Computational Modeling of Probiotic Recovery from 3D-Bioprinted Scaffolds for Localized Vaginal Application

Abstract

Lactobacilli, play a beneficial role in the female reproductive tract (FRT), regulating pH via lactic acid metabolism to help maintain a healthy environment. Bacterial vaginosis (BV) is characterized by a dysregulated flora in which anaerobes such as Gardnerella vaginalis (Gardnerella) create a less acidic environment. Current treatment focuses on antibiotic administration, including metronidazole, clindamycin, or tinidazole; however, lack of patient compliance as well as antibiotic resistance may contribute to 50% recurrence within a year. Recently, locally administered probiotics such as Lactobacillus crispatus (L. crispatus) have been evaluated as a prophylactic against recurrence. To mitigate the lack of patient compliance, sustained probiotic delivery has been proposed via 3D-bioprinted delivery vehicles. Successful delivery depends on a variety of vehicle fabrication parameters influencing timing and rate of probiotic recovery; detailed evaluation of these parameters would benefit from computational modeling complementary to experimental evaluation. This study implements a novel simulation platform to evaluate sustained delivery of probiotics from 3D-bioprinted scaffolds, taking into consideration bacterial lactic acid production and associated pH changes. The results show that the timing and rate of probiotic recovery can be realistically simulated based on fabrication parameters that affect scaffold degradation and probiotic survival. Longer term, the proposed approach could help personalize localized probiotic delivery to the FRT to advance women’s health.

1. Introduction

Bacterial vaginosis (BV) is the most common lower genital tract condition affecting women of reproductive age [1]. BV occurs due to an imbalance of the microbiota in the female reproductive tract (FRT), leading to an excess of anaerobes such as Gardnerella vaginalis (Gardnerella). Symptoms may include itching, dysuria, abnormal vaginal discharge, and foul-smelling odor. The condition can longer term lead to enhanced risk of gynecological complications such as pre-term labor and sexually transmitted diseases [2].

The current standard treatment of oral or topical antibiotics, although effective in the short term, also depletes beneficial bacteria. One of the roles of beneficial bacteria is to regulate vaginal pH by metabolizing glucose into lactic acid [3]. Healthy Lactobacilli typically reside in lower pH environments (< 4.5) whereas Gardnerella thrive at higher pH [4]; depletion of beneficial bacteria may promote higher vaginal pH that favors recurrence [5]. Although not specific to BV, elevated pH is a condition in Amsel’s criteria, commonly used to diagnose BV [6]. An ideal treatment would target Gardnerella and other undesirable bacteria while simultaneously promoting a favorable pH environment for probiotics [4]. Beneficial bacteria such as L. crispatus could be applied as probiotic treatment to achieve proactive prevention of recurrence via production of antibacterial agents and re-establishment of a healthy vaginal flora. Efficacy of probiotic administration has been investigated along with adjunct antibiotic [7–10], and has been confirmed in clinical trials, such as that of Lactin-V and others [11, 12].

Limitations exist, however, for probiotic administration to ensure efficacy, including frequent, if not daily, applications, which present a challenge for user adherence. Sustained probiotic delivery via a single treatment would, therefore, be highly desirable. 3D-bioprinted scaffolds using bio-ink that includes probiotics have been recently proposed for sustained recovery of L. crispatus into the female reproductive tract (FRT) [13]. The features of a print, such as geometry, loading, composition, viscosity, and degradation, can be precisely tailored to fit a specific need [14]. While 3D prints have shown promise in cartilage engineering and soft tissue engineering [14, 15], and antiviral drug delivery via intravaginal rings (IVRs) [16, 17], localized delivery for BV treatment has only been recently explored [18, 19].

It is not immediately apparent how to determine optimal parameter values for sustained probiotic recovery from 3D-bioprinted scaffolds, as experimental constraints preclude systematic evaluation of these values, especially when considering the longer term goal of personalized treatment. Bioprinted scaffold characteristics that can alter probiotic recovery include fabrication parameters, material properties, architecture and morphology, amount and distribution of incorporated probiotic, affinity for water, porosity, viscosity, salt and protein concentrations, pH, and system geometry. Dependent on all these parameters, the degradation rate is of critical importance for a vehicle that can offer sustained delivery of active pharmaceutical ingredients (API) (in this case, probiotics) incorporated into the biomaterial. This rate, influencing the timing and amount of probiotic recovery, requires tuning based on the fabrication parameters to promote probiotic viability while avoiding detrimental physiochemical interactions, such as could occur with antibiotic locally co-delivered with the probiotic and leading to probiotic loss.

