PWI-001 - COMPARISON OF METHODS OF FIRST-IN-HUMAN DOSE PREDICTION FOR ADENO-ASSOCIATED VIRUS-MEDIATED GENE THERAPIES IN FABRY DISEASE

F. Gibbons¹, N. Boukharov², N. Chen³, K. Sun⁴; ¹Preclinical and Translational Sciences, Takeda Pharmaceuticals, Cambridge, MA, United States, ²Takeda Pharmaceuticals, Cambridge, MA, ³Preclinical and Translational Sciences, Takeda Pharmaceuticals, ⁴Data Science Institute, Takeda Pharmaceuticals.

Background: Adeno-associated virus gene therapies (AAV-GT) enable durable non-integrating delivery of genetic material, producing transgene products that can complement endogenous enzymes whose abundance or activity may be pathogenic. Several challenges raise the stakes of first-in-human (FIH) dose selection: (i) target tissues in humans are not transduced as efficiently as mouse (ii) anti-capsid antibodies raised following the first dose obviate a second one and (iii) regulatory guidance advocates starting with a potentially therapeutic dose.

Methods: We examined three methods for FIH dose prediction of AAV-GT, developed using data for hemophilia A and B. Two apply allometry to a gene-efficiency factor (GEF) (“Tang GEF”, PMID: 33884615; “Zou GEF”, PMID: 36374990), while a third establishes a non-linear relationship between dose and a species-independent “normalized protein” (“Zou NP”, PMID: 36367416). We digitized non-clinical and clinical data from literature for several Fabry disease AAV-GTs. Since there exists debate over the appropriate target concentration, we assessed predictive performance by geometric mean ratio of predicted-to-actual dose (GMRD) at measured clinical serum concentrations.

Results: We find that scaling parameters can be well estimated for all three methods and differ appreciably from those found for hemophilia A and B. For a given measured clinical serum concentration, the Zou GEF allometric method predicted a dose closest to the actual dose used (i.e., GMRD being close to 1, see Table 1). Systematic biases in the other two methods are much larger than Zou GEF.

Conclusion: To our knowledge, this is the first application of these methods to Fabry disease. In this limited dataset, using the clinical concentrations observed, the allometric method of Zou GEF came closest to predicting the required dose.