Pharmacokinetic model of plasma FVIII exposure after AAV-FVIII delivery defines an immunogenicity threshold. (A) These graphs display the FVIII production (or exposure) rate, k_FVIII, and the corresponding cumulative FVIII exposure, AUC_FVIII, for each AAV-FVIII–treated animal through day 15 (n = 66). Red curves represent animals that developed an inhibitor response, and black curves represent the animals that did not develop inhibitors. The proposed immunogenicity threshold is outlined in blue and was determined by linear regression with shared parameters. (B) ID₅₀ values were calculated by simple logistic regression analysis (ID₅₀ = the value predicted to induce inhibitor development in 50% of AAV-FVIII–treated subjects): (i) day 5 k_FVIII, (ii) day 10 k_FVIII, (iii) day 5 AUC_FVIII, (iv) day 10 AUC_FVIII (likelihood ratio test [LRT] of the β1 variable was set at P < .05 for significance and brackets indicate the 95% CI). (C) Contingency analysis was performed based on the inhibitor incidence (count) of animals that met the k_FVIII threshold by day 10 (2-sided Fisher’s exact test with significance set at P < .05; brackets indicate 95% CI; ARI, attributable risk increase; RR, risk ratio). (D) Survival curve indicates occurrence of inhibitors for animals with day 5 k_FVIII values ≥ ID₅₀ compared with those below the ID₅₀. The median time to inhibitors with k_FVIII ≥ 4.49 = 30 days. (χ² = 66.95, df = 1, P < .0001). (E) Correlation of day 5 k_FVIII values with the time to anti-FVIII IgG incidence. (Spearman r = −0.7333; 95% CI, −0.8907, −0.4189; P = .0002). (F) Median day 5 k_FVIII values by vector dose for each AAV-FVIII vector. Error bars indicate IQR. Values below the red dashed line indicate a day 5 k_FVIII = 0, meaning that there was no detectable FVIII activity by chromogenic assay 5 days after vector delivery. Both the vector design and dose factors have a significant impact on the day 5 k_FVIII (main effects 2-way ANOVA, P < .0001).