Genetic Influence of Genetic Factors Influencing the Desmopressin's Efficacy in Mild/Moderate Hemophilia A

GIDEMHA is an observational, retrospective and multicentric clinical-biological study conducted in Hemophilia Treatment Centers (HTC) of the French Grand-Ouest interregion including HTCs of Angers, Brest, Caen, Nantes, Rennes and Tours. Objectives of the GIDEMHA study are: * Description of the post-desmopressin (DDAVP) FVIII pharmacokinetics (PK) in a large retrospective cohort of patients with mild/moderate HA, * Research of patients-related factors influencing this FVIII PK, * Building of predictive population- and Bayesian-based models Inclusion criteria: * Males at any ages with a mild or moderate hemophilia * Therapeutic test with DDAVP realized since 2010, * Factor VIII levels measurements realized at least 2 times during the therapeutic test, just before the DDAVP infusion and 30 or 60 minutes after, * Complete genotyping of the F8 gene for genetic diagnosis of hemophilia Exclusion criteria: * Patients with an anti-factor VIII inhibitor * Refusal to participate in the study * Unable to understand the study's French letter of non-opposition and information Description of the DDAVP therapeutic tests: The procedure of the DDAVP therapeutic test was identical for all investigator centers as recommended by international and French guidelines. DDAVP was always administered intravenously at a dose of 0.3-0.4 μg.kg-1 diluted in 50 mL of saline solution over 30 minutes. Hemostatic parameters were required to have been evaluated before and at least 30 or 60 minutes after the DDAVP infusion. Subsequent measurements performed at T2h, T4h and T6h after the infusion are also recorded during the test. Collected data: All data collected in this study were issued from the medical files at the moment of the DDAVP therapeutic test. They include: * FVIII activity levels measured with a one-stage clotting assay from plasmas collected in 0.109 M sodium citrate (fresh or stored at -80°C). These FVIII levels were measured just before and after the DDAVP infusion (until 24 hours if available). * Molecular analysis of the F8 gene. * Blood group * DDAVP doses * Von Willebrand factor levels during the DDAVP test * Age * Weight * Polymorphisms of genes influencing the FVIII clearance if available * Blood DDAVP levels after DDAVP infusions (only for patients of the HTC of Rennes and included during the last 5 years) Pre/Post-DDAVP pharmacodynamic parameters The following pharmacokinetic parameters were calculated using FVIII activity versus time: basal FVIII, FVIII peak (highest level measured after DDAVP administration), FVIII recovery (recFVIII= peak FVIII / basal FVIII), FVIII half-life (FVIII T1/2) and clearance, and FVIII area under the curve (FVIII AUC). All parameter except T1/2 were estimated using non compartmental method. FVIII AUC was calculated using the trapezoidal from FVIII activity versus dosing time extrapolated to baseline, based on the last observed concentration. The elimination rate constant (Ke) was calculated using one-compartment model approach with the following equation: C=C0\*e(-Ke.t) where C, C0, Ke and t denote respectively, the post-DDAVP FVIII activity, peak FVIII activity, the elimination rate constant, and time after DDAVP administration. Goodness of fit statistic for the terminal elimination phase was adjusted for the number of points used in the estimation of Ke, and only those R2\>0.90 were conserved for further analyses. FVIII T1/2 was calculated as Ln(2)/Ke. T1/2 and AUC were calculated only if FVIII levels were measured at least at 4 different times, with basal and peak FVIII levels being available. To validate the method, the investigators carried out 2 additional FVIII measurements (at 12h and 24h) in the last 10 patients enrolled. This allowed to compare the FVIII T1/2 obtained from 5 points (from T0 to T6h) with that obtained from 7 points (from T0 to T24h). Scores to measure the response to DDAVP To qualitatively assess the biological response to DDAVP, the investigators used criteria previously reported by Stoof et al \[11\]: * The absolute response was either "complete" (peak FVIII ≥0.5 IU.mL-1), "partial" (FVIII ≥0.3 - \<0.5 IU.mL-1) or "null" (FVIII \<0.3 IU.mL-1). * The relative response was defined as "complete" (recFVIII \>3), "partial" (recFVIII ≥2 - \<3) or "null" (recFVIII \<2). Two other scores, absolute duration and relative duration, were built to determine the evolution of the DDAVP response over time. They also comprised each 3 groups (short, medium and long) based on the data collected from all patients included, with at least 20% of patients per group: * The Absolute duration determined the time that the FVIII level was ≥ 0.5 IU.mL-1 after the FVIII peak. It was either "short" (≤3h), "medium" (\>3h - ≤6h), or "long" (\>6h). * The Relative duration corresponded to the FVIII T1/2 and was either "short" (≤3h), "medium" (\>3h - \<5h) or "long" (≥5h). These 4 scores were determined for all patients, but statistical analyses by mutation were performed only for the so-called "hot spot" F8 mutations, defined as ≥5 patients having a similar gene defect. Statistical analyses Descriptive characteristics were analyzed with median values, their 25-75% interquartile ranges (IQR) and minimum-maximum values (MIN-MAX). Non-parametric Kruskal-Wallis and Mann-Whitney tests were used to compare continuous variables between the groups. Fisher's exact test was performed to compare proportions in contingency tables and the Odds ratio was calculated. A univariate linear logistic regression was used for the paired comparison of continuous values. Survival curves with FVIII ≥0.5 IU.mL-1 of different F8 hot spot mutations were compared by the Kaplan-Meier method. For qualitative values of AR and AD of the 4 mutation groups, positive predictive value (PPV), negative predictive value (NPV), sensibility and specificity were calculated. An approximate 95% confidence interval was determined (95% CI) for every statistical analysis and a p-value \<0.05 was considered statistically significant. SPSS 17.0 (SPSS Inc. Chicago, IL, USA) and GraphPad 5.0 (Prism Software Inc. San Diego CA) were used to perform the statistical analyses.