Mathematical modeling allows for evaluation of multiple parameters independently or in combination in a systematic manner. In particular, kinetic models enable simulation and prediction of microbial processes for in vitro or in vivo applications. Modeling strategies can be divided into three general types: fully principle-based (“white box”), fully empirically-based (“black box”), and semi principle-based or semi empirically-based (“gray box”) [20]. In this study, finding the relation between the delivery of probiotics from 3D-bioprinted scaffolds and scaffold parameters is tackled as a principle-based exercise, which does not depend on correlating extensive empirical measurements to scaffold parameter values. Previous mathematical modeling by Clark et al. [21] and Halwes et al. [22] identified key delivery vehicle parameters for potential application of therapeutics to the FRT. This study builds upon the work of Halwes et al. and previous related work by Casalini et al., studying the degradation of PLGA polymer chains [23], as well as multicompartmental models by Katz and coworkers evaluating antiviral drug delivery in the vaginal environment [24, 25], to implement a model that can simulate the timing and rate of probiotic recovery from 3D-bioprinted scaffolds. The long term goal is to provide a template for future vehicle design and probiotic incorporation. Furthermore, a modeling platform enabling preclinical evaluation of sustained probiotic recovery as a function of scaffold fabrication parameters would advance the goal of effective incorporation of live cells in 3D bioprinting.

2. Methods

2.1. Scaffold Synthesis

Bioprinted L. crispatus-containing scaffolds were fabricated for calibration of the model parameters. First, bio-ink was developed with a formulated ratio of 10:2 w/w bovine skin gelatin (Sigma, G9391–100G): sodium alginate (MP Biomedicals, 218295) that was determined to be most optimal for pressure-assisted extrusion for defined construction of bioprint geometries [13]. With the addition of De Man, Rogosa, and Sharpe (MRS) broth (Sigma 69966) broth, the bio-ink was incubated overnight at 37°C for homogeneity of materials. Following incubation, subcultured L. crispatus was centrifuged, re-suspended, and added into the gelatin alginate bioink at a loading concentration of 5 × 10⁷ CFU/mg to complete the formulation. Utilizing an Allevi 3 bioprinter, the bio-ink was loaded into a syringe barrel and installed in printing head at 37°C. The bio-ink extruded with a 34G needle at 115 psi in a defined annular morphology into which the scaffolds were crosslinked with calcium chloride and genipin to retain structural integrity [26, 27].

2.2. Probiotic Recovery and Lactic Acid Production

Probiotic recovery and lactic acid production from the L. crispatus delivered by the scaffolds were evaluated in vitro. Recovery was assessed 1 through 28 days, for which scaffolds were washed in 5 mL of PBS and placed in fresh MRS prior to each timepoint. Supernatants were serially diluted by adding 20 μL to 180 μL of MRS broth, and 5 μL of each sample dilution was plated on MRS agar plates and placed in anaerobic conditions at 37°C for 48 hr. Afterwards, colonies were CFU enumerated to determine cumulative probiotic recovery.

Lactic acid production from L. crispatus-containing scaffolds was determined through serial dilutions of supernatant centrifuged at 2500xg for 5 min for separation of bacteria from the supernatant. Supernatants underwent 10-fold serial dilutions in which D(−) lactic acid was determined with a D-Lactic acid/L-lactic acid detection kit (R-biopharma; Darmstadt, Germany). The amount of L. crispatus recovery and D(−) lactic acid production were used for initial model calibration.

2.3. Mathematical Model

The model domain represents a 3D-bioprinted silicone scaffold as the delivery vehicle for encapsulation of a gelatin-alginate bio-ink containing L. crispatus. Recovery of probiotic occurs as the scaffold degrades. The delay to begin this recovery depends on the scaffold degradation rate, which is a function of the fabrication parameters. Following Halwes et al [22], the geometry represents a single vehicle placed over a layer of vaginal tissue (Figure 1), simulating the FRT environment. The scaffold is assumed to be in fluid, with a domain computationally represented as a cylindrical compartment with radius 0.5 cm ($r_F$) and thickness 0.5 cm ($h_F$) (Figure 1B).

Figure 1.

System geometry: (A) Isometric view; (B) system cross-section. Adapted from Halwes et al. [22].

