Drug-Coated Balloon Surface Roughness Modulates Intraprocedural Contact Mechanics and Delivery Efficiency: An In-Silico Study

1. Introduction

The global prevalence of peripheral artery disease (PAD) has increased significantly, with 1,466 diagnoses per 100,000 individuals worldwide in 2019 ¹. This rising trend is also evident in the United States, where 2020 estimates indicate that PAD prevalence ranged from 19 to 21 million individuals ². Following the approval of drug-coated balloons (DCBs) nearly a decade ago, DCBs have seen rapid adoption in the U.S., wherein the primary drug payloads are Paclitaxel (PTX) and derivative compounds ³. An analysis of 62,054 arterial interventions performed from September 2016 to December 2019 showed that DCBs were used in 34.4% (n=21,352) of these procedures ⁴. Recent meta-analyses demonstrate that DCB angioplasty is more effective than plain old balloon (POB) angioplasty for PAD, significantly reducing the risk of revascularization, restenosis, late lumen loss, and major adverse events ⁵. Despite these benefits, some concerns remain regarding the long-term safety of PTX-coated devices, particularly potential systemic toxicity due to off-target delivery ⁶. It is important to note that even reduced doses of PTX may elevate systemic toxicity and the risk of all-cause mortality, particularly in individuals with pre-existing cardiovascular conditions ⁷. The long-term effectiveness and deleterious side-effects of DCB angioplasty is dependent on the PTX dosage administered, underscoring the necessity for continued improvement in drug transfer efficiency during balloon deployment ⁸.

DCB structural contact modeling has been sparsely used in device design, even though the contact between the coating and arterial wall obviously influences drug transfer and thus subsequent therapeutic performance ⁹. Contact between these inherently non-uniform surfaces make the relevant mechanics multiscale and challenges identification of key biophysical interactions governing transfer efficiency during DCB deployment ¹⁰. Finite element approaches may offer a solution, as they have been employed to analyze other endovascular device-artery interactions and identify arterial pressure, arterial compliance, lesion complexity, and stent design as factors underlying the success of vascular therapy ^11,12. In the DCB context, prior research has qualitatively emphasized the importance of coating microstructure and surface topography in determining the nature of contact and drug transfer efficiency during balloon inflation ^9,13,14. Distinct coating microstructures, such as needle-like versus spherical shapes, differentially influence coating penetration into tissue and interfacial contact mechanics, thereby affecting drug delivery ¹⁵. Surface modifications, such as ultraviolet-ozone plasma treatment, can also alter surface roughness to enhance coating adherence and drug transfer efficiency ¹⁶.

Existing literature has explored the relationship between DCB-arterial wall contact parameters, including contact area, contact load, and applied displacement ^17,18. We seek to build upon these studies by understanding the operative contact area between the balloon and the arterial wall due to variations in coating composition and microstructure ^9,12,19. We previously analyzed the surface topography of novel DCB coatings that deliver both PTX and the anti-contractile drug Valsartan (VAL), where we identified a compositional dependence of surface roughness and associated topographical parameters, including the arithmetical mean deviation (R_a) and root mean square deviation (R_q) of a microstructural surface map ²⁰. Here we consider urea-based coating variants that are distinguished by the dosing ratio of VAL to PTX, specifically [VAL/PTX] = 0, 0.25, 0.5, and 1. The experimentally obtained topography of these coatings were used to generate contacting surfaces in a finite element model of DCB deployment. A commercial finite element package was utilized to perform elastic contact modeling, wherein model predictions yield measures of intra-procedural PTX and VAL delivery efficiency as a function of both coating compositional and balloon deployment parameters.

2. Materials and Methods

2.1. Quantification of Coating Surface Area

We previously quantified the surface roughness of urea-based coating formulations via texture analysis of scanning electron microscope (SEM) images (Figure 1A) ²⁰. Here we directly used these images to first extract point clouds of the coating surfaces (Figure 1B), then applied an interpolation function to recover a continuous surface profile (Figure 1C), and finally used the function to create parametric surfaces amenable to finite element modeling (Figure 1D). This 3D geometry provided a quantification of urea-based surface area variation of PTX coatings with titrated VAL content, wherein each coating variant was reconstructed within a standardized geometrical domain (100x100x25 µm³) reflective of typical for urea-based coatings ²¹.

