A Predictive Model for the Development of High-Responding Inhibitor In Children with Severe and Moderate Hemophilia A- Results of a Multicenter Cohort Study

Abstract Abstract 710 Background: Pediatric cohort studies in previously untreated [PUPs] children with hemophilia A [HA] identified potential risk factors for clinical meaningful inhibitor development [IH], among them HA severity < 1%, high risk gene mutations [HRS], high dose FVIII replacement therapy [HD] and use of recombinant FVIII concentrates [rFVIII]. The aim of the present study was to develop a simple score to rate inhibitor development in HA children. Patients and methods: 184 consecutively ascertained PUPs [enrollment between 1982 and 2008] with HA [severity < 5%] followed over a period of 200 exposure days were randomly selected into two groups [G1; G2] of 92 patients each. Three variables [HRS: 2 points], [HD: median single dose > 30 IU/kgbw within 8 to 12 weeks after therapy start] and [rFVIII] 1 point each, were incorporated into a risk assessment model, which was i) evaluated in patient group G1 [pilot cohort: patients not included in the CANAL study] and ii) validated in the second group [G2: confirmation cohort]. The risk discrimination cut-off was calculated using two independent statistical methods [ROC-analysis: MedCalc 11.0.0; CART regression tree: CART 6.0]. In addition, model specific cumulative IH-free survival [Cox regression analysis: hazard ratio (HR) and 95% confidence interval (CI)], negative predictive values [NPV], sensitivity and specificity were calculated. In Cox regression analysis adjustment was performed for i) age at first symptomatic bleed [proxy for early start of prophylaxis], ii) severity of factor VIII-deficiency, and iii) for treatment periods [1982-2008: divided into 3 to 5 year intervals]. Results: The risk discrimination cut-off in the pilot cohort G1 was calculated as “2”. Model sensitivity was 78.8 % and specificity was 49.8% respectively. In the confirmation cohort G2 seventeen children with HA developed a high responding IH [18.5%]: 16 out of 52 in children with a risk score > 2 [30.8%] and one out of 40 with a risk score < 2 [2.5%: p<0.001]. Compared to children with a risk score < 2 the cumulative IH-free survival in HA patients with a risk score > 2 was significantly reduced with an adjusted HR [CI] of 11.2 [1.5-83.5]. The corresponding NPV to rule out a high responding IH development in the confirmation cohort was 92% [80.5-98.5]. Conclusion: The model is able to rule out the development of high responding IH in white children with HA < 5% with good discriminative ability. Thus the model can be instrumental to individualize treatment strategies in HA patients. The score needs to be validated in other populations to test its generalizability. Disclosures: No relevant conflicts of interest to declare.