree relatives of affected individuals. Assuming that mutation frequency for disease causation is 0.1% and also that PBC is caused by a single gene, then this level of risk among affected first-degree relatives implies the penetrance among carriers of PBC susceptibility alleles to be 6% and among non-carriers to be 0.4%. A single gene model for PBC, however, is simplistic and not consistent with available epidemiologic data. If instead, an assumption of 4 disease causative loci is made and risk is thus assigned to each one of these loci, then the genotype-specific risk would be 2.6%, the locus-specific relative risk would be a more modest 3.75, and the disease allelic frequency in carriers would be 10%. In the studies proposed here, we will begin by studying 300 cases and 300 controls for PBC association with specific candidate genes discussed above. The power to detect association in any case-control study is highly dependent upon the LD between the disease mutation and the SNPs at closely linked loci as well as the number of genes studied (about 40). If we assume a high level of disequilibrium, the disequilibrium coefficient (D') being 0.9 and, to be conservative, apply Bonferroni's correction for the simultaneous evaluation of 40 loci, then achieving 5% power in this study will require obtaining a single SNP test p-value of 0.00125. Based on a disease allele frequency of 10% (as calculated above), and the analysis of 300 cases and 300 controls, the power to detect a SNP marker-disease association would then be 84%. The study can be further powered by increasing numbers of cases and/or controls. Thus, to conserve resources, genotyping will be initially carried out using 300 cases and 300 controls, but an additional 300 controls will be further genotyped for any markers showing significant results at the p=0.01 level. Genome-wide association analyses will also be carried out initially on the 300 cases and controls. However, because a genome-wide screen is likely to yield many false positive loci-disease associations, any loci showing association at the 5% level will be confirmed using a replicate set of 300 cases and controls and/or trio families, once 200 such families are available.