2.2. Contact Model: Domain Creation

A 2D axisymmetric model was developed to simulate intraprocedural DCB-arterial wall contact, with the axial direction along the globally-defined positive Y-axis and the radial direction aligned with the positive X-axis. This model consisted of three geometric domains: a balloon, a coating, and the arterial wall. Prior to imposing a geometrical rotation about the axis of symmetry, all domains were rectangular with balloon, coating, and arterial wall dimensions of 50 µm x 3 mm, ~25 µm x 3mm, and 0.5 mm x 6 mm, respectively. A rotational symmetry of order 4 (270°) was then applied across all domains, creating a 3D cylindrical representation of the DCB-arterial wall contact environment.

2.2.1. Balloon Domain

Balloon material properties were motivated by a standard semi-compliant catheter that expands up to 30% upon inflation ¹². The balloon material was accordingly modeled as a homogeneous, isotropic, linear, elastic solid, with a Young’s modulus of 414 MPa, a Poisson’s ratio ν=0.4, and density ρ=916 kg/m³ based on reported values for Pebax balloons ^12,22–24.

2.2.2. Coating Domain

We considered four coating domains, each with unique, experimentally-determined surface microstructures resulting from the varied drug payloads. Prior to axisymmetric domain rotation, all 2D coating domains had a length of 3 mm and a nonuniform width due to the surface microstructure. We utilized a polygon function to connect and close the distinct interpolation curves extracted from the SEM images, but positioning them at specific coordinates such that the mean coating thickness was ~25 µm in all cases. The coating formulation integrates two therapeutic agents, PTX and VAL, with urea serving as the excipient at a 1000 mg/mL concentration ²⁵. This excipient concentration drives coating crystallization and confers both the structural integrity and drug release characteristics ^26,27. Although the underlying urea crystals inherently exhibit mechanical anisotropy due to the alignment of needle-like surface features ¹⁵, we approximate the coating as an isotropic material with topography derived from SEM roughness profiles. The irregular, multi-directional nature of the observed surface features suggests that this simplification, which greatly reduces computational cost, reasonably reflects the critical surface features dictating contact mechanics ^28-31. The urea-based coating layer was modeled as a homogeneous, isotropic linear elastic solid with a Young’s modulus of 2 GPa ²⁹, a Poisson’s ratio of 0.3 ³⁰, and a density of 700 kg/m^{3 31}.

2.2.3. Arterial Wall Domain

A 2D axisymmetric model of a vessel was created as a rectangle with L = 6 mm and W = 0.5 mm, which, upon revolution, generates an ideal cylinder mimicking the geometry of a healthy femoral arterial wall ³². The vessel length considered in each case was twice the length of the balloon/coating domains, ensuring full DCB containment and preventing any interference from imposed boundary conditions on simulated balloon expansion. The selected thickness represents a physiologically realistic dimension for the healthy femoral artery, particularly relevant for the combined intima-media layer.

The mechanical behavior of the vessel was characterized using the 3rd-order Mooney-Rivlin hyperelastic constitutive model ^33,34. The assigned geometry and model parameters were based on reported values for femoral arteries (Table 1), with the tissue considered as a nearly incompressible solid with Poisson’s ratio of 0.475 ¹², a density of ρ=0.983 g/mL ³⁵ and Young’s modulus Е=9Mpa ³⁶.

2.3. Simulated Deployment

DCB-arterial wall contact due to a prescribed balloon displacement and subsequent drug transport upon coating contact with the arterial wall were simulated with a commercial finite element software (COMSOL^TM) within a time-dependent 2D axisymmetric framework as described above. The contact pairs were defined as the nonuniform coating layer as the source boundary and the arterial wall as the destination boundary (Figure 2A) ³⁷. The source was meshed using a triangular mesh to accommodate the sharp edges of the coating’s irregular surface, while a uniform rectangular mesh was employed for the featureless destination boundary (arterial wall) (Figure 2B). Mesh convergence studies were conducted by iteratively increasing mesh density within each domain, with the stop condition set as a difference of results between two successive steps less than 2%. To achieve accurate convergent results and save computation time, several factors were considered in the simulation. These include: (1) a prescribed displacement value applied on the top surface of the balloon-coating layer to mimic the actual balloon inflation-deflation process; (2) a first-order rectangular function used to apply the load for ramping; and (3) ensuring the mesh on the destination surface is always finer than the source boundary (Figure 2C).

