NORDTREAT Prospective Study on Inflammatory Bowel Disease

The primary aim is to identify molecular markers (e.g. in the blood and stools) for discrimination between individuals diagnosed with inflammatory bowel disease (IBD) inclusive Crohn's disease (CD), ulcerative colitis (UC) and inflammatory bowel disease unclassified (IBD-U), and those without (non-IBD). Participants with suspected IBD at baseline, with various disease pathways, will be evaluated again using a 1-year cohort study. Diagnosis and clinical outcome will be evaluated at referral and after 1 year of observation.

The secondary aims are, in addition to the molecular information, to investigate whether the inclusion of information on clinical and lifestyle factors as well as combination hereof (e.g. gene-environment interaction analyses) can improve the predictive potential of identifying IBD and distinguish the prognosis.

Study design: A prospective Nordic multicenter study on prognostic factors for the diagnosis and characterization of IBD among patients referred to the hospital on suspicion of IBD. A panel of possible prognostic biomarkers for diagnostic purposes will be applied to all participants.

Setting: All patients referred due to a suspicion of IBD to the departments of gastroenterology in Odense University Hospital, Svendborg Hospital, Vejle Hospital, Esbjerg Hospital and potentially Hospital of Southern Jutland, Aabenraa will be invited to take part in the study. Participants will be included from January 2022 for a 1-year period or until 800 participants in the Nordic study (up to 400 in Denmark) have been included. The follow-up period is one-year including visits and questionnaires and an additional nine years of follow-up by the use of register data. Biological material will be obtained four times for participants with IBD, at week-2/0, and 12, 26 and 52 after the diagnosis has been established. Participants where IBD is not established (non-IBD) will only have biological material obtained at baseline (that is visit -2/0) and will have a clinical interview after 52 weeks. All participants are treated according to standard clinical practice by the clinical departments.

Clinical data consist of personal data, data on health and disease, lifestyle, laboratory measures, and disease activity scores including patient-reported outcome measures (PROMs), clinical assessments, and laboratory data. Each participant will fill out validated questionnaires on disease activity, quality of life, and lifestyle using electronic links.

Data management: Study data registered by clinicians, study nurses and technicians will be stored in the web-based case report form (eCRF) Viedoc (Viedoc™, Uppsala, Sweden) or REDCap (Open Patient data Explorative Network (OPEN) at Odense University Hospital) for the diet questionnaire. The questionnaires are in Danish and the participants will have access to the questionnaires via an electronic link sent to their personal, electronic mailbox (REDCap) or via an investigator provided link to MyViedoc. All data will be stored in secure research storage facilities.

Statistical methods: We will develop multivariable prediction models relating multiple predictors for a particular individual to the probability of or risk for the presence (diagnosis of IBD at baseline) or future occurrence (prognosis) of a particular outcome such as severe IBD within the first year. Predictors such as biomarkers are covariates, explored as prognostic factors (independent variables).

The primary hypothesis is that the final biomarker will define the participant population into two groups each consisting of 50%: a biomarker positive group whereof 80% will be diagnosed with IBD and a biomarker negative group whereof 20% will be diagnosed with IBD.

For a comparison of two independent binomial proportions using the likelihood ratio statistic with a Chi-square approximation with a two-sided significance level of 0.05, a total sample size of 800 assuming a ratio biomarker positivity-to-negativity of 1 to 1 has an approximate power of 100% when the proportions of being diagnosed with IBD are 0.8 and 0.2. If we assume that we will have 800 study participants, we will potentially ("rule of thumb") be able to build a statistical model with as many as 80 covariates in the multivariable model.

The associations of the suspected important biomarkers with other variables will be tested with non-parametric tests: with Spearman rank correlation (rs) for continuous variables, and the Wilcoxon rank-sum test or Kruskal-Wallis test, including a Wilcoxon-type test for trend across ordered groups where appropriate, for categorical variables. In general, logistic regression models will be used with individual marker as the exposure variables and the clinical response as the outcome (dependent variable). The analyses will be adjusted simultaneously for sex, age, and prescribed targeted therapy. Potential interaction between biomarker status (positive/negative) and specific drug type will be analyzed. Covariates (biomarkers) consist of various measures within genetics, transcriptomic, microbiomes, and proteomics.

Sample size considerations: Assuming that biomarker positivity constitutes 50% of the individuals enrolled at baseline, if our event rate is 10% on average, a total sample size of 800 patients (i.e. 400 biomarker positive) we will have a very good statistical power (99.7%) to detect a difference between proportions having surgery of 10% points (15% and 5%, respectively). If we decide to split the data set into two (2×400 individuals), in order to first build the model, and subsequently validate it in the second independent dataset, we will have 91.8% statistical power to detect a difference between groups. If the prognostic value of our biomarkers is not that effective separating the number of patients with severe IBD at week 52, we will still have more than 90% power to detect a difference between biomarker groups of 6% points (e.g. 10% and 4%, respectively).

Another consideration is the number of events (individuals having severe IBD at week 52) per variable (EPV) considered for inclusion in the multivariable model. For logistic regression modelling the EPV should be at least ten times the number of potential prognostic variables that could be included in the model. As a consequence of this logic, our expected sample size of 800 individuals (having 80 events) will, with reasonable confidence, allow us to create a multivariable model with up to 8 covariates simultaneously.

Project organization: NORDTREAT is part of a larger Nordic project (DK, SE, NO and IS) where regular meetings will be held between the partners. Collaborative research and material transfer agreements will be conducted with the national and international collaborators.

In addition to the scientific reporting of results, major findings with translational implications will be communicated to health professionals, patient organizations, public health policy makers, and to the general public through various media and news activities.