Bleeding Data across Baseline FIX Expression Levels in People with Hemophilia B: An Analysis Using the 'Factor Expression Study'

Introduction

Complications such as spontaneous and trauma-related bleeding events typically experienced by people with hemophilia B (PWHB) are associated with long-term joint damage and chronic pain, and burdensome treatment with intravenous factor IX administration.

Gene therapy, designed to enable the endogenous production of the missing clotting factor, has potential for curative benefit in PWHB (Dolan et al, 2018). Due to its link to risk for bleeding episodes, factor expression level (FEL) is commonly used as an endpoint in hemophilia gene therapy trials. However, little data currently exist linking FEL to bleeding risk in PWHB, most notably within the mild range. As such, the aim of this analysis was to examine the relationship between annual bleed rate (ABR) data across baseline FEL in PWHB.

Methods

Data from adult non-inhibitor PWHB, across Europe and the United States (US) who received clotting factor on-demand (OD), were drawn from the 'Cost of HaEmophilia in adults: a Socioeconomic Survey' (CHESS) studies. The CHESS studies are retrospective, burden-of-illness studies in people with hemophilia A or B, capturing the economic and humanistic burden associated with living with hemophilia. Additional data were collected to supplement the existing CHESS studies, particularly in people with exogenous FEL in the mild and moderate range. ABR was defined as the physician-reported number of bleed events experienced by the patient in the 12 months to study capture. A generalized linear model (GLM) was used to analyze variation in ABR data across FEL, adjusting for covariates age, body mass index (BMI), and blood-borne viruses. Following this, a multivariable restricted cubic spline (RCS) GLM regression was performed to create, model, and test for the potential non-linear relationship between FEL and ABR. The RCS regression employed 3 knots, located at baseline FEL values of 1, 5, and 10, and controlled once again for age, BMI, and blood-borne viruses.

Results

A total of 407 adult non-inhibitor PWHB, receiving an OD therapy regimen and with information on ABR, were profiled. The GLM provided adequate fit for the modeling of bleed data; the average marginal effect at the mean was computed from the GLM regression outputs. After controlling for the effects of all other model covariates, the regression analysis showed a significant association between FEL and ABR; for every 1% increase in FEL, the average ABR decreased by 0.08 units (p<0.001). The results of the RCS regression found a significant non-linear relationship between FEL and ABR, ceteris paribus (p<0.001).

Conclusions

The results of this analysis found baseline FEL to be significantly associated with ABR in PWHB; as baseline FEL increased, ABR reduced. This highlights the clinical importance of new hemophilia gene therapies potentially increasing FEL to that of the mild or non-hemophilic range in terms of reducing patient burden through the better prevention of bleeding events in PWHB.