2.4. Model Boundary Conditions

For simplicity, the same boundary conditions and pores sizes as in [22] are assumed. Two boundaries are considered: the scaffold center ($r=0$) and its periphery ($r=R$). Since symmetry is assumed at the scaffold center, one can define:

$${\left.\frac{\partial C_i}{\partial r}\right|}_{r=0}=0$$ (1)

$${\left.\frac{\partial D_i}{\partial r}\right|}_{r=0}=0$$ (2)

with $C_i$ as concentration of a species and $D_i$ as its diffusion coefficient.

Since mass transfer occurs at the scaffold periphery involving transport of probiotics, monomers, oligomers, water, and probiotics to the surrounding aqueous environment, this boundary condition is:

$${\left.-D_i(r=R)\frac{\partial C_i}{\partial r}\right|}_{r=R}=k_{C,i}^{ext}(C_{b,i}-C_i\left(r=R\right))$$ (3)

$C_{b,i}$ is concentration of a particular species in the exterior environment, which is assumed 0 for the monomer, oligomer, and probiotic, and 0.055 mol cm⁻³ for water. This assumption is based on the turnover of surrounding fluid in physiological systems, which represents a sink condition. The Sherwood number is a dimensionless value utilized in mass transfer to find the ratio of convective mass transfer to diffusive mass transport, and is used here to estimate the mass transfer coefficient $k_{C,i}^{ext}$. Here, the Sherwood number depends on cylinder radius $R$ and diffusion coefficient of each species $D_{i,w}$:

$$Sh=2=\frac{2k_{C,i}^{ext}R}{D_{i,w}}$$ (4)

2.5. Simulation of Scaffold Degradation

As reviewed in [28], degradation models can be established phenomenologically, probabilistically, or empirically [29]. Briefly, phenomenological models build upon governing equations describing diffusion, reaction, and dissolution. These models typically describe bulk degradation due to reaction with the environment as well as mixed surface and bulk degradation, and are applicable to various conditions, device geometries, and polymer types. Probabilistic models build upon probability distributions of molecular kinetics related to diffusion mass transfer and chemical reactions underlying the degradation of the material. In contrast, empirical models typically apply regression techniques to measure the relationships between variables, using data from controlled degradation experiments.

This study applies a phenomenological approach to model the scaffold degradation. Silicone consists of polymerized siloxanes and, as such, is expected to undergo polymeric degradation. We have recently shown that drug release from 3D printed scaffolds follows kinetics similar to that of release from polymeric matrices.[30]. Accordingly, the degradation kinetics implemented in [22] (simulating polymer degradation of a single PLGA fiber) are adapted here to represent silicone polymer scaffold degradation. Briefly, statistical moments are applied to polymer chains of length $n>9$,

$${\mu }_k={\sum }_{n=1}^{\mathrm{\infty }}(({n)}^k{C}_n)$$ (5)

where the k^th statistical moment depends on the degree of polymerization $n$ and concentration of polymers of $n$ chain length $C_n$. Thus, ${\mu }_0$ is concentration of polymer per unit volume, ${\mu }_1$ is concentration of monomers per unit volume, and ${\mu }_2$ represents polymer polydispersity. From these three statistical moments, the polymer concentration change in time can be defined.

$$\frac{\partial C_M}{\partial t}=\frac{1}{r}\frac{\partial }{\partial r}\left(D_{M}r\frac{\partial C_M}{\partial r}\right)+2k_d{C}_W({\mu }_0-C_M){\mu }_0$$ (6)

$$\frac{\partial C_n}{\partial t}=\frac{1}{r}\frac{\partial }{\partial r}\left(D_{n}r\frac{\partial C_n}{\partial r}\right)+2k_d{C}_W({\mu }_0-\sum _{j=1}^n{C}_j){\mu }_0-(n-1)k_d{C}_W{C}_n{\mu }_0,2\le n\le 9$$ (7)

$C_M$ and $C_W$ represent concentration of polymer monomers and concentration of water, respectively, while $r$ represents cylinder radius, $D_M$ and $D_n$ represent diffusion coefficients of monomer and polymer oligomer in water, and $k_d$ is a kinetic rate constant for scaffold degradation. Assuming bulk erosion, diffusion of water into the polymer matrix can be modeled as:

$$\frac{\partial C_W}{\partial t}=\frac{1}{r}\frac{\partial }{\partial r}\left(D_{W}r\frac{\partial C_W}{\partial r}\right)-k_d{C}_W\left({\mu }_1-{\mu }_0\right){\mu }_0$$ (8)

