A process-driven approach for manufacturing personalized intravaginal rings using droplet deposition modeling

Abstract

Droplet Deposition Modeling (DDM) is an additive manufacturing technology with potential applications in personalized drug delivery systems, enabling the production of drug products tailored to individual patient characteristics. This study explores the application of DDM in manufacturing customized intravaginal rings (IVRs) using β-estradiol as a model drug, focusing on how three key printing parameters—infill density (ID), droplet aspect ratio (DAR), and discharge rate (DR)—affect the performance characteristics of 3D-printed IVRs. These characteristics include surface morphology, porosity, weight, mechanical and viscoelastic properties, as well as in vitro drug release. The 3D-printed IVRs exhibited high dimensional accuracy and uniformity within batches. The results demonstrated that increasing ID from 70% to 90% and DR from 56% to 62% led to higher weight, Shore M hardness, compression strength, Young’s modulus, storage modulus, and loss modulus, while decreasing porosity. On the other hand, increasing DAR from 1.1 to 1.3 produced opposing trends. Drug release was predominantly influenced by variations in ID and DAR, with higher ID values associated with slower drug release rates. These findings provide insight into the mechanisms governing the DDM process in pharmaceutical applications and highlight the importance of printing parameters for optimizing the quality and functionality of personalized IVRs. This work demonstrated the potential of additive manufacturing in enabling personalized medicine and distributed manufacturing of drug delivery systems, contributing to the advancement of patient-specific healthcare solutions.

1. Introduction

Intravaginal rings (IVRs) are flexible, toroidal-shaped drug delivery devices manufactured from polymeric materials with a dispersed active pharmaceutical ingredient (API) for localized release in the vaginal cavity. IVRs offer several advantages, including enhanced safety, localized drug delivery, controlled release of drugs, and high patient compliance (Boyd et al., 2020; Tiboni et al., 2021). While commercially available IVRs have primarily been used as contraceptive devices and for hormone replacement therapy (Boyd et al., 2020), recent research has increasingly focused on their potential as promising delivery systems for microbicides against sexually transmitted infections and diseases (Tietz and Klein, 2019; Verstraete et al., 2017). Currently, marketed IVRs are manufactured with fixed sizes, shapes, mechanical properties, and dosing regimens, following a one-size-fits-all principle, which fails to address the diverse needs, anatomical variations, and personal preferences of individual patients (Koutsamanis et al., 2021; Sierra-Vega et al., 2024). This has led to a growing need for personalized IVRs that may be tailored to address these diverse needs. However, the mass production of such personalized IVRs remains challenging with the traditional manufacturing processes based on injection molding or hot melt extrusion, which are not easily adaptable for customization due to high setup costs, tooling requirements, and validation complexities that make small-batch variants economically unfeasible. Alternatively, additive manufacturing, more commonly known as three-dimensional (3D) printing, is an advanced manufacturing technology with the potential to produce customized drug products tailored to the specific needs of patients (Alvarez-Lorenzo et al., 2024; FDA, 2023; Trivedi et al., 2018).

Droplet deposition modeling (DDM) is a 3D printing technology that enables the manufacturing of complex thermoplastic components using standard material pellets. This eliminates the need for solvents, specialized resins or filaments, and multiple melting cycles (Sierra-Vega et al., 2025). The potential of DDM for the manufacturing of solid oral dosage forms and IVRs has been demonstrated in previous studies (Ebrahimi et al., 2024; McDonagh et al., 2022a, b, 2023; Sierra-Vega et al., 2024; Welsh et al., 2019; Zhang et al., 2021). DDM process begins with melting pellets of plastic materials in a plasticizing cylinder, after which a closable piezoelectric shut-off nozzle operates at high-frequency deposits droplets that join to create a continuous droplet chain, as shown in Fig. 1(a). However, as a relatively new technique, the full understanding of how printing parameters in DDM affect the quality and performance of 3D-printed drug products is still evolving. Several factors – including material attributes and process and geometric parameters – may influence the performance of the final drug product. Understanding the relationships among these factors is crucial for ensuring consistent product quality and optimizing DDM for pharmaceutical applications.

Fig. 1.

Visualization of the DDM process: (a) Most important components of the Freeformer. Effect of (b) ID and (c) DAR (when ID of 100% is used) on droplet deposition.

In DDM, the design space of the printing parameters should be defined based on the critical attributes of the printing material being processed and the performance of the drug product. Infill-density (ID) is the main parameter to control the filling degree, while the drop geometry and deposition are controlled mainly by layer thickness (LT), droplet aspect ratio (DAR), and discharge rate (DR). The ID, LT and DAR are parameters set by slicer software, while DR is a process parameter that is tuned by the printer during the development of the manufacturing process. The slicer software divides the 3D model into cross-sectional layers of a specific LT and generates parallel deposition paths spaced at a distance T for each layer. This distance T is inversely related to ID, as shown in Fig. 1(b). Increasing the ID results in a decrease of the distance T, allowing for a greater density of droplets within each layer. ID does not depend on the melted material pressure at the nozzle head. Previous research has shown that reducing the ID increases porosity and enhances drug release from solid oral dosage forms (Ebrahimi et al., 2024; McDonagh et al., 2022a, b, 2023; Zhang et al., 2021). For example, Zhang et al., observed a linear relationship between ID and release rate in tablets based on hypromellose acetate succinate and polyethylene oxide, manufactured via DDM, suggesting that ID could be used as a tailoring parameter for personalization of drug release from swellable and erodible systems (Zhang et al., 2021). The impact of ID on both mechanical properties and drug release has also been studied in drug loaded IVRs manufactured through DDM. For instance, DDM was used to improve the release properties of poorly water-soluble dapivirine, demonstrating that decreasing the ID from 100 % to 10 % enhanced dapivirine release from IVRs (Welsh et al., 2019). Our previous study also highlighted the impact of ID on the mechanical properties and drug release of β-estradiol-based IVRs, showing that increasing ID reduces porosity and release rate while enhancing mechanical properties, weight, and drug dosage of IVRs (Sierra-Vega et al., 2024). In addition, our previous study included a detailed characterization of the thermal stability of β-estradiol, as well as its molecular and physical interactions with TPU during the 3D printing process. These results confirmed that β-estradiol maintained its stability throughout the process temperature scheme used in this study (Sierra-Vega et al., 2024).