2.4. Transport Equations

PTX and VAL transfer to and transport within the arterial tissue was simulated with a transient diffusion model, thereby neglecting the contribution of advection to drug transport within the wall ^6,35-36. Thus, concentration-dependent drug kinetics are first-order and depend on assignment of an effective diffusion coefficient (D_eff), which is drug- and domain-specific:

${\displaystyle \frac{\partial C}{\partial t}}=D_{eff}{\nabla }^2\mathrm{C}$………………………….1

While VAL shows some capacity to penetrate transdermal tissues and undergo passive diffusion, its overall tissue distribution is limited, with no explicit diffusivity constant found in the given data ⁴⁰. The available data mostly point to its pharmacokinetics involving limited tissue permeation and binding mostly to plasma proteins ^41,42 we integrated theoretical calculations with MonteCarlo random walk simulation ⁴³ to calculate the diffusion coefficient in both environments (urea coating and artery), initially, the Stokes-Einstein equation was used (Equation 2)

$D_{eff}={\displaystyle \frac{{\kappa }_{B}T}{6\pi \eta r}}$……………………………2

This involved defining the key parameters—specifically, the Boltzmann constant (kB), absolute temperature (T), VAL radius (r), and the viscosities (η)—for both the artery and urea coating at 310 K. All constants and parameters are summarized in Table 2. Monte Carlo simulation was performed for each environment to provide an empirical estimation of the diffusion coefficient. A total of 10,000 particles were simulated over 1,000 time steps, with each step lasting 1 millisecond.

During each time step, the displacement of each particle was randomly generated from a normal distribution, effectively mimicking Brownian motion. The mean squared displacement (MSD) of all particles from their initial positions was then calculated, and this MSD, along with the total simulation time, was used to estimate the diffusion coefficient for each environment using the relationship described in Equation 3.

$\mathrm{M}\mathrm{S}\mathrm{D}\left(\mathrm{t}\right)=6D_{eff}\mathrm{t}$ ………………………….3

2.5. Boundary Conditions

Arterial wall contact with the coating was defined as a contact pair within the structural contact model (Figure 2A). A zero-flux boundary condition was applied between the coating-lumen interface and the coating-balloon interface. A perfect sink condition was applied to the end wall layer. All other boundaries remained open throughout the balloon inflation-deflation simulation. The initial drug molar concentration of PTX in the coating was set to 140 mol/m³ for a coating thickness of 25 µm, which is equivalent to a real surface concentration of 3 µg/mm², across all four models. Conversely, the initial volume concentration of VAL varied monotonically to match the experimental [VAL/PTX] surface concentration ratios of 0, 0.25, 0.5, and 1, corresponding to VAL molar concentrations of 0, 68.75, 137.5, and 275 mol/m³, respectively.

2.6. Simulation

To facilitate transient solution convergence, we employed a fully coupled approach in which all degrees of freedom embodied by the governing equation system are simultaneously solved for at each simulation time step. This strategy, while computationally intensive, provided convergence stability when implemented with an automatic, highly nonlinear (Newton) solver. The governing system of equations was integrated using a fifth-order backward differencing scheme with variable time stepping and default values of relative and absolute tolerances. To ensure a smooth and convergent simulation output, we applied ramping functions for surface displacements (as opposed to step-wise functions). The applied balloon displacement to simulate DCB deployment was 50, 60, 70, and 80 µm (Figure 3), values in line with the expected range of post-contact inflations for typical DCBs ^23,44.

3. Results and Discussion

3.1. Surface Area Characterization

Our findings indicate that surface topography, characterized by peaks and valleys structures, is non-uniformly distributed across all patch regimes (Figure 1). The presence of these microscopic irregularities demonstrates that the true surface area within a regime always exceeds its idealized geometric surface area (Figure 4), which is considered along with standard measures of coating consistency over the balloon surface.

