A burn injury model using an ex vivo biologically active human skin platform: examples of limitations of novel alternative methods (NAMs) and compensatory insights from computational modeling
Authors:
Gary An, Chase Cockrell
Body of Abstract:
Background: Novel alternative methods (NAMs) are microphysiological systems (MPS) that use human-derived cells intended to reduce and potentially replace animal models. It is still an open question as to whether MPSs can reproduce the organismal/system-level complexity seen in clinical disease. In silico methods have been proposed as a way of integrating multiple MPSs; we have previously described a modular architecture using agent-based models to provide this capability. We apply this concept to modeling the burn patient, and herein propose to use an ex vivo biologically active human skin model sourced from GenoskinTM to generate experimental data to calibrate a previously developed modular component, the Burn Agent-based Model (BABM), of our Burn Digital Twin.
Methods: Burn wounds were induced on Genoskin TM Hyposkin modules (2 cm diameter) using a 25 W contact wood burning tool with a 1 cm diameter flat tip preheated to between 200 °C and 300 °C. The four experimental groups (n = 3/group) were: control, 5 s contact, 15 s contact and 30 s contact. Hyposkin modules were cultivated in an incubator (37 °C, 5 % CO2, 95 % humidity) for 7 days with daily renewal of support medium with exchanged media sent for Luminex analysis of the following mediators: TNF, IL1a/b, IL1RA, IL4, IL6, IL8, IL10, IL12, IFNg and GCSF. Tissue at the end of the experiment (Day 7) was sent for tissue Luminex and histology. Experimental data was used to calibrate our previously developed Burn Agent Based Model using our Active Learning Pipeline/Model Rule Matrix
Results: Histological analysis demonstrated reproducible burn injuries with 5 s burns = superficial partial thickness (SPT) burns, 15 s burns = deep partial thickness (DPT) burns and 30 s burns = full thickness (FT) burns. In terms of trajectories of measured cytokines in the support media, TNF levels peaked at 24 hrs and subsequently decreased, IL1a and IL12 plateaued at 24 hrs and were sustained, IL1b peaked at 48-72 hrs and subsequently decreased, IL1a, IL4, IL6, IL10 and IFNg plateaued at 48 hrs and were sustained, IL1RA rose until Day 5 then decreased, and IL8 and GCSF rose throughout the experiment. There were no statistical differences in the media mediator levels based on duration/depth of burn, though tissue levels of IL1a, IL1b, IL10, IL12 and GCSF were highest in the 5 s/SPT burns. Calibration of the BABM to this data required exclusion of systemic perfusion/circulating inflammatory cells to render it insensitive to burn depth.
Conclusion: We produced SPT, DPT and FT contact burns in an ex vivo human skin model, but had the counter intuitive result that increasing burn depth and amount of tissue damage did not manifest in increased mediator levels. Calibration of the BABM required exclusion of circulating inflammatory cells, suggesting an avenue by which computational experiments can provide direction to overcoming the limitations of a NAM in representing systemic disease.