While the studies mentioned above primarily focus on ID as a controlling factor for the performance attributes of 3D-printed drug products in DDM, the DAR also plays an important role in optimizing the geometry and deposition of the droplets within the printed layers. Fig. 1 (c) illustrates the effect of DAR on droplet deposition in the DDM process. When an ID of 100 % is used, the distance $T$ corresponds to the ideal width ($W_{id}$) of the droplets. The DAR is defined as the ratio of the $W_{id}$ of the droplet to the length (Eq. (1). It is important to note that DAR does not directly influence the actual width ($W_{ac}$) of the droplets deposited. The $W_{ac}$ is determined by the thermal and rheological properties of the polymer used for printing. Variations in DAR may impact the spacing between the centers of two successive droplets and the similarity between $W_{id}$ and $W_{ac}$, as illustrated in Fig. 1(c). When $W_{ac}=W_{\mathit{id}}$, the droplets are ideally tangent to one another, and the layers are filled according to the geometry defined by the slicer software. However, an increase in DAR results in $W_{ac}<W_{id}$, which leads to less droplets deposited within the layer, more void spaces, and the emergence of printing defects that can affect the product’s quality. Conversely, a decrease in DAR results in $W_{ac}>W_{id}$, which reduces the void space in the layer and may cause the deposition to exceed the nominal profile (Profili et al., 2025).

$$\mathit{D}\mathit{A}\mathit{R}=\frac{{\mathit{W}}_{\mathit{i}\mathit{d}}}{\mathit{L}\mathit{T}}$$ (1)

Another key parameter affecting drop generation in DDM is the DR, which defines the volume of melted material extruded in each droplet. DR governs the volumetric flow during the process and represents the screw stroke increment $L$ compared to a reference stroke increment $L_0$ of 0.115 mm (Eq. (2), as shown in Fig. 1(a) (Eisele et al., 2023). DR depends on the pressure applied by the screw on the melted material and the closure frequency. The droplet volume calculated by the slicing software is independent of the DR. At a constant DAR, an increase in DR leads to a higher flow of the material in each droplet, which results in a larger $W_{ac}$ (Fig. 1(c)), while a decrease in DR results in a smaller $W_{ac}$. Optimizing both DAR and DR, along with the ID, is then critical to avoid any printing defects and to ensure acceptable and consistent deposition patterns of the droplets throughout the printing job. As the customization parameters for personalizing the performance of IVRs, this optimization is essential to meet the specific needs of individual patients.

$$DR=\frac{L}{L_o}=\frac{L}{0.115mm}\times 100\mathrm{\%}$$ (2)

Previous research has demonstrated that DAR and DR significantly impact the mechanical properties of 3D-printed solid objects in DDM, even when ID is kept constant. Specifically, previous studies have shown that reducing DAR may enhance the mechanical properties of 3D-printed solids, assuming all other printing parameters remain unchanged (Charlon and Soulestin, 2020; Hentschel et al., 2023). Similarly, using low DR was linked to reduced porosity, density, and tensile strength of 3D-printed objects (Charlon and Soulestin, 2020; Eisele et al., 2023). However, the effects of DAR, DR, and their interactions with ID on properties and critical performance attributes of drug products have not yet been investigated.

The current study aimed to define a design space for a DDM process to manufacture customized β-estradiol-based IVRs tailored to meet the specific needs of patients. β-Estradiol was selected as a model active ingredient due to its efficacy in treating postmenopausal vaginal atrophy and other estrogen-deficiency–related disorders. However, systemic administration of estradiol has been linked to an increased risk of adverse events such as venous thromboembolism and stroke (Abou-Ismail et al., 2020), emphasizing the need for delivery systems that reduce systemic exposure and allow individualized dosing. Building upon our previous work (Sierra-Vega et al., 2024), this study investigated the effects of ID, DAR and DR and their interactions on the customization responses of IVR through a printing design space constructed based on central composite design (CCD). The hypothesis underlying this research was that controlled changes in ID, DAR and DR could be leveraged to customize physicochemical, mechanical, and drug release characteristics of 3D-printed IVRs. TPU was employed as the polymeric carrier for IVRs, while a water-soluble polymer, Armat11, was used as a supporting material. The customization responses included weight, porosity, Shore M hardness, Young’s modulus, compression strength, storage modulus, loss modulus, as well as the cumulative percentage and rate of β-estradiol released from these additively manufactured IVRs.

2. Materials and methods

2.1. Materials

Pellets of an amorphous thermoplastic polyurethane (TPU) elastomer (Elastollan 1170 A 10 000) from BASF (New York, USA) were used as a polymeric carrier. β-estradiol (1,3,5-Estratriene-3,17-β-diol) from MilliporeSigma (St. Louis, MO) was used as a drug model. Castor oil from Fisher Scientific (Hampton, NH) was used to coat the pellets before extrusion. ARMAT 11 (USP grade) from Arburg GmbH + Co KG (Lossburg, Germany) was used as the support material. Sodium chloride, potassium hydroxide, calcium hydroxide, bovine serum albumin, lactic acid, acetic acid, glycerol, urea, glucose, sodium azide, and Tween 80 from MilliporeSigma (St. Louis, MO) were used for the release studies. High-performance liquid chromatography (HPLC) grade water was obtained by Milli pore Direct Q3 UV ultrapure water.

2.2. Preparation of printing materials

The oil-coating method was used to load β-estradiol onto the surface of TPU (Tappa et al., 2017). TPU was dried before mixing in a vacuum oven at 80 ± 1°C for 2.5 h to ensure a moisture content ≤0.04 %. β-estradiol was dissolved in castor oil at a concentration of 25 %w/w using a magnetic stirrer. This β-estradiol in castor oil blend was then blended with TPU pellets at a loading of 0.4 %w/w using a Turbula blender (WAB group, Old Bridge, NJ, USA) for ten minutes. The loading concentration of β-estradiol in TPU was ~0.1 %w/w.

