Abstract
Introduction: It was well-known that severe-type patients with hemophilia A (PwHA) had great variability in bleeding phenotypes. Factors effecting bleeding patterns of PwHA include at least treatment modality and interindividual various procoagulant and anticoagulant levels. We aimed to investigate what clinical variables could predict bleeding frequency of severe PwHA and to develop models for predicting bleeding phenotypes among severe PwHA with/without FVIII prophylaxis therapy.
Methods and materials: Totally 51 severe-type previously-treated PwHA from two Hemophilia Centers in Taiwan were enrolled, who received standard half life (SHL) rFVIII products with complete consecutive bleeding records at least more than 6 months, and their medical charts 2017－2018 were retrospectively viewed. The clinical data were collected for analysis, including age, body mass index (BMI), body weight (BW), ABO blood groups, hemoglobin (Hb), hematocrit (Hct), HCV infecton, HIV infection, treatment modality, baseline VWF levels, and genetic defects. Baseline VWF activity meant the data via VWF:ACL activity or VWF:RCo. Clinical variables for annualized bleeding rate (ABR) and annualized joint bleeding rate (AJBR) were evaluated by multivariate linear regression (MVLR) analysis.
Results: The cohort of 51 severe-type PwHA included 8 boys and 43 adults, aged 8-64. For treatment modality, there were 19 patients receiving episodic treatment (ET) and 32 receiving prophylaxis therapy (PT) with intermediate-dose standard half life (SHL) rFVIII. The mean study period was 11.9 months, range 10－14.5 months. Among them, there were 31 with HCV infection and 4 with HIV infection. PwHA with non-O blood group were 31 and those with O blood group 20. The mean baseline VWF:Ag was 115.6±55.5%, range 50%－294.7%. The mean baseline VWF:activity was 105.4±52.1%, range 41.3%－307%. ABR of ET group and PT group were 46.1±29.2 and 6.8±7.1, respectively. (p<0.0001***) AJBR of ET group and PT group were 37.3±27.7 and 6.0±6.8, respectively. (p=0.0001***) By MVLR analysis, both treatment modality and baseline VWF:Ag were recognized as inverse predictors of ABR and AJBR, and HCV infection recognized a predictor for AJBR. Age, inhibitor histroy, BMI, BW, ABO blood groups, Hct, Hb, HIV infection, and missense mutation or not were eliminated as predictors. The predictive equations by MVLR were as the following two:
(1) Predictive ABR = 56.5 - 37.8 * (Treatment model) - 11.8 * baseline VWF:Ag (IU/mL).
(2) Predictive e AJBR = 41.9 - 28.6 * (Treatment model) - 12.0 * baseline VWF:Ag (IU/mL) + 10.0 * (HCV infection).
(1 if Treatment model is PT, 0 if Treatment model is ET. 1 if HCV infection or anti-HCV antibody is positive, 0 if HCV infection or HCV antibody is negative.) Separate prediction models developed from MVLR analysis could explain 52.51% of the ABR variability and 50.56% of the AJBR variability. The correlation between predicted and observed bleeding frequency was significantly strong.(P-rank>0.7, p-value<0.0001***) Mean difference between predicted ABRs and observed ABRs was 1.75 and that between predicted AJBRs and observed AJBRs was 1.27. Predicted ABR deviated <21 (<2 per month) of observed ABR in 42/50 patients (84%). Predicted AJBR deviated < 24 (<2 per month) ofobserved AJBR in 44/50 patients (88%).
Conclusion: Prophylaxis therapy and baseline VWF:Ag levels were the strongest two inverse predictors for ABR and AJBR. Positive HCV infection was another predictor for AJBR. The prediction models provided with an insight into personal bleeding quantified patterns and may identify PwHA with high bleeding risks based on individual characteristics of treatment modality, baseline VWF:Ag, and HCV infection. Our approach is of help for individualized treatment and refining of dosing strategies.
Disclosures
No relevant conflicts of interest to declare.
Severe Type of COVID-19: Pathogenesis, Warning Indicators and Treatment
