P388 Prediction model to safely CEASE anti-TNF therapy in Crohn’s disease: validation of a predictive diagnostic tool for the cessation of anti-TNF treatment in CD in a Dutch population

Abstract

Abstract Background Tools for patient identification to safely cease anti-TNF therapy in Crohn’s disease (CD) patients are urgently needed. After an individual participant data meta-analysis (IPD-MA) a predictive diagnostic tool has been developed for cessation of anti-TNF therapy in CD. This study aims to validate this tool. Methods A retrospective study was conducted, in 16 Dutch Hospitals, of CD patients in whom anti-TNF therapy was ceased. Inclusion criteria were anti-TNF therapy use >6 months, start of anti-TNF therapy due to luminal CD and remission as an indication for cessation. Collected baseline demographic, clinical, biochemical, treatment and imaging data were included; age, gender, smoking, Montreal classification, disease- and remission duration, history of surgery, type of anti-TNF medication, previous or concomitant immunosuppressant, thiopurines level, previous anti-TNF therapy, anti-TNF therapy duration, haemoglobin, leukocytes, thrombocytes, albumin, C-reactive protein, anti-TNF serum concentration, anti-infliximab/adalimumab antibodies, remission at MRI/endoscopy, additional stop reason other than remission. The primary outcome was documented relapse of CD that necessitated (re)introduction of biologicals, corticosteroids or immune-suppressants or surgery. Results A total of 523 CD patients (333 females (63%), median age 40 years (IQR 32 – 53)) were included. 293 (56%) patients experienced a relapse after anti-TNF cessation after a median follow-up of 30.2 months (IQR 15–51). The relapse rate was 33% (95% CI 31–34) and 53% (95% CI 52–53) after 1 and 2 years, respectively. The discriminative ability of the prediction model in this external validation cohort (Table 1) equalled that of a previous IPD-MA with a C-statistic of 0.59. An update of the model with faecal calprotectin resulted in a C-statistic of 0.60 [0.55–0.63] and a reported calibration slope of 0.69. Conclusion A previously developed predictive diagnostic tool to safely cease anti-TNF therapy in CD has been validated, however, showed moderate performance in this external cohort. A further update of the model with biochemical and histological data is necessary to improve our ability to adequately select patients for cessation of anti-TNF therapy and is currently being performed.