2.3. Droplet deposition modeling process

An Arburg Plastic Freeformer (APF) 300–3X, manufactured by Arburg GmbH + Co KG, Lossburg, Germany, was used for printing of IVRs. This printer was equipped with three independent material dispensing units, each fitted with a heated nozzle having a diameter of 0.2 mm. The printer operated in three continuous steps: feeding, plasticizing, and droplet deposition. Initially, granular material was fed into the hopper and dispensed gravimetrically. Subsequently, the material was melted in the plasticizing cylinder, which had an internal diameter of 15 mm heated by two heating zones: zone 1 and zone 2 (Sierra-Vega et al., 2025). The cylinder contained a rotating and longitudinally movable screw, which helped to reach the necessary pressure for the deposition process. A non-return valve ensured that once the desired pressure was achieved, no material was drawn back into the system. After plasticization, the molten material entered the discharge unit (210 °C), where a piezoelectric shut-off nozzle operating at a discharge frequency of 250 Hz precisely controlled the ejection of the material in the form of discrete droplets. These droplets were deposited in a continuous chain onto a vacuum-held polycarbonate sheet hold by vacuum on the build platform in a heated chamber (90 °C). The build platform was capable of movement along three axes to facilitate the layer-by-layer printing process. Notably, this temperature scheme for extrusion and deposition of β-estradiol loaded TPU matrix were developed and optimized in our previous study (Sierra-Vega et al., 2024).

2.4. IVR design and DDM printing space

The torus matrix IVRs were designed with an outer diameter of 35 mm and a cross-section diameter of 5.75 mm using SolidWorks software (version SP 4.1, Dassault Systems SolidWorks Corp., Waltham, MA). These dimensions were selected to demonstrate the manufacturing flexibility of 3D printing while allowing for efficient material usage and comprehensive evaluation of printing parameters. STL (standard triangle language) files were generated and loaded into the Arburg Freeformer slicing software v.2.30 (ARBURG GmbH + Co KG, Lossburg, Germany). A central composite design (CCD) was used to investigate the main and interaction effects of the printing parameters on the mechanical and drug release characteristics of the IVRs. This CCD was constructed and analyzed using JMP statistical software (JMP 17, SAS Institute Inc., Cary, North Carolina, USA). An orthogonal failure edge of 1.287 % was implemented in the CCD. Supplementary Fig. S1 shows the cube design of the CCD used in this study. The input parameters used in the CCD were ID, DAR, and DR. A total of 17 printing runs, including three central points, namely R8, R9, and R10, were performed, as shown in Table 1. The levels of ID, DAR and DR in CCD ranged from 50 % to 90 %, 1.2 to 1.3, and 56 % to 62 %, respectively. These levels were established based on the results stated in our previous study to avoid any printing defects observed outside this design space (Sierra-Vega et al., 2024). These printing defects included excessive over-extrusion, lamination, imperfect fusion, non-uniform porosity, non-uniform density, layer separation and stringing (Sierra-Vega et al., 2024). The response variables in CCD were weight, porosity, shore M hardness, Young’s modulus, compression force, storage modulus, loss modulus, percent β-estradiol released at 14, 21, 35, 42, and 49 days from IVRs and release rate. The statistical analysis was performed with a 95 % confidence level, and p-values < 0.05 were considered statistically significant. The R², p-values and Root Mean Square Error (RMSE) were used as key indicators to quantify how well the regression model fits the dataset.

Table 1.

Central composite design matrix for DDM parameters in IVR manufacturing.

Run	DAR	ID (%)	DR (%)
R1	1.18	70	59
R2	1.2	50	56
R3	1.2	50	62
R4	1.2	90	56
R5	1.2	90	62
R6	1.25	44.3	59
R7	1.25	70	55.1
R8	1.25	70	59
R9	1.25	70	59
R10	1.25	70	59
R11	1.25	70	62.9
R12	1.25	95.7	59
R13	1.3	50	56
R14	1.3	50	62
R15	1.3	90	56
R16	1.3	90	62
R17	1.31	70	59

Table 2 lists the fixed DDM process parameters across all runs to optimize the consistency, quality, and performance of IVRs. The temperatures for the discharge unit (nozzle temperature) (210 °C), heated zone 2 (180 °C), and heated zone 1 (160 °C), and chamber (90 °C) were selected to control the viscosity and flow of TPU, ensuring optimal droplet formation and deposition while preventing material degradation (Hentschel et al., 2020; Hentschel et al., 2023). Dosage strokes were set at 6 mm for loaded drug TPU and 9 mm for ARMAT 11 to achieve consistent material flow, with back pressures of 40 bar and 50 bar, respectively, regulating the extrusion process and ensuring stable droplet formation. The screw speeds were fixed at 3 m/min for loaded drug TPU and 4 m/min for ARMAT 11 to ensure proper homogenization of the material within the plasticating cylinder. An LT of 0.2 mm was selected to match the nozzle diameter, allowing for high-resolution deposition, while an increment angle of 90° was selected to ensure uniformity of layer buildup. Following the manufacturer’s recommendations, an ID of 90 %, a DAR of 1.66, and a DR of 90 % were used for ARMAT 11 in all runs to ensure consistent printing quality of the support material.

Table 2.

Fixed printing parameters for DDM manufacturing of IVRs.

Parameter	TPU *	ARMAT 11
Temperature zone 1 (°C)	160	160
Temperature zone 2 (°C)	180	190
Temperature discharge unit (°C)	210	210
Chamber temperature (°C)	90	90
Dosage strokes (mm)	6	9
Back pressure (bar)	40	50
Screw speed (m/min)	3	4
Layer height (mm)	0.2	0.2
Increment angle (°)	90	90
ID (%)	–	90
DAR	–	1.66
DR (%)	–	90

^*ID, DAR and DR for TPU were included as factors in the CCD.