Using the statistical moments, diffusion and degradation of polymers of chain length greater than 9 can be defined [22]:

$$\frac{\partial {\mu }_0}{\partial t}=\sum _{j=1}^9\frac{1}{r}\frac{\partial }{\partial r}\left(D_{j}r\frac{\partial C_j}{\partial r}\right)+k_d{C}_W\left({\mu }_1-{\mu }_0\right){\mu }_0$$ (9)

$$\frac{\partial {\mu }_1}{\partial t}=\sum _{j=1}^9\frac{j}{r}\frac{\partial }{\partial r}\left(D_{j}r\frac{\partial C_j}{\partial r}\right)$$ (10)

$$\frac{\partial {\mu }_2}{\partial t}=\sum _{j=1}^9\frac{j^2}{r}\frac{\partial }{\partial r}\left(D_{j}r\frac{\partial C_j}{\partial r}\right)+\frac{k_d{C}_W{\mu }_0}{3}\left({\mu }_1-2\frac{{\mu }_2^2}{{\mu }_1}+\frac{{{\mu }_2\mu }_1}{{\mu }_0}\right)$$ (11)

These equations define polymer degradation and diffusion in the model, establishing a basis to calculate the timing and recovery of probiotic from the perspective of the scaffold interior.

2.6. Simulation of Controlled Release

A delay in probiotic recovery is implemented in the model to control the timing for the recovery to begin. The delay is set using a threshold $T$ for scaffold degradation status $S$, such that the threshold for scaffold degradation is met at the desired delay. For instance, if the threshold is set at 90%, a 2-day delay would signify that 90% of the scaffold has degraded by 2 days.

$$S=\frac{{\sum }_{i=1}^9{C}_i}{{\mu }_0}$$ (12)

$S$ is calculated by using concentrations of polymers of varying chain lengths. $C_i$ represents concentration of polymer chains of length 1 (monomer) to 9. $S$ is calculated at each time step of the simulation to see if the criterion has been met.

2.7. Simulation of Probiotic Recovery

Bacterial concentration within the scaffold is assumed to be at a therapeutic concentration (e.g., >5×10⁷ colony forming units (CFU)/mg). As such, it is assumed that probiotics can be treated as an aggregate entity and that the effect of individual bacteria behavior is negligible. Accordingly, probiotic concentration ($C_P$) over time and probiotic growth rate $k_P$ are represented as:

$$\frac{d{C}_p}{dt}=k_P·C_P-k_D·C_P-(k_{AD}·C_D+k_R·({\mu }_0-{\mu }_{0,t}))·C_P·S$$ (13)

$$k_P=k_{Pmax}·\left(\frac{C_G}{K_S+C_G}\right)·\left(1-\frac{C_A}{C_{Amax}}\right)$$ (14)

Further details on the dynamics represented by these equations are in Supplement. The concentrations for glucose $G_G$ and lactic acid $G_A$ are further described in Supplement.

2.8. Model Variables and Parameters

Model variables are summarized in Supplementary Table 1, while system parameters (Supplementary Table 2) were calibrated as described in Supplement. The MATLAB Partial Differential Equation toolbox was used to numerically solve the model equations.

3. Results

3.1. Experimental Characterization of Fabricated Scaffolds

Crosslinked, L. crispatus-containing scaffolds had previously demonstrated optimal mechanical integrity to retain a well-defined architecture in simulated vaginal fluid (SVF) and support bacterial proliferation [13]. Here, scaffolds exhibited probiotic recovery at a concentration of 4 ×10⁸ CFU/mg per day and 10¹⁰ CFU/mg after 28 d, cumulatively (Figure 2). Scaffolds produced greater than 40 mg of D-lactic acid /mg scaffold after 28 d, cumulatively.

Figure 2.

Characterization of 3D-Bioprinted Scaffolds. (A) Probiotic cumulative recovery profile (inset shows representative scaffold image). (B) D-Lactic acid production.

3.2. Calibration of Mathematical Model

The cumulative recovery of probiotics simulated by the model was first calibrated to the experimental results obtained in vitro (Figure 2A), under the conditions of no delay in probiotic recovery, i.e., scaffold degradation leads to immediate probiotic recovery (Figure 3A). This calibration was accomplished by setting the model parameters as described in Methods (and Supplement) to match the trend and amount of recovery observed experimentally. The corresponding overall lactic acid production calculated by the model (Figure 3B) is lower than experimentally observed (Figure 2B) as expected, since the environment external to the scaffold in which bacteria can continue proliferating and produce lactic acid is not simulated.

