Modeling genetic susceptibility to multiple sclerosis

Abstract

The main aim of this thesis was to investigate genetic and environmental factors and their role in the etiology of Multiple Sclerosis (MS) by using comprehensive registry data or novel computationally intense methods. To date, over 100 genes associated with MS have been identified, but how they interact in the risk for the disease is not yet fully understood. The presence of high prevalence clusters has led researchers to believe that there might be as yet unidentified rare variant involved in the disease etiology. In Paper I, we attempted to search for these rare variants by using a population based linkage approach, estimating haplotypes shared between individuals inherited by descent from some common ancestor. One significant hit was found on chromosome 19, but due to methodological problems the result should be interpreted with caution.

MS is commonly attributed high familial risks, decreasing with relatedness, which indicates a large genetic component involved in the disease etiology. In Paper II, nationwide registry data was used to reinvestigate the familial risks and estimate the proportion of genetics and environment contributing to disease etiology. The relative risks estimated were lower than usually reported, with a sibling relative risk of 7.1 and no significant differences between the sexes. The heritability was estimated to be 64% and the environmental 36% with a non-significant shared environmental component of 1%.

In Paper III, the women-to-men ratio for MS in Sweden was reinvestigated. MS is a disease more common in women than men, and an increase in the women-to-men ratio has been reported in several countries. However, a report from Sweden did not show this increase in women and Paper III extended this report using data from nationwide registers. An increase among women compared to men was identified, and when comparing against the previous study, an inclusion bias, presumably caused by a higher mortality rate among the oldest men, was identified.

One framework used to model complex diseases such as MS is the sufficent cause model, also known as Rothman's pie model. This model hypotehsizes that a disease can be caused by several mechanisms, or pies, each consisting of a set of different factors and when all factors are present they will inevitably cause disease. Paper IV extends this model into a stochastic version and presents an algorithm that can estimate the probability that an a priori suggested mechanism has caused disease in a certain individual. The algorithm showed high classification accuracy on synthetic data; however it needs further investigation of its properties.

In conclusion, this thesis revise the familial risks for MS to more moderate levels, with no differences between the sexes, and confirms the global trend of an increasing women-to-men ratio. No rare variants contributing to MS on population level were identified. We also present a probabilitic version of Rothman's pie model, showing promising results on synthetic data.