(T1530-02-07) Integrated HypoSkin® Model and PBPK Modeling as a Bioavailability Predicting Tool for Subcutaneous Delivery of Protein Therapeutics

Purpose: Subcutaneous (SC) delivery has emerged as a preferred route of administration for many biologics. Existing literature data indicate a wide range human SC bioavailability (BA) for monoclonal antibodies (mAbs) while mechanisms behind poor SC bioavailability are either unknown or poorly understood. Lack of Reliable models presents a big hurdle for predicting the BA of mAbs following SC administration in humans. In this work, we assessed the feasibility of using an human ex vivo HypoSkin® model for predicting BA in humans.  A mini-PBPK Model incorporating ex vivo data was used for SC bioavailability prediction.

Methods: HypoSkin® human ex-vivo model from Genoskin was cultured in culture medium for 7 days with medium changed every day. A total 5 clinical investigated mAbs with known BA were radiolabeled with [¹²⁵I]. The [¹²⁵I] labeled mAbs were characterized by SEC-HPLC coupled with a radio detector. Radioactivity was quantified using liquid scintillation counting (LSC) and UV-Vis spectrophotometer. 100uL of purified antibody was injected directly into adipose tissue of HypoSkin® model to mimic SC injection. Serial samples were over a period of 7 days and collected culture medium were first measured using LSC to quantify % injection dose (%ID) in medium and in tissue. Intact mAbs in the culture medium were then separated by TCA precipitation and counted using LSC to determine the intact/catabolism ratio of each mAbs. In the next step, we developed a mini-PBPK model that includes compartments for blood, tissues, lymph drainage, and a compartment for subcutaneous injection that links to the lymph compartment. The intact/catabolized ratio was utilized to inform the fraction (F%) of mAb that undergoes lymphatic absorption. Meanwhile, the gradient of the remaining percentage of the % ID was employed to estimate the lymph absorption rate. The mini-PBPK model was developed using Berkeley Madonna version 10.4.2.

Results: The ¹²⁵I labeling of antibodies were achieved with high radioactivity and purity examined by SEC-HPLC. With 100uL mAbs injected into 0.5cm depth underneath the epidermis, no significant leaking was observed, while the skin morphology remains unaltered through 7 days. Analyzing the % ID in the collected medium showed a significantly different tissue penetration rate among tested mAbs. In general, mAbs with higher human SC bioavailability possessed a faster tissue penetration rate ex-vivo. Evaluating the intact/catabolized ratio after day 7 further demonstrated the difference among tested mAbs. In our experiment, mAbs with higher human SC bioavailability typically showed a higher %ID accumulation in the collection medium with larger intact/catabolized ratio, whereas the mAbs with lower SC bioavailability largely trapped in the skin tissue with a larger proportion as catabolized antibodies. The rank order of tissue penetration rate and accumulated intact/catabolized ratio in medium was consistent with the literature data. We carried out simulation of PK profiles for two mAbs, anti-LTα and Omalizumab. The model's predictions were in good alignment with the published clinical PK data, evidencing the model's potential for applying Genoskin and PBPK for predicting SC bioavailability.

Conclusion: Investigating SC injection using an ex-vivo human skin model could provide a reliable quantitative approach to predict SC absorption of mAbs. Combining this ex-vivo data with PBPK modeling can provide a reliable approach to predict human SC bioavailability of antibodies post SC delivery.

Figure 1. % ID accumulated in the tissue post S.C. injection

Figure 2. % ID of intact and catabolized antibody in the culture medium post S.C. injection

Figure 3. Human PK simulations of anti-LTα and Omalizumab