Figure 3.

Model calibration to experimental data under conditions of no delay to begin probiotic recovery. (A) Cumulative recovery profile and (B) corresponding lactic acid production simulated by the model (excludes lactic acid in the environment external to the scaffold).

33. Simulation of Scaffold Degradation

The scaffold degradation rate $k_d$ has a direct bearing on probiotic interactions with the surrounding environment, probiotic recovery, and eventual lactic acid production. A nonlinear relationship between the degradation rate and the time to reach a certain amount of degradation can be established in silico based on the calibrated set of parameters (Figure 4). Without loss of generality, for the simulations of delayed release an arbitrary threshold of 90% degradation was chosen before the start of probiotic recovery.

Figure 4.

Time to reach a particular scaffold degradation vs. scaffold degradation rate ($k_d$). As the scaffold degradation rate increases, the corresponding time for the scaffold to degrade is decreased. Points from left to right correspond to degradation rates: 0.014, 0.019, 0.030, 0.068, and 0.130.

3.4. Simulation of Probiotic Recovery

Simulations of probiotic recovery evaluating a range of recovery rates $k_R$ show that the amount recovered increases as the delay to begin recovery due to initial scaffold degradation is increased (Figure 5). The recovery reflects probiotic proliferation as well as the physiochemical interactions that could induce probiotic death, such as interactions with antibiotic locally co-delivered with the probiotic (as a function of the rate $k_{AD}$, chosen here to be non-zero (Supplementary Table 2) to illustrate the system dynamics). Furthermore, alteration of the probiotic recovery rate may impact the recovery kinetics; as the rate decreases, the recovery tends to saturate (Figure 5).

Figure 5.

Cumulative probiotic recovery profiles due to varying recovery rates and delays to begin this recovery (perspective is from the interior of scaffold, implying that higher recovery rate leads to lower probiotic): (A) 2d delay. (B) 5d delay. (C) 10d delay. (D) 15d delay. Recovery rate $k_R$ varies as specified in the insets.

The time it would take to recover a certain amount of probiotic (% initial concentration) as a function of the recovery rate and the delay to start recovery is in Figure 6. The time to reach certain probiotic amounts (as a function of recovery rate) increases as the delay to begin recovery increases, but the time decreases as the recovery rate increases, providing a means to select the rate of recovery based on the desired amount of recovery and the delay, with the latter depending on the scaffold fabrication parameters. For example, if a 50% probiotic recovery is desired with a 5-day delay before recovery begins, then a recovery rate of 0.025 would require 100 h after the initial 5-day delay (Figure 6B).

Figure 6.

Time to reach various probiotic recovery amounts as a function of the delay to begin recovery and the recovery rate $k_R$ (perspective is from the interior of the scaffold). (A) 2d delay. (B) 5d delay. (C) 10d delay. (D) 15d delay.

3.5. Simulation of Lactic Acid Production

The initial condition of the model features an intra-scaffold pH of 3.3, considered to be a (lower bound) equilibrium pH set by the amount of lactic acid inside the scaffold independent of the vaginal microbiome pH. A healthy vaginal pH is considered to be ~4 [31]; hence, probiotics would need to produce sufficient lactic acid to maintain this pH intra-vaginally. The simulations enable evaluating the delay in lactic acid production into the surrounding environment as the delay to begin probiotic recovery increases (Figure 7). The time it takes to attain a certain level of production increases nonlinearly as the delay to begin recovery is increased. The results further show that lactic acid production is mostly insensitive to the recovery rate, as the diffusion of lactic acid into the environment precedes the bacterial recovery out of the scaffold.

Figure 7:

Lactic acid production over time for various delays to begin probiotic recovery due to scaffold properties (perspective is from the exterior of the scaffold). Probiotic recovery rate ${\mathrm{k}}_{\mathrm{R}}$ is set to (A) 1.00×10⁻⁵ and (B) 1.00×10⁻¹.

To further assess the potential effect of the probiotic recovery rate on the lactic acid production, Figure 8 shows that the time it takes to reach a certain percent of lactic acid production (and corresponding pH shift) depend linearly on the delay to begin this recovery. With the given set of parameter values, variation in the recovery rate itself does not seem to affect this trend.

Figure 8.

Lactic acid production and pH shift due to probiotic recovery from bioprinted scaffolds as a function of the delay to begin recovery and the recovery rate $k_R$. Model-simulated times to reach (A) 50% and (B) 95% of lactic acid production, and (C) 50% and (D) 95% shift in pH (from the perspective from the exterior of the scaffold) are shown.

