(T0930-10-67) A Sensitive and Specific Human Primary Stem Cell-Based In Vitro Assay for Predicting Gastrointestinal Toxicity Risk

Purpose: Gastrointestinal toxicities (GITs)are the most frequently reported drug-induced adverse events (AEs) in clinical trials. The incidence rate of GIT reflects (1) high intestinal drug exposure, particularly for orally administered therapeutics; and (2) an inherent sensitivity of the rapidly dividing stem/progenitor cell populations that maintain intestinal self-renewal, differentiation, and barrier formation/maintenance. Disruption to these populations can lead to impaired GI barrier integrity, giving rise to a plethora of clinical phenotypes including diarrhea, inflammation, and increased risk of infection from opportunistic pathogens. As common nonclinical animal models exhibit poor translatability to human clinical outcomes, GITs are often not discovered until late-stage clinical development. Furthermore, tumor-derived cell models, such as Caco-2 cells, exhibit limited predictive power as a transformed epithelium with abnormal growth and differentiation characteristics. As such, there is an unmet need for human-based cell models capable of informing GI safety risks in early drug discovery and lead optimization stages. Ideally, such a model would closely reprise natural human GI physiology and cellular composition, and enable reproducible, robust, and reliable assessments of drug effects on intestinal renewal and tissue function. To this end, Altis has developed a cell-based assay by repurposing our current model, RepliGut® Planar, to produce a 96-well plate GI Toxicity screening assay that combines high-throughput readouts for both cell proliferation and barrier formation. This cell model has the potential to fill the unmet need for an accurate drug screening human stem cell-based model.

Methods: Human intestinal tissue was obtained post-mortem via an established Organ Procurement Organization following consent of family under strict ethical guidelines established by the Organ Procurement Transplantation Network (OPTN). The stem-cells derived from the tissue, were then plated on a 96 well-plate Transwell® filters and tested over a five-day protocol in which proliferative cells were exposed to test articles on day 3 in culture, which included both marketed drugs with (e.g., Bortezomib, Afatinib, Idarubicin and Colchicine) and without (e.g., Verapamil and Nadolol) clinical evidence of clinical GITs. All compounds were dosed over 6-orders of magnitude, overlapping the highest observed plasma concentrations in clinical trials (C_max). On study day 5 (48 hours post-dosing), cell health was assessed as the proliferative capacity (measured using EdU/DAPI staining/imaging) and intestinal barrier properties (measured as the trans-epithelial electrical resistance or TEER). For both endpoints, the toxic concentrations which produced a 50% decrease relative to vehicle (TC₅₀) were calculated and benchmarked against the clinical C_max to assess human translatability. The assays were repeated using cells derived from multiple donor tissues to determine donor-to-donor variability.

Results: Overall, the positive GIT-associated drugs achieved TEER TC₅₀ values within < 20-fold of the clinical C_max, while the known non-toxic compounds did not impact TEER. Bortezomib, which reversibly binds to the 26S proteasome and blocks the degradation of pro-apoptotic factors, produced the anticipated non-selective loss of both proliferative and non-proliferative cell populations. Conversely, Colchicine Afatinib, and Idarubicin, which more selectively target cell proliferation pathways, were ~30 times more toxic to the proliferative versus the non-proliferative cell populations. The negative control compounds did not impact either the proliferative or non-proliferative cell populations significantly.

Conclusion: The study outcomes were consistent with a high degree of predictive accuracy within this assay, both with respect to (1) correctly identifying drugs associated with clinical diarrhea incidence versus negative controls; and (2) establishing TC₅₀/C_max values at < 20-fold. Furthermore, the assays were capable of distinguishing cell population selective vs non-selective mechanisms underlying GITs. Overall, the model system provided an efficient and informative screen for GI toxicity that integrated sensitive and robust endpoints, 96-well plate testing capabilities, and provided insights into the mechanisms underlying the observed toxicity.