Computational calibration of a novel valve-on-chip device replicating native aortic valve hemodynamics

Introduction: : Calcific aortic valve disease (CAVD) remains the most common aortic valve (AV) disease. It is characterized by progressive fibro-calcific remodeling and thickening of the valve leaflets, which may result in the obstruction of the cardiac outflow. While CAVD has been recognized as the result of cell-mediated processes including inflammation, remodeling, osteogenesis, apoptosis and necrosis, the triggers remain largely unknown. Clinical observations and animal studies have suggested correlations between mechanical stresses and AV biology. Separate in vitro studies conducted in our lab have indicated that fluid wall shear stress (fWSS) and tensile strain (tS) alterations separately promote the same inflammatory and remodeling processes as those observed in early CAVD pathogenesis. However, to date, in vitro models have failed to capture the full complexity of AV hemodynamics, which combines dynamic tS and side-specific fWSS. To address this gap, we designed a novel valve-on-chip (VoC) device. The device consists of a multi-layered collagen-based gel embedded in a rectangular polydimethylsiloxane (PDMS) chamber, mimicking the tri-layered structure and cellular composition of a native leaflet. Silicone rubber sheets attached to the top and bottom of the gel chamber are coupled to a linear actuator to subject the gel to tS. PDMS flow chambers mounted on the silicone sheets and fed by a programmable syringe pump subject the top and bottom gel surfaces to side-specific fWSS (Figure1A). The objective of this study was to determine computationally the operating conditions (i.e., linear actuator displacement and pump mass flow rates) replicating the native AV leaflet fWSS and tS environments.

Materials and

Methods: : A finite element analysis (FEA) was performed in ANSYS Mechanical to determine the linear actuator displacement subjecting the gel to the native leaflet circumferential stretch. The different materials were modeled as hyperelastic Mooney-Rivlin or Yeoh models calibrated against published test data. The initial displacement waveform was estimated from Hooke’s law. At the end of the initial FEA run, a correction factor (native/predicted tS) was calculated and applied to the initial displacement to yield a corrected displacement. This procedure was repeated until the root-mean-square (RMS) error between the predicted and native tS became smaller than 5% of the native tS amplitude. The determination of the pump operating conditions yielding the native leaflet side-specific fWSS was performed via fluid-structure interaction (FSI) modeling in ANSYS CFX and Mechanical. The flow equations were solved in their arbitrary Lagrangian Eulerian form. The final displacement determined by FEA was prescribed on the silicone sheets. The working fluid (DMEM) was modeled as Newtonian. The flow channel outlets were modeled as openings with zero-gauge pressure. Initial fluid velocities prescribed at both inlets to generate the ventricularis and fibrosa fWSS were obtained from the analytical solution of flow in a rectangular channel. Following the initial FSI run, the fWSS predicted on both gel surfaces were extracted and a correction factor (native/predicted fWSS) was used to generate corrected velocities. The process was reiterated until the error in temporal shear magnitude (TSM) and oscillatory shear index (OSI) between the predicted and native fWSS signals were smaller than 5%.

Results, Conclusions, and Discussions::

Results: The linear actuator displacement waveform obtained after four FEA iterations generated a volume-averaged strain in the gel with a normalized RMS error of 0.8% relative to the native tS waveform (Figure 1B). The flow velocities obtained after one FSI run generated TSM errors of 27% and 1%, and normalized OSI errors of 4.8% and 2.8% relative to the native ventricularis and fibrosa fWSS, respectively (Figure 1C).

Conclusions and Discussions: Overall, the results suggest the ability of this novel VoC device to subject a gel-based valvular cell construct to the native tS and FWSS environments of AV leaflets. The linear actuator displacement obtained by FEA closely matched the native tS experienced by the base of AV leaflets. The initial inlet velocities obtained from rectangular channel flow theory were able to achieve fWSS environments that closely matched the magnitude of the native fibrosa fWSS and the directional characteristics of both the native fibrosa and ventricularis fWSS. Additional FSI iterations are needed to refine further the inlet velocity of the ventricularis flow channel. Similar analyses are being conducted to isolate the operating conditions achieving the native hemodynamic stress environments of bicuspid aortic valve (BAV) leaflets. Future steps include the practical implementation of this VoC with trilayered cell constructs mimicking the native leaflet structure and composition, and the assessment of cellular response to the native AV and BAV hemodynamics.

Acknowledgements and/or References (Optional):: This work was supported by the American Heart Association (grant 25AIREA1376304)