Interspecies Scaling of Transgene Products for Viral Vector Gene Therapies: Method Assessment Using Data from Eleven Viral Vectors.

ABSTRACT

The prediction of transgene product expression in human is important to guide first-in-human (FIH) dose selection for viral vector-based gene replacement therapies. Recently, allometric scaling from preclinical data and interspecies normalization of dose-response (D-R) relationship have been used to predict human transgene product expression of adeno-associated virus (AAV) vectors. In this study, we assessed two interspecies allometric scaling methods and two dose-response methods in predicting human transgene product expression of nine intravenously administered AAV vectors, one intramuscularly administered AAV vector, and one intravesical administered adenoviral vector. Among the four methods, normalized D-R method generated the highest prediction accuracy, with geometric mean fold error (GMFE) of 2.9 folds and 75% predictions within fivefold deviations of observed human transgene product levels. The vg/kg-based D-R method worked well for locally delivered vectors but substantially overpredicted human transgene product levels of some hemophilia A and B vectors. For both intravenously and locally administered vectors, the prediction accuracy of allometric scaling using body weight^-0.25 (AS by W^-0.25) was superior to allometric scaling using log(body weight) (AS by logW). This study successfully extended the use of allometric scaling and interspecies D-R normalization methods for human transgene product prediction from intravenous viral vectors to locally delivered viral vectors.