(T1230-10-57) Biorelevant Dissolution Testing in Tiny-TIMsg to Improve Predictivity of Food Effect on Oral Delivery of BCS Class II Compound

Purpose: A physiologically-based biopharmaceutics (PBB) model was developed for oral administration of a BCS class II compound to a healthy adult volunteer population, using a USP II apparatus to inform fasted state drug dissolution from the formulation. A significant food effect was observed from the PK data which could not be captured using the USP II dissolution methodology or pure in silico modelling. Therefore, the tiny-TIMsg, part of the SurroGUT™ platform previously applied to predict food effect for PBB modelling¹, was implemented as an advanced in vitro dissolution model since it is able to mimic the interplay of mixed micelle and viscous effects of a real food meal, dynamic secretions of acid, bicarbonate and bile salts and biorelevant peristaltic motility and gastric emptying time on drug release and dissolution in the gastrointestinal (GI) tract in vitro.

Methods: A two compartmental PK model was developed for oral administration of a BCS class II compound from an immediate release formulation using GastroPlus® (v10.1, Simulations Plus, Lancaster, CA). A middle-out approach was taken to optimise the compartmental PK model in the fasted state. Compound properties were informed by a combination of in vitro measurements and ADMET Predictor® (where in vitro data were unavailable). For the fasted state, a dissolution profile obtained from the USP II was integrated, wherein a formulation was added to 900 mL sodium phosphate buffer (0.1 M, pH 7.0) with 1 % (w/w) Tween 60 and agitated at 100 rpm. For the fed state, multiple approaches were taken to capture food effect on drug release and dissolution. Firstly, drug dissolution was modelled from an IR tablet using Gastroplus®. Alternatively, in vitro dissolution profiles obtained from A) the USP II and B) the tiny-TIMsg were integrated into the PBB model. The tiny-TIMsg is an in vitro model of the stomach and small intestine that allows oral administration of a dosage form with a real food meal, realistic simulation of dynamic gastric and intestinal secretions, peristaltic motility and physiologically-relevant gastric transit and emptying times. Importantly for BCS II compounds, transit from the gastric to intestinal compartment facilitates precipitation to be captured and in situ digestion of a real food meal and physiological bile salt concentrations reproduces any micelle-mediated solubility enhancement of API that may be expected in vivo. The formulation was administered to the stomach compartment with a high-fat meal, mimicking fed state oral administration in vivo. Passive absorption of simulated small intestinal fluid through a filtration unit facilitated the removal of drug-saturated media from the system, similar to a flow-through dissolution apparatus or passive absorption of intestinal fluids in vivo. Filtrate and lumen samples were collected at predefined time intervals and analysed using ultra-performance liquid chromatography (UPLC) to quantify the amount of drug present. The resulting dissolution profile was imported to Gastroplus®.

Results: A PBB model was built to predict the food effect of a BCS Class II compound. Figure 1 (A) shows a good prediction in the fasted state, informed by USP II dissolution data. Figure 1 (B) shows the predicted and clinical PK data in the fed state. Due to the difficulty of predicting food effect on drug release and dissolution, different techniques were incorporated into the model and compared, namely a purely in silico model, the USP II and tiny-TIMsg. The purely in silico approach performed the least well, significantly underpredicting drug exposure. When informed by the USP II dissolution profile, the predicted PK profile peaked early, failing to capture the extended dissolution time in fed state. This resulted in a significant overprediction of C_max and an early t_max, with predicted to observed ratios of 1.40 and 0.30, respectively. A significant improvement was observed using tiny-TIMsg, with predicted to observed ratios for C_max and t_maxof 0.95 and 1.10, respectively, and a successful prediction of AUC (ratio of 0.95). This is in line with previous successful applications of tiny-TIMsg for PBB modelling^1,2.

Conclusion: A PBB model was successfully built to predict the human PK performance following oral administration of a BCS II compound. USP II adequately described fasted state drug release and dissolution, however failed to capture food effect. In contrast, a fed-state dissolution profile from the tiny-TIMsg significantly improved the PBB model, enabling a more accurate prediction of fed-state PK profile. For compounds with poor aqueous solubility, it can be difficult to predict drug dissolution in conventional in vitro apparatus and purely in silico. The tiny-TIMsg is a valuable in vitro tool to model the complex phenomena governing drug release and dissolution in the GI tract.

References: 1. Jia Liu, Karthik Nagapudi, Po-Chang Chiang (2024). Evaluating Utilization of Tiny-TIM to Assess the Effect of Food on the Absorptions of Oral Drugs and Its Application on Biopharmaceutical Modeling. Journal of Pharmaceutical Sciences, Volume 113, Issue 6, Pages 1586-1596. 10.1016/j.xphs.2024.01.009.
2. Chiang, Po-Chang & Liu, Jia & Nagapudi, Karthik & Wu, Ricky & Dolton, Michael & Salehi, Niloufar & Amidon, Gregory (2022). Evaluating the IVIVC by Combining Tiny-TIM Outputs and Compartmental PK Model to Predict Oral Exposure for Different Formulations of Ibuprofen. Journal of Pharmaceutical Sciences. 111. 10.1016/j.xphs.2022.01.024.

Observed PK profile following oral administration of compound X in the fasted (A) and fed (B) state. The fasted state prediction used USP II dissolution profile whereas multiple methods were employed for fed state, modeling dissolution in silico and in vitro using the USP II (magenta) and tiny TIM (orange).