2.5. Physicochemical characterization of IVRs

2.5.1. Scanning electron microscopy (SEM)

A JCM-7000 NeoScope^™ Benchtop SEM (JEOL LTD, Hertfordshire, UK) was used to visualize the surface and cross-section morphological characteristics of the IVRs. A high vacuum was applied in the observation chamber. The accelerating voltage was adjusted at 5 kV at the magnifications x30 and x100. All samples were attached to SEM stubs using double adhesive tape. SEM main software (JEOL Ltd., Hertfordshire, UK) was used to control the equipment.

2.5.2. Weight and dimensions

The weight of individual IVRs from all printing runs was measured using a 4-point balance (Mettler Toledo, Columbia, MD). The horizontal cross-sectional diameter and height of the IVRs were measured using 200 mm digital calipers (Starrett 798, Athol, MA, USA). The weight and cross-sectional diameters for six IVRs were determined and reported as mean and standard deviation.

2.5.3. Porosity

The internal microstructure of the IVRs were visualized and analyzed using a high-resolution X-ray Micro-CT (μCT) scanner (SkyScan 1272, Bruker, Kontich, Belgium) operated at 70 kV of source voltage and at a constant source current of 142 μA. Samples were scanned over a 360° range with a rotation step of 0.8°, and 1.55 s exposure time per frame. An aluminum filter of 0.55 mm was used in the scanning procedure. The resulting 2D images were reconstructed using NRecon and Data-Viewer software (Version 2.2.0.6, SkyScan, Kontich, Belgium). Porosity analysis was conducted using CTAn software (version 1.16.4.1, SkyScan, Kontich, Belgium), which uses a global thresholding and segmentation approach to distinguish solid material from void space. CTAn calculates porosity in terms of open, closed, and total porosity. Open porosity corresponds to the volume of open pores as a percentage of the volume of interest (VOI), while closed porosity is calculated as the volume of closed pores as a percentage of the total material volume. The μCT total porosity was calculated as the volume of all pores (open and closed pores) as a percent of the total solid plus pore (open and closed) volume within the VOI (Bruker, 2020). Notably, the percent of total porosity does not always equal the sum of the percent open porosity and the percent closed porosity (Bruker, 2020).

2.5.4. Shore M hardness

The Shore M hardness of the 3D-printed IVRs was measured using a digital durometer hardness tester (HPE-III-M series, Bareiss, Germany). This instrument was equipped with a micro-M gauge with a sharp 30° angle indenter and a maximum spring force of 78 g. All Shore M hardness measurements were conducted under controlled conditions of temperature (22 °C ± 2°C). Three IVR samples were randomly selected from each manufactured batch for these measurements.

2.5.5. Tensile and compression strength

The tensile strength of IVRs was tested using a texture analyzer (TA. TX. Plus, Stable Microsystems Ltd, Surrey, UK) under controlled conditions of temperature (22 °C ± 2 °C). Each IVR was cut into segments approximately 30 mm in length prior to testing. Tensile grips were installed on the texture analyzer, leaving a 10 mm gap between the two grips. The test was performed in tension mode using a “return to start” protocol. The IVR sample was held with the tensile grips, ensuring it was free of initial tension. The test was programmed using Exponent software (Stable Microsystems Ltd., Godalming, UK), applying a tensile load at a constant speed of 0.5 mm/s. The measured forces were recorded as a function of both time and displacement. The stress and strain were calculated based on the force–displacement data using Eqs. (3) and (4), respectively, where σ is the stress (N/m²), F is the force acting parallel to the area, A is the cross-sectional area, ∈ denotes strain (dimensionless), $\Delta l$ is the change in length, and $l_0$ is the initial length. Young’s modulus was calculated from the slope of the linear region in the stress–strain curve.

$$\sigma =\frac{F}{A}$$ (3)

$$\in =\frac{\mathrm{\Delta }l}{l_o}$$ (4)

The compression strength of IVRs was also determined using the texture analyzer. Each ring was placed in the ring holder and compressed by a probe attached to the moveable arm of the Texture Analyzer. The test was performed in compression mode with a target distance of 5.0 mm and a test speed of 2 mm/s. Six compression cycles were applied to each IVR sample, and the maximum compression forces for the last five compression cycles were averaged.

2.5.6. Storage modulus, loss modulus and Tan $\mathit{\delta }$

A dynamic mechanical analyzer (Q800, TA instruments, Leatherhead, UK) was used to determine the storage modulus, loss modulus and Tan $\mathrm{\delta }$ of the IVRs (n = 3). The multifrequency strain mode test was performed under a nitrogen atmosphere using a tension clamp. The 3D-printed IVRs were previously bisected and standardized to a length of 30 mm. The DMA testing was performed on one half of the ring with a preload force of 0.01 N. The material was heated at a constant rate of 3 °C/min from 35 °C to 150 °C. The material was then deformed at a constant amplitude of 15 μm over a frequency range from 0.01 to 200 Hz while mechanical properties such as storage modulus (G’), loss modulus (G”) and Tan δ were recorded as a function of temperature.

2.5.7. Drug content Assay

IVRs were radially sectioned into ~7 mm segments with a scalpel. Each segment (n = 3) was immersed in 2 mL tetrahydrofuran and sonicated for 30 min to ensure complete disintegration. Methanol (16 mL) was then added and the mixture vortexed for 30 s. An aliquot (1 mL) was filtered through a 0.4 μm PVDF membrane to remove particulates. A 500 μL portion of the filtrate was diluted 1:1 with methanol (additional 500 μL) and the resulting solution was injected into the HPLC for analysis.

