We present a combined discovery study (from 2004) and a validation study (from 2014) in a prospective design (the transomic Advanced Study of Inflammatory Bowel Disease) where clinical, biochemical, histological and transcript data where retrospectively tested to identify biomarkers of clinical outcome 1 year after disease diagnosis of UC. Mucosal TNF transcripts showed high test reliability for predicting severe outcome after 1 year in UC in both studies but was not ideal to discriminate between mild, moderate and severe disease. Moreover, when the TNF transcripts were combined with histological activity (RHI) scores, the test improved its diagnostic reliability. Mucosal cut-off values for TNF and RHI scores determined in the calibration cohort displayed a high test performance with specificity of 0.99 and a diagnostic odds-ratio (DOR) of 54 in the prospective validation study. Thus, mucosal TNF transcript combined with a histological score at debut of disease can likely identify patients who experience severe outcomes during the first year. This is an important step towards personalizing treatment in IBD and may be used as a criterion for selecting candidates for top-down treatment of anti-TNF. However, this awaits further studies.
We have tested a broad spectrum of potential factors that could, alone, or in combinations, predict clinical outcome in the first year of diagnosis. The clinical outcomes were defined as the highest treatment level required for achieving disease remission during the first year of disease, in a step-up treatment approach. The broad/wide selection of variables including various combinations did not have the necessary precision to discriminate between mild, moderate and severe outcomes. However mucosal TNF transcript in combination with the histological RHI score was able to predict, with high precision, the most severe colitis outcomes needing biological or surgical treatment, within the first year of disease. The validated cut-off values (TNF ≥ 18000, RHI ≥ 9) showed a high specificity to predict severe outcome and a DOR as high as 54. From a clinical point of view, these cut–off values indicate a need of anti-TNF therapy during the first year after diagnosis with high reliability, and therefore of high clinical value and utility in the management of IBD/UC. In order to use a biomarker for selection for top-down treatment, a high PPV is necessary to avoid excessive use of biologics. Our proposed biomarker shows a PPV of 0.89 meaning that 9 out of 10 positives will be correctly identified as severe outcome.
A step-up treatment approach represents well-established international guidelines.8–10 One drawback of this approach is that patients in the severe outcome group often experience a period of poor response during the gradual escalation of treatment intensity until an adequate response is obtained. In some cases, one may lose an important window of opportunity for optimal effect of biologics leading to permanent structural damage and/or need of surgery. The impact of early treatment before development of severe disease is not completely investigated. However, the top down approach published by D’Haens et al indicated that immunosuppressive therapy was superior to a step-up approach in patients with Crohn’s disease.36 Moreover, it is well documented that induction of treatment to remission reduces later hospitalization, whereas conflicting results exist for colectomy in two studies.37, 38
The use of molecular data from the mucosa represents a novel approach and is an easily available tool, with high utility for clinicians to individually tailor therapy in UC. Endoscopic biopsies are routinely taken at diagnosis and surveillance of IBD. Thus, the logistics of measuring mucosal TNF transcript are simple, as biopsies are readily available and samples do not require freezing prior to analysis.31
Our study contributes with new knowledge in the scientific field of personalized therapy in UC.15 16 We know that treatment to remission improves long-term clinical outcome.39, 40 The main question is: Can a top-down therapy of the most severe forms of disease have an effect on the natural course of disease? This awaits future studies.
The strength of this prospective designed, combined discovery and validation study is that we have retrospectively searched for and validated biomarkers for treatment at debut of UC, using a broad search of clinical, histological and analytical factors including mucosal immune transcripts. Moreover, this is part of the transomic Advanced Study of Inflammatory Bowel disease (ASIB) study where parallel studies of the epigenome, transcriptome, proteome and metabolome are ongoing.33, 41–45. This transomic approach at debut of UC will be performed and correlated to long-term clinical outcome. Therefore, the upcoming transomic data from the ASIB study and from several ongoing studies such as the PREDICTS study will not only search for therapeutic but also prognostic and natural course biomarkers.17 The weakness of the study includes the lack of endoscopic diagnosis at one year, which would have given insight into endoscopic status and endoscopic remission rates according to treatment levels. Additionally, the decision to use or not use steroids at time of diagnosis is dependent on the subjective decision of the clinicians. This may be one explanation for the small differences detected between the mild and moderate treatment group.
In conclusion, the combined information of mucosal TNF transcription and histological score at debut of UC can predict severe outcome and the need for anti-TNF therapy. This is of great clinical utility and may contribute to a personalized medicine approach in UC.