4. Discussion

This study implements a modeling platform to simulate sustained probiotic recovery from 3D-bioprinted scaffolds, with the longer term goal to inform the formulation and development of delivery vehicles for localized administration to prevent recurrent BV. 3D bioprinted scaffolds were first fabricated to experimentally evaluate probiotic recovery and lactic acid release. The model was calibrated to the experimental data, in which there was no delay to the probiotic recovery. This delay depends on the fabrication parameters and would be of consequence when considering co-delivery of antibiotics that could interact with the probiotic recovery. Then, the model was used to explore the relationship between this delay and the recovery rate. The results show the feasibility of simulating the timing and rate of probiotic recovery as a function of the scaffold characteristics, including its degradation rate, dependent on the fabrication parameters.

Based on the given parameter set, a nonlinear relationship between scaffold degradation rate and scaffold degradation is established, with the rate reflecting the scaffold physical characteristics (Figure 4). While the times to reach particular amounts of degradation vary significantly at low degradation rates (from 360 hours (for 90%) to 155 hours (for 10%) when the rate is 0.014), these times are all <24 hours at the highest rate (0.130). The model further shows that the probiotic recovery may saturate at lower recovery rates (Figure 5) and highlights the potential effects of physiochemical interactions that could induce probiotic death, e.g., with antibiotic locally co-delivered with the probiotic (here, included as a nonzero rate $k_{AD}$). The nonlinear nature of the probiotic recovery as a function of the scaffold parameters is reflected by the amount recovered as a function of the delay to begin recovery and the recovery rate (Figure 6).

The simulations show that the overall lactic acid produced increases only slightly as the recovery rate increases, since the lactic acid diffusion from the scaffold is based on the overall probiotic amount and not necessarily on the amount of probiotic recovery (Figure 7). With the given parameter set, the time to reach a certain percentage of lactic acid production (and corresponding pH shift) are linearly proportional to the delay to begin recovery (Figure 8). This time is mainly sensitive to this delay, indicating that lactic acid diffusing from within the scaffold could exert beneficial effects on the surrounding exterior environment without first having to wait for the bacteria to detach from the scaffold.

The proposed modeling platform could longer term enable the personalization of localized probiotic delivery to the FRT. This would build upon the notion of individualized care by simulation of different process parameters and target object characteristics in healthcare, e.g., as in insulin drug delivery (see [32]). It is further noted that the scaffold fabrication method could influence the probiotic recovery. Common polymer 3D printing methods include fused deposition modeling (FDM), selective laser sintering (SLS), digital light processing (DLP), and stereolithography (SLA). FDM, which involves extrusion of the material as a filament, is considered an economical method and was used for the scaffolds analyzed in this study. Further details on classification of 3D printing technologies can be found in [33, 34]. Future work could explore variation in probiotic recovery based on different fabrication methods.

Validation of the proposed model will require in vitro as well as in vivo evaluation. Accordingly, future work will seek to validate simulated probiotic recovery and lactic acid production from scaffolds when bioprinted with L. crispatus or other probiotics compared to experimental results based on variation in the scaffold fabrication parameters. The model could be expanded to simulate BV and thereby evaluate the competition between L. crispatus and Gardnerella to predict treatment efficacy. The co-delivery of a variety of antibiotics (e.g., metronidazole, clindamycin, or tinidazole) could also be further explored. Lastly, potential effects on scaffold degradation by fluid flow [35] in the vaginal environment deserve further study. Cervico-vaginal secretions are variable depending on age, time in menstrual cycle, pregnancy, and menopause [31, 36], and, as such, could induce variations in intra-vaginal probiotic recovery and local lactic acid concentration.

Conclusions

This study implements a novel platform to model 3D bioprinted scaffold degradation and probiotic recovery, and to evaluate the timing and rate of probiotic recovery based on the fabrication parameters. The results highlight the nonlinear dynamics of the probiotic recovery from 3D-bioprinted constructs for vaginal application. As personalized medicine evolves, customization of therapy could benefit from computer-aided design and 3D printing of API formulations informed by modeling. In addition to fabrication parameters, a variety of geometries could be evaluated via simulation, e.g., capsules or IVRs, with varying diameters, thicknesses, lengths, or widths. The model presented here offers a step towards more effectively meeting therapeutic benchmarks for individualized care.