2.5.8. In vitro release testing (IVRT)

In vitro drug release characteristics were evaluated under accelerated release conditions using a shaking water bath and a simulated vaginal fluid (SVF) as the release medium. The SVF was prepared according to the recipe reported by Owen and Katz (Owen and Katz, 1999), by dissolving 3.51 g sodium chloride, 1.40 g potassium hydroxide, 0.22 g calcium hydroxide, 0.018 g bovine serum albumin, 2.0 g lactic acid, 1.0 acetic acid, 0.16 g glycerol, 0.4 g urea, 5.0 g glucose in 1000 mL of deionized water. The pH was adjusted to 4.2 using HCl. The acceleration of drug release was induced by heating the media to 42 ± 0.5 °C and continuous agitation at 80 RPM. These accelerated release conditions were previously optimized to preserve the drug release mechanism while accelerating release rates for comparative evaluation of IVR performance (Sierra-Vega et al., 2024). This experimental design ensured maintenance of the sink conditions throughout the test. Given the β-estradiol solubility of 15.1 μg/mL in the release medium, each intravaginal ring was immersed in 45 mL of SVF to prevent saturation. Aliquots were collected at predetermined intervals (7, 21, 28, 35, 42, and 49 days), and cumulative β-estradiol release was quantified using HPLC analysis. All dissolution tests were performed in triplicate for each IVR run.

The Higuchi model was used to fit the drug release data (Higuchi, 1963). The Higuchi model generally interprets the release from matrix-type systems containing dispersed drug particles that occurs through fluid-filled pores, where the release rate decreases with time due to the depletion of the drug in the matrix. Eq. (5) describes the Higuchi model, where $M_t$ is the cumulative amount of drug released at time $t$, and $\mathrm{k}$ is the Higuchi constant $\left(\mathrm{\%}/\right.{\mathit{day}}^{0.5})$.

$$M_t=k\sqrt{t}$$ (5)

2.5.9. HPLC analysis

β-estradiol concentrations in IVRT samples were determined using an Agilent 1260 HPLC system (Agilent 1260 series, Agilent Technologies, Santa Clara, CA) equipped with a binary solvent pump, autosampler, photodiode array detector, thermostatic column compartment, and chem station chromatographic software (Agilent Technologies, Santa Clara, CA). A Phenomenex Kinetex C18 column (2.6 μm, 4.6 × 150 mm) with an isocratic mixture of acetonitrile, water, and THF in the ratio of 45:50:5 and flow rate of 0.8 mL/min at 37 °C was used to elute β-estradiol. The mobile phase was a mixture of acetonitrile, water, and THF in the ratio of 45:50:5 and was pumped isocratically in the system at a flow rate of 0.8 mL/min. Standard β-estradiol solutions (100 μL) in the range of 0.05 to 10 μg/mL were injected into the HPLC column, and β-estradiol was detected at 225 nm. The HPLC method was developed in-house and fully validated over a three-day period.

3. Results and discussion

3.1. Microstructure of the IVRs

The surface characteristics of IVRs are important factors that influence both tactile and visual perception, as well as the efficacy of IVR therapy (Vargas et al., 2019; Welsh et al., 2019). The SEM micrographs presented in Fig. 2, provide detailed analysis of the morphological features of 3D-printed IVRs. These images include cross-sectional views at magnifications of x30 and x100, as well as top views at x30, for IVRs from printing runs R1, R7, R6, R11, R12 and R17. These micrographs reveal distinct morphological variations in the IVRs, which results from the modification of one printing parameter while maintaining the other two parameters constant.

Fig. 2.

Scanning electron micrographs of cross-section and top surface of 3D-printed IVRs (R1, R17, R6, R12, R7 and R11).

The effect of ID on the morphology of IVRs was investigated by comparing SEM images from runs R6 and R12, where ID was varied from 44.3 % to 95.7 %, while the DAR and DR were fixed at 1.25 and 59 %, respectively. Cross-sectional views of these samples show that increasing the ID reduces the interstitial space between droplet chains and minimizes void areas. In particular, the droplet chains visible in the IVR sample from R6 become indistinguishable in the IVR sample from R12. Top-view images further demonstrate that higher ID values contribute to a smoother, more compact surface with reduced void presence. The reduced spacing between droplet chains of the melt blend at higher ID increases the area of contact between polymer droplets.

The SEM micrographs of IVRs from runs R1 and R17 in Fig. 2, illustrate the effect of DAR on the morphology of IVRs. In these runs, DAR varied from 1.18 to 1.31, while the ID and DR were kept constant at 70 % and 59 %, respectively. The SEM images showed that IVR from R1 exhibits a homogeneous structure with fewer visible droplet chain separations, while IVR from R17 exhibits distinct droplet chains and more noticeable voids. The top-view images further show that the increase in DAR contributes to a rough surface finish, with large voids and increased deformation in the droplet chains in IVR of R17. In contrast, when only the DR was altered from 55.1 % (R7) to 62.9 % (R11), the IVR surface exhibited slight densification and fewer imperfections; however, the droplet chains remained visible, suggesting less impact on the overall surface smoothness compared to the variation in ID and DAR.

3.2. Dimensional analysis of IVRs

The weight and geometrical accuracy of the IVRs are critical for personalization, as they directly influence drug loading. The box plots in Fig. 3 summarize the weight, horizontal cross-sectional (CS) diameter, and height of the IVRs across all manufacturing runs. As shown in Fig. 3 (a), the heaviest IVRs were manufactured in run R4 (DAR = 1.2, ID = 90 %, DR = 56 %), while the lightest were from run R13 (DAR = 1.3, ID = 50 %, DR = 56 %). All runs exhibited low standard deviations, highlighting consistent manufacturing. Similarly, Fig. 3(b) demonstrated minimal variations in the horizontal CS diameters, which ranged from 5.32 mm (R15) to 5.64 mm (R8) and remained within 10 % of the target CAD model dimensions, indicating a low horizontal deviation. In contrast, height measurements (Fig. 3(c)) ranged from 4.28 mm (R13) to 5.93 mm (R5), with low standard deviations. Notably, most IVR height values fell below the lower 10 % threshold of the CAD model, suggesting significant vertical deviation compared to the deviation of horizontal CS diameter. The low standard deviations within each printing run highlight the high batch-to-batch reproducibility of the DDM process. The observed variations in weight and geometry were attributable to the intentional modulation of ID, DAR and DR within the design space.

Fig. 3.

Box plots for (a) weight, (b) horizontal CS diameter, and (c) height measurements of IVRs under printing conditions listed in Table 1. The blue dashed line and red dash-dot line represent the CAD diameter (5.75 mm), and it is ±10 % dimensional variations, respectively.

