Background. Replacement therapy in severe hemophilia A patients is complicated by formation of inhibitors in around 25% of children. Identification of patients at the highest risk may help to tailor personalized treatment strategies.

Objectives. To develop a scoring system that may be used to identify patients at the highest risk of inhibitor development at first treatment.

Methods. We used the data from a retrospective multicentre cohort study (the Canal cohort) of patients with severe hemophilia A (factor VIII (FVIII)less than 0.01 IU/ml), born 1990–2000, followed at least until their 50th exposure day. Presence of inhibitor was defined as twice a positive inhibitor titer and a decreased FVIII recovery. Based on the coefficients of a logistic regression model, a weighted risk-score was developed. Shrinkage of regression coefficients was used to control for overfitting. The discriminative ability of the risk-score was expressed as the area under the curve (AUC) of a receiver operating characteristic curve.

Results. Out of the 366 patients from the Canal cohort, 284 children were selected of whom 78 developed an inhibitor (27%). Logistic regression analysis revealed 3 independent risk factors for inhibitor development: positive family history, intensive treatment (at least 5 consecutive days) at the first FVIII exposure, and high risk FVIII gene mutations. Table 1 presents odds ratios of the uni- and multivariate analyses, and the risk score. The AUC for the risk-score was 0,724. Table 2 shows that the model is able to separate high and low risk patients from patients with an intermediate risk of 25%.

Conclusions. The risk score includes positive family history of inhibitors, high risk factor VIII gene mutation and intensive treatment at first FVIII exposure. This risk score can recognize patients with a doubled risk for inhibitor development and may be used to guide inhibitor preventive treatment strategies.

Table 1.

Uni- and Multivariate analysis and risk-score.

OR (CI), UnivariateOR (CI), Multivariatep-valueRisk-Score
CI = confidence interval 95% 
Positive Family History of Inhibitors 4.0 (1.7–9.4) 3.0 (1.2–7.7) ,022 3 
High risk FVIII Gene Mutation 3.6 (1.8–7.2) 3.7 (1.8–7.9) ,001 4 
Intensive Treatment at 1st FVIII exposure 6.8 (3.6–12.9) 7.2 (3.4–15.1) ,000 6 
OR (CI), UnivariateOR (CI), Multivariatep-valueRisk-Score
CI = confidence interval 95% 
Positive Family History of Inhibitors 4.0 (1.7–9.4) 3.0 (1.2–7.7) ,022 3 
High risk FVIII Gene Mutation 3.6 (1.8–7.2) 3.7 (1.8–7.9) ,001 4 
Intensive Treatment at 1st FVIII exposure 6.8 (3.6–12.9) 7.2 (3.4–15.1) ,000 6 
View Large
Table 2.

Calibration of the risk-score.

ModelTotal n° patientsPredicted n° InhibitorsObserved n° InhibitorsPositive Predictive ValueNegative Predictive Value
LR= Low Risk, MR= Medium Risk, HR= High risk 
LR:0 72 7 6 0,34 0,92 
MR:3–4 153 39 38 0,25 0,69 
HR: >4 59 32 34 0,58 0,80 
ModelTotal n° patientsPredicted n° InhibitorsObserved n° InhibitorsPositive Predictive ValueNegative Predictive Value
LR= Low Risk, MR= Medium Risk, HR= High risk 
LR:0 72 7 6 0,34 0,92 
MR:3–4 153 39 38 0,25 0,69 
HR: >4 59 32 34 0,58 0,80 
View Large

Topics:
antibodies, child, hemophilia a, factor viii, calibration, family history, logistic regression

Author notes

Disclosure: No relevant conflicts of interest to declare.

2007, The American Society of Hematology
2007
Sign in via your Institution