3.2. Contact Mechanics

Our results demonstrate that surfaces with high inherent roughness, such as 0[VAL/PTX], maintain a substantial contact area even at low loads due to their numerous peaks providing an extensive topological surface (Figure 5A-D). At low loads, a relatively smooth surface, such as 0.25,1 [VAL/PTX], initially presents a larger nominal contact area compared to a rougher surface like 0.5 [VAL/PTX] (Figure 5A, B) because the smooth surface forms a single, continuous contact patch, and its area growth saturate, leading to an insignificant increase in the contact area. For all coating variants, the contact area demonstrates a nonlinear increase with applied low loads (Figure 5E). However, a clear linear relationship emerges in the contact area, primarily due to the coating’s stiff nature. This stiffness causes the peaks to largely retain their shape, effectively acting as tiny indenters that deform the softer artery. The artery’s elastic properties enable it to intimately conform to the rough coating’s topography as the load increases. This conformity effectively engages more microstructural peaks, thereby increasing the effective contact area and confirming that surface roughness further enhances tissue-coating contact area (Figure 6).

The average contact stress, expressed by Von Mises stress in Pascals (Pa), is plotted against different applied displacements (Figure 7). While initial contact on a rough surface concentrates stress at peaks, the high deformability of the artery allows these peaks to distribute the load over a larger area. This mechanism reduces extreme local pressure points that would otherwise occur with hard, rough surfaces.

3.3. Impact of the Coating Roughness on the Amount of Drug Transfer

Increased surface roughness on DCBs results in a higher amount of PTX transfer to the artery (Figure 8). This is attributed to the fact that, as we stated before, a rougher surface offers a larger total contact area for drug release. This phenomenon of a rougher surface promotes superior mechanical interlocking with the artery, particularly as load increases. The irregularities promote faster transport of the drug away from the surface by offering more escape routes or contact points at the interface, thereby accelerating the achievement of steady-state drug release (Figure 8A, D). In contrast, the smooth surface provides a less efficient diffusion source, hindering drug transfer and slowing the achievement of steady-state release. PTX has a significantly higher molecular weight, approximately 853.9 g/mol, which is nearly double that of VAL, at approximately 435.52 g/mol. This substantial difference in molecular weight is a primary factor explaining the distinct behaviors observed between the two drugs ( and Figure 9). PTX’s larger size leads to greater resistance and slower movement within both the coating and the arterial tissue, resulting in lower diffusion coefficients D_eff (5.41e^-11, 1.99e^-13) m²/s in the healthy artery and coating, respectively³³. In contrast, VAL, being a relatively low molecular weight drug, shows higher diffusivity. Our Monte Carlo simulation estimated its diffusivity at 1.3×10⁻¹⁰ m²/s in the healthy artery and 4.54×10⁻¹¹ m²/s in the coating. This enhanced diffusivity allows VAL to easily navigate the internal matrix of the coating. The greater mobility of VAL subsequently overshadow any potential influence of surface topography on its release characteristics (Figure 9).

3.3. Study Limitations

Our in-silico study focuses on intraprocedural contact mechanics and drug transfer, which, while vital, does not extend to predicting the chronic effects of different roughness profiles on drug retention or systemic distribution. The primary simplification involves the reduction of complex anatomical and physiological factors to a computational model, which omits the full spectrum of in vivo variability. Another study is needed to precisely adjust the microstructural elements of the drug coating based on its composition to achieve an optimal balance between micro-scale surface roughness and drug transfer efficiency. Such considerations motivate further exploration of novel drug payloads for integration into next-generation medical devices.

4. Conclusions

Our investigation demonstrates that the interplay of surface roughness, applied load, and contact area is crucial for DCB performance. By strategically incorporating an anticontractile compound like VAL at a half ratio into the existing urea-based DCB coating, we achieved a dual-purpose design. This modification, despite slightly altering roughness, allows for susbsequent evaluation of VAL within DCB payloads while effectively preservingthe inherent microstructure and contact mechanics of traditional urea-based coatings, thereby maintaining the primary function of PTX delivery/efficiency.