The effects of the printing parameters on the weight, horizontal CS diameter, and height of the IVRs were investigated using CCD regression models. PLS regressions yielded strong correlations for both weight and height, with R² > 0.99 and p-values < 0.0001. However, the horizontal CS diameters showed low variability across runs and no statistically significant relationship with any of the investigated parameters was found (R² < 0.35 and p-value > 0.05). The observed sensitivity of the IVR height – but not of the horizontal CS diameter – to the printing parameters was attributed to two factors. Firstly, the layer-by-layer deposition process inherently introduced vertical gravimetrical compression, as upper layers exerted stress on lower layers, potentially causing deformation and affecting the height of IVRs. This effect was more pronounced at lower ID, where increased spacing between droplet chains in underlying layers led to deformation and reduced height. Secondly, horizontal dimensions of IVRs in DDM process were constrained by a predefined contour line deposited before the internal droplet chains of each layer, which maintained the horizontal stability and minimize variations in the horizontal CS diameter.

The effect plots in Fig. 4(a) illustrate the impact of ID, DAR, DR, and their significant interactions on the weight and height of IVRs. An effect plot displays the change in the response variable as a factor varies from a low to a high level, with all other factors held constant at their average values. The error bars represent a 95 % confidence interval. An effect was considered statistically significant when the confidence interval does not include zero. Among the printing parameters, ID was found to be the most significant factor influencing the weight of IVRs. Specifically, increasing ID resulted in a substantial increase in the weight of the IVRs. For example, increasing ID from 50 to 90 % at a DAR of 1.2 and DR of 56 % led to an increase in the weight of IVRs from 1.41 g (R2) to 2.37 g (R4). As ID increased, the distance between adjacent droplet chains decreased; leading to a higher deposition of melted polymeric material within each printed layer (Fig. 1(b)), which directly contributed to the increased weight of the IVRs. These findings align with those reported in our previous study (Sierra-Vega et al., 2024).

Fig. 4.

(a) Plot of the effects of printing parameters on the weight and height of 3D-printed IVRs; and (b) Surface plot for the weight of IVRs as a function of ID and DAR (Red spheres represent the experimentally measured values for each DOE run).

An inverse relationship was observed between the weight of IVRs and DAR. Increasing DAR from 1.2 to 1.3 at a DR of 56 % and an ID of 50 % resulted in a decrease of IVR weight from 1.41 g (R2) to 1.04 g (R13), respectively. This weight reduction may be linked to the increased distance between droplets at higher DAR, as shown in Fig. 1(c), leading to a lower overall mass of the melted material deposited per unit length of the droplet chain. The results in Fig. 4(a) demonstrated that the interaction between DAR and ID negatively influenced the weight of IVRs. While DAR had a significant effect on weight at an ID of 50 %, this effect was diminished when an ID of 90 % was used, as ID became the most dominant factor in determining the weight of IVRs (Fig. 4(b)). Furthermore, the results in Fig. 4(a) highlighted a positive effect of DR on the weight of IVRs. When DAR was fixed at 1.2 and ID at 50 %, increasing DR from 56 % to 62 % resulted in an increase in the weight of IVRs from 1.4 g (R2) to 1.55 g (R3). This positive effect of DR on the weight of IVRs may be attributed to the densification of droplets within the printing path, leading to a higher overall material deposition and, consequently, increased weight. These observations underscore the importance of considering these three factors – ID, DAR and DR – not only for customizing the degree of filling or density of the IVRs but also for achieving a desired performance to meet specific requirements for clinical applications.

Fig. 4(a) illustrates the effect of ID, DAR, and DR on the height of IVRs, with no significant interaction observed among these factors. The results indicated that increasing either ID or DR led to an increase in the height of IVRs and reduced deviations from the target CAD model. As discussed previously, higher ID or DR narrowed the gaps between adjacent droplet chains within a layer, providing a more stable underlying support structure that restricted vertical deformation during subsequent layer deposition. In contrast, an increase in DAR widened the spacing between droplet chains within a layer, creating more voids and reduced interlayer support. The increased void volume facilitates vertical deformation during deposition, leading to greater height deviation from the CAD model. As shown in Fig. 2, increasing DAR from 1.18 to 1.31, resulted in greater spacing between droplet chains and the formation of more void spaces.

3.3. Porosity of the IVRs

The porosity of the IVRs was analyzed using high-resolution μCT imaging. The Supplementary Fig. S2 presents reconstructed images of IVRs from runs R2, R3, R4, and R13 obtained by the μCT analysis. The results in Fig. 5(a) illustrated the percentages of both open and closed porosity across all printing runs. Deliberate changes in ID, DAR and DR within the design space of CCD caused variations in both open and closed porosity, which subsequently affected the total porosity of the IVRs. Specifically, IVRs from run R13, printed with an ID of 50 %, a DAR of 1.3, and a DR of 56 %, exhibited the highest porosity, while IVRs from run R5, printed with an ID of 90 %, a DAR of 1.2, and a DR of 62 %, exhibited the lowest porosity. All IVRs showed closed porosity values below 2.5 %, with open porosity varying between 10.7 % (R5) and 58.6 % (R14). The total porosity ranged from 10.9 and 58.6 %. These findings suggested that the total porosity of the 3D-printed IVRs was predominantly determined by the open porosity. As shown in Supplementary Fig. S3, variations in ID, DAR, and DR did not produce significant changes in the spatial distribution of pores within the ring matrix. It may be noteworthy that μCT analysis may have limitations in detecting micropores that may be formed within a droplet chain because of droplet deposition, deformation, and fusion. These limitations may result in an underestimation of the recorded closed porosity values in μCT analysis (Elkhoury et al., 2019).

Fig. 5.

(a) Closed and open porosity values for IVRs across all manufacturing runs; and (b) Plot of the effects of printing parameters on the open and total porosity of IVRs.

