Are genetic risk scores for arthritis and cardiovascular events predictive in arthritis patients on a population level?

The main aim of the project was to investigate to what extent known genetic risk factors for rheumatoid arthritis (RA) and ankylosing spondylitis (AS) contribute to explaining the risk of these diseases and of cardiovascular disease in individuals with RA. The research used data from the population-based Nord-Trøndelag Health Study (HUNT), so that the patients could be compared to a large number of controls without RA or AS. RA and AS lead to joint inflammation (arthritis) and are caused by environmental factors that interact with a strong genetic predisposition. These patients also have an increased risk of cardiovascular events, due to mechanisms related to their arthritis and other unfavorable risk factors such as smoking, blood lipid disturbances and hypertension. Currently, many genetic factors predisposing to these diseases are known, but their importance on a population level has not been well characterized. One of the project aims was therefore to investigate the role of genetic risk factors compared to non-genetic risk factors for disease development. We also wanted to study whether genetic risk factors for arthritis contribute to cardiovascular events in these patients, and whether genetic risk factors for cardiovascular events identified in the general population have the same effects in persons with RA. Individuals with RA or AS in HUNT were identified and compared to approximately 75,000 other HUNT participants. HUNT provided questionnaire data regarding health and lifestyle, measurement data including height, weight, and results of blood tests, as well as results from genotyping. The genetic data were summarized as risk scores. The scores give an estimate of each person?s risk of developing RA or AS, or of suffering a myocardial infarction if having a diagnosis of RA. Linkages were performed to the National Causes of Death Registry and the Myocardial Infarction Registry in Nord-Trøndelag, with a follow-up time of approximately 17 years. We first studied the incidence of RA and AS in the HUNT population, comparing the participants? own answers with diagnoses from hospital case files. The results showed that many who believe that they have one of these diagnoses actually do not. The incidence of RA in HUNT was comparable to similar populations, approximately 770 per 100,000 persons. The incidence of AS was approximately 260 per 100,000 persons, i.e. higher than previously reported in a mixed Norwegian population. A genetic predisposition is the strongest risk factor for AS. In the next study we constructed a genetic risk score including almost all known genetic risk variants at the time. The score was associated with AS development, but many persons with high scores did not have AS. The ability to predict AS was improved when the carrier state of the HLA-B27 genetic variant was also considered. The HLA-B27 test is already in clinical use in the diagnostic work-up of persons with AS symptoms. An important result was that even when combining the new risk score, HLA-B27 carrier state, and some clinical variables, prediction of AS was not accurate enough to be clinically useful in individual persons. Systematic knowledge is lacking on how to best select useful genetic variants to include in a risk score. Based on previous studies, we started with 269 genetic risk variants for RA and compared different ways to develop risk scores. A score including 27 variants was found to best for prediction of RA. If more variants were included, separating persons with and without RA became more difficult. The explanation is probably that only a few risk variants are common for all patients, whereas there the distribution of the other variants shows more variation. The ability to predict RA was clearly improved when non-genetic variables were also included. It was evident that none of the risk scores for RA were clinically useful for individual diagnostics, even when including the non-genetic variables. We also demonstrated that it is of great importance which statistical characteristics are employed to evaluate the usefulness of a genetic risk score. This has not always been emphasized in previous publications regarding genetic risk scores, so many published scores appear more promising than they are. In the last substudy, we investigated genetic risk factors for myocardial infarction in HUNT participants with and without RA. A manuscript has been submitted for evaluation in an international journal. The results will be made publicly available when the paper has been published.

Formålet med prosjektet var kunnskapsutvikling som på sikt kan gi bedre diagnostikk. Det gav økt forståelse av hvorfor det er vanskelig å finne genetiske risikoskårer som skiller mellom syke og friske og kunnskap om viktige metodologiske forskjeller når formålet er prediksjon av inflammatorisk leddsykdom versus å utrede underliggende genetiske mekanismer for sykdom. Prosjektet bekreftet hvilke statistiske egenskaper ved risikoskårene som er viktige for vurdering av klinisk anvendelighet, samt viktigheten av å teste skårene i populasjonsbaserte studier og andre settinger hvor sykdomsforekomsten er normal eller lett økt. Samlet bidrog prosjektet til å klargjøre viktige rammer for utvikling av bedre risikoskårer for inflammatorisk leddsykdom. Vi tror dette vil ha betydning for fremtidig arbeid på feltet både nasjonalt og internasjonalt gjennom våre samarbeidspartnere, og vil bidra til realisme og trolig utvikling av nye metoder som kan redusere de nåværende begrensningene.

Rheumatoid arthritis (RA) and ankylosing spondylitis (AS) are chronic inflammatory joint diseases posing a significant burden to patients, their families, and society. They are caused by an interaction between a strong genetic predisposition and environmental factors. RA patients also have increased mortality rates and higher risk of cardiovascular disease, including myocardial infarctions and strokes. The objectives of the project are to test whether risk scores based on published genetic risk variants may predict the risk of RA or AS, if genetic risk scores for RA and for cardiovascular events may predict the risk of mortality or myocardial infarctions in RA patients, and if prediction is improved when adding non-genetic variables. The project uses data from the population-based Nord-Trøndelag Health Study performed in 1995-7 and 2006-8, including questionnaire data on health and lifestyle, data from a clinical examination, and blood samples. Using hospital files, diagnoses were ascertained in 545 RA cases and 190 AS cases. Control data are available from approximately 36,000 individuals. Genotyping of a large number of genetic variants has been done, and linkages have been performed to the National Causes of Death Registry and the Myocardial Infarction Registry of Middle Norway. The main challenge is whether genotypes for all the previously published genetic variations will be available. The findings from the project will increase our understanding of how important the published genetic risk variants are relative to non-genetic risk factors in a general population, how important the contribution from disease-specific risk genes are to the overall risk of cardiovascular events in RA patients, and whether genetic risk factors for cardiovascular events identified in the general population have the same effects in RA patients. Improved predictive risk scores may also permit identification and preventive measures in individuals at high risk of disease development.