Search in the Abstract Database

Abstracts Search 2019

P485 Prediction Model Incorporating Pharmacokinetics Calculates Probability of Endoscopic Healing in Patients with ulcerative colitis Starting Infliximab Therapy

N. Vande Casteele*1,2, V. Jairath2,3,4, J. Jeyarajah2, P. S. Dulai1,2, S. Singh1, B. G. Feagan2,3,4, W. J. Sandborn1,2

1University of California San Diego, Department of Medicine, La Jolla, USA, 2Robarts Clinical Trials, Inc., London, Canada, 3University of Western Ontario, Medicine, London, Canada, 4University of Western Ontario, Epidemiology and Biostatistics, London, Canada

Background

Infliximab (IFX) is effective treatment for moderate to severe ulcerative colitis (UC), however baseline parameters associated with, and probability of achieving endoscopic healing during induction and maintenance therapy are unknown.

Methods

Data from the ACT-1 and -2 trials encompassing 484 IFX-treated UC patients were analysed. A two-compartment population pharmacokinetic model was used to calculate baseline IFX clearance (CL). The Mayo endoscopic score was available at Weeks (W) 0, 8 and 30. Three logistic regression prediction models were developed using the ACT-1 dataset and externally validated using the ACT-2 dataset. The models evaluated W0 variables for prediction of endoscopic healing (MES ≤ 1) at W8 and W30, and W8 variables for prediction of endoscopic healing at W30. An online tool to calculate the probability of achieving endoscopic healing in individual patients was also created.

Results

IFX CL, stool frequency, and rectal bleeding at W0 were independently associated with endoscopic healing at W8 with an area under the curve (AUC [95% confidence interval]) of 0.73 (0.66–0.79) and 0.67 (0.60–0.74) for the derivation and validation models, respectively. IFX CL, stool frequency, white blood cell count, and weight at W0 were independently associated with achieving endoscopic healing at W30 with an AUC of 0.68 (0.62–0.75) and 0.67 (0.61–0.74) for the derivation and validation models, respectively. Rectal bleeding, stool frequency, white blood cell count, and albumin at W8 were independently associated with achieving endoscopic healing at W30 with an AUC of 0.83 (0.78–0.89) and 0.78 (0.72–0.84), for the derivation and validation models, respectively. Odds ratios for the factors predictive of endoscopic healing are shown in Table 1.

Table 1. Odds ratios for W0 and W8 factors predictive of endoscopic healing in patients receiving IFX. Variable selection was based on univariable selection (p < 0.15) followed by a forward stepwise multi-variable logistic regression model (p < 0.1).

Patient-level probabilities for endoscopic healing at W8 and/or W30 can be calculated using a free online tool available at http://premedibd.com. The predicted probability of endoscopic healing at W8 for a hypothetical UC patient starting IFX therapy using the online tool is shown in Figure 1.

Figure 1. Probability of W8 endoscopic healing in a hypothetical UC patient. A population pharmacokinetic model uses sex and albumin to calculate W0 IFX CL, which is incorporated into the prediction model with stool frequency and rectal bleeding.

Conclusion

Three models were developed and externally validated to calculate the probability of endoscopic healing in individual patients with UC during IFX induction and/or maintenance therapy based on IFX CL, patient demographics and disease activity measures at W0 and/or W8.