The effect of the investigated parameters and their interactions on the open, closed, and total porosity of IVRs was analyzed using the regression models in CCD. The PLS regressions predicted open and total porosity with R² > 0.96 and p-values < 0.05. However, due to the low variation and smaller values observed for the closed porosity, no significant relationship was found between the closed porosity and any of the investigated printing parameters (R² < 0.65 and p-value > 0.05). The effect plot in Fig. 5(b) illustrates the influence of ID, DAR, and DR on the open and total porosity of IVRs. As shown in Fig. 5(b), a significant inverse proportionality was observed between ID and both the open and total porosity. These results were in good agreement with the SEM images (Fig. 2), where 3D-printed IVRs processed with an ID of 50 % exhibited large void spaces between adjacent droplet chains. During the layer-by-layer deposition process, when the printer deposits a new layer, voids would appear in those regions where the spaces occur in the previous layer. As ID increased, the area of contact of the adjacent chain of droplets increased, resulting in a lower number and size of pores. Similar behavior had been observed in other published studies in literature, where an increase in the ID decreased the porosity of 3D-printed IVRs (McCoy et al., 2019; Sierra-Vega et al., 2024). Fig. 5(b) also indicated that DAR and DR had no statistically significant effect on either open or total porosity (p-value > 0.05). In contrast, a previous study reported significant effects of DAR and DR on the porosity of acrylonitrile–butadiene–styrene cuboids printed using DDM (Eisele et al., 2023). That study examined broader ranges of DAR (1.1–1.7), and DR (25–155 %) while keeping ID constant at 100 %. This study used narrower ranges for DAR (1.2–1.3) and DR (56–62 %), and a wider range for ID (66–90 %), which may account for the differing results.

Porosity is a critical factor in the customization of IVRs, as porosity may influence both their mechanical properties and overall performance. Open porosity can facilitate the filtration of liquids within the core of the IVRs, which may affect their drug-release characteristics. In contrast, closed porosity, while less accessible, can impact the material’s structural integrity and contribute to weight reduction. Therefore, precise control of the porosity through the adjustment of these printing parameters, namely DAR, DR, and ID, is essential for personalizing the performance of IVRs printed in DDM to meet specific patient needs.

3.4. Shore M hardness, Young’s modulus, and compression strength

The mechanical properties of IVRs are related to their comfort, safety, and efficacy. IVRs should be flexible enough for easy insertion and adherence to therapy yet sufficiently rigid to prevent undesired expulsion (Johnson et al., 2010). In the current study, the investigated mechanical properties of IVR included Shore M hardness, Young’s modulus, and compression strength. Intentional changes in ID, DAR, and DR caused measurable variations in these mechanical properties. Within the printing runs, Shore M hardness values ranged from 36.87 (R13) to 79.5 (R5), while Young’s modulus and maximum compression strength measurements ranged from 1.53 (R13) to 7.93 N/mm² (R5) and from 6.66 (R6) and 52.67 N (R5), respectively. IVRs with the highest mechanical properties were obtained at higher ID and DR and lower DAR (R5).

The results in Fig. 6(a) show the main effects and significant interactions of DAR, ID and DR on Shore M hardness, Young’s modulus, and maximum compression strength of IVRs. All PLS regression models in CCD fit analysis exhibited R² > 0.98 and p-value < 0.05. The ID was the most significant printing parameter affecting these three mechanical properties. Increasing ID led to a significant increase in Shore M hardness, Young’s modulus, and maximum compression strength. This enhancement was attributed to the formation of denser, less porous structures with increased area of contact between adjacent droplet chains during deposition, as illustrated in Fig. 2. These conditions promoted a chain entanglement within the IVR matrix and facilitated the development of a more robust and densely packed 3D network structure, with improved stress distribution throughout the material, thereby enhanced the mechanical strength of IVRs and reduced their susceptibility to indentation and deformation under applied load. This observation aligns with the results shown in Fig. 6(b) and (c), where an increase in Shore M hardness was associated with lower porosity, and greater weight correlated with higher compressive strength, respectively. As porosity increased, the number of structural breakthroughs also increased, diminishing Shore M hardness and maximum strength.

Fig. 6.

(a) Plot of the effects of printing parameters and significant two-factor interactions on Shore M hardness, Young’s modulus, and compression strength; (b) Plot of Shore M hardness as a function of total porosity; (c) Plot of Maximum Compression strength as a function of the weight (d) Interaction plot for Young’s modulus as a function of ID and DAR; (e) Interaction plot for Compression strength as a function of ID and DR.

A significant positive correlation was observed between the DR and the mechanical properties of IVRs (p-value < 0.05). As DR increased from 56 to 62 %, the amount of polymeric material deposited per droplet also increased. This led to denser surface structures where individual droplet chains came into contact (Fig. 2), forming a more complex structure that improved the stress distribution in the IVR and consequently enhanced their mechanical properties. However, an increase in DAR resulted in reduced mechanical properties of IVRs, as shown in Fig. 6(a). A higher DAR increased the spacing between adjacent droplet chains, reducing the area of contact between droplets within the IVRs matrix. This structural change decreased resistance to indentation and lowered the force required for compression or stretching in IVRs of similar dimensions. The effect plots in Fig. 6(a) also highlight a significant interaction between ID and DAR for Young’s modulus and maximum compression strength (p-value < 0.05). As shown in Fig. 6(d), at lower DAR, ID had a high impact on Young’s modulus; nevertheless, this effect becomes less significant at higher DAR due to the increased spacing between droplet chains, which became the dominant factor influencing mechanical properties. Additionally, Fig. 6(a) shows that the interaction between ID and DR significantly impacted maximum compression strength. As shown in Fig. 6(e), the effect of ID became more significant at higher DR, driven by the significant increment in the area of contact between the droplets, which improved mechanical stability.

The mechanical properties results indicated that the ID, DAR, and DR significantly influence the Shore M hardness, Young’s modulus, and maximum compression strength of the IVRs, which are key properties determining their flexibility, elasticity, and strength. Shore M hardness plays a role in patient comfort during insertion. The compression strength describes the force required to deform the IVR under compression during insertion, while Young’s modulus provides insight into its stiffness and ability to recover elastically after deformation. IVRs with lower mechanical properties may be more prone to involuntary expulsion, while highly rigid IVRs may be difficult to insert, potentially causing tissue irritation or inflammation. Consequently, optimizing printing parameters to achieve IVRs with balanced flexibility and stiffness is essential for patient comfort. For comparison, Shore M hardness and maximum compression strength were measured for marketed IVRs − Femring, NuvaRing, and Estring − using the same methodology. The Shore M hardness values were 87.4 (1.2), 58.9 (1), and 53.8 (0.6) for NuvaRing, Femring and Estring, respectively, while their corresponding compression strengths were 3.5 (0.2), 4.2 (0.3), and 8.3 (0.5) N. These results were consistent with the mechanical properties of the 3D-printed IVRs, indicating that the 3D-printed IVRs can be tailored to have mechanical properties similar to the performance of the commercially available IVRs.

3.5. Storage modulus, loss modulus, and Tan δ

Viscoelastic properties of the materials provide valuable information for determining the patient acceptability of IVRs. The storage modulus represents the ability of the material to store and recover energy during the cyclic loading and unloading, while the loss modulus indicates the energy loss as heat or dissipated during one cycle of sinusoidal deformation. Tan δ is the ratio of the loss modulus to the storage modulus. Tan δ < 1 suggests that the material behaves like an elastic solid, while Tan δ > 1 indicates a more viscous liquid-like behavior. The results in Fig. 7 show the storage modulus, loss modulus and Tan δ of the IVR across all runs as a function of temperature. At 30 °C, the storage modulus ranged from 1.55 MPa (R13) to 9.10 MPa (R4). However, at 140 °C, this range narrowed significantly, from 0.078 MPa (R13) to 1.47 MPa (R5). The observed decrease in the storage modulus with increasing temperature reflected that the TPU matrix becomes softer and less rigid at higher temperatures. This may be due to the molecular chains of TPU undergoing microstructural changes, gaining increased mobility as the temperature rises, ultimately leading to a reduction in its elastic properties. The results presented in Fig. 7(b) revealed that the loss modulus of the IVR decreased as the temperature increased from 30 °C to 140 °C. Notably, this reduction was less pronounced than the decrease observed for the storage modulus. At 30 °C, the loss modulus values ranged from 0.53 MPa (R4) to 0.11 MPa (R6), while at 140 °C, the loss modules ranged from 0.32 MPa (R5) to 0.029 MPa (R13). This decline in the loss modulus may be due to the reduced viscosity of the material at higher temperatures. For all IVR runs, the loss modulus remained lower than the storage modulus, resulting in Tan δ < 1, as shown in Fig. 7(c). This indicated that the TPU matrix in the IVRs behaved like an elastic solid.

Fig. 7.

Plots of (a) Storage modulus, (b) Loss modulus, and (c) Tan δ as a function of temperature for 3D-printed IVR samples from all manufacturing runs.

The plots in Fig. 8(a) present the main effects and significant interactions of ID, DAR and DR on the storage and loss moduli of the IVRs at 30 °C. The PLS regression models yielded R² values exceeding 0.98 and p-values below 0.0001. ID was the most significant factor impacting both storage and loss moduli, while the DR showed no significant effect. An increase in ID led to higher values for both moduli. This increase was attributed to an enhanced area of contact between and within the layers. This relationship was confirmed by a stronger linear correlation observed between weight and storage modulus at 30° C (R² = 0.98), as shown in Fig. 8(b). A similar trend between weight and loss modulus was observed (Supplementary Fig. S4). Conversely, increasing DAR resulted in a decrease in both moduli. A higher DAR led to greater distance between droplets, which reduced the rigidity of IVRs. These findings align with the mechanical properties of IVRs, where a higher DAR also resulted in lower Shore M hardness and compression strength. Additionally, Fig. 8(a) highlights a significant negative interaction between DAR and ID, which decreases both storage and loss moduli. The contour plot in Fig. 8(c) shows that although ID increases the loss modulus, this effect was less pronounced when IVRs were printed at a higher DAR of 1.3, due to the larger distance between the droplets deposited at higher DARs, which decreased the area of contact between the droplet chains.

Fig. 8.

(a) Plot of the effects of printing parameters and significant two-factor interactions on the storage modulus and loss modulus; (b) Plot of storage modulus at 30 °C as a function of IVR weight; and (c) Contour plot showing the loss modulus as a function of ID and DAR.

As described previously, the deliberate changes in ID, DAR and DR affect both the storage and loss moduli of IVRs. These two properties are important for assuring the performance of IVRs, and patient compliance. The storage modulus is key to understanding the ability of the material to store energy, directly impacting the mechanical performance and comfort of IVRs during clinical use. A higher storage modulus could lead to difficulties during insertion and an increased risk of tissue damage and inflammation, while an insufficient storage modulus may result in inadequate retention, causing slipping or expulsion. On the other hand, the loss modulus provides information on the energy dissipation of the material, providing insight into the wearability and durability of the IVRs. The optimization of storage and loss moduli is essential for enhancing drug loading and release characteristics, ensuring that the IVRs deliver the appropriate dose consistently over time, thus supporting patient adherence and improving therapeutic outcomes (Morrow Guthrie et al., 2015).

3.6. Drug content

The changes in ID, DAR and DR also affected the total mass of β-estradiol incorporated into the IVRs. The average drug contents and their standard deviations (n = 3) for the IVRs are presented in Supplementary Fig. S5. The mean drug content values ranged from 1.08 (0.07) mg (R6) and 2.36 (0.07) mg (R5). However, when normalized to the weight of each IVR, the results showed a relatively consistent loading of approximately 0.1 % w/w for all experimental runs indicating no degradation nor process loss during manufacturing. The regression model to predict the drug content yielded an R² value of 0.99 and a p-value < 0.0001. The plots in Fig. 9(a) show the main effects of ID, DAR and DR on the drug content of the IVRs, with no significant interaction observed among these factors (p-value > 0.05). Increasing ID and DR resulted in higher drug content, whereas increasing DAR led to a decrease in drