Methodologic pitfalls in the determination of genetic anticipation: The case of Crohn disease

Background: The term genetic anticipation is used when genetically transmit ted disease manifests at increasingly younger ages with each succeeding gen eration: that is, if the offspring of patients develop the disease, they wi ll tend to do so at an earlier age than their parents. In some monogenetic disorders, genetic anticipation has a biological basis in expanded genetic triplet repeats; however, some have claimed that it occurs in polygenic dis orders, such as Crohn disease, in which its mechanism cannot be explained. Objective: To show how apparent changes in age at diagnosis of Crohn diseas e between generations, which could suggest genetic anticipation, can be an artifact of inadequate analysis based on age at diagnosis in cohorts that h ave not been followed for a sufficiently long time. Design: Comparison of ages at diagnosis of Crohn disease among different bi rth cohorts, before and after adjustment for observation time. Setting: Meyerhoff Center for Inflammatory Bowel Disease, Johns Hopkins Hos pital, Baltimore, Maryland. Patients: 928 consecutive outpatients with Crohn disease. Measurements: Trends in age at diagnosis of Crohn disease among birth cohor ts were determined by calculating Pearson correlation coefficients and perf orming Kaplan-Meyer analysis before and after adjustment for observation ti me. Adjustment for observation time was performed by ensuring that the time during which all included patients were at risk for Crohn disease was equa l and that all patients had developed disease by the end of the risk period . Results: Mean age at diagnosis decreased by approximately 5 years with each subsequent 10-year birth cohort on the basis of crude cross-sectional data that could suggest genetic anticipation between generations. However, afte r adjustment for observation time, the age at diagnosis decreased minimally , if at all, with each successive generation. Conclusions: Apparent genetic anticipation can be explained by observationa l biases without invoking any additional genetic influences. Claims for gen etic anticipation must be based on methods that properly account for the du ration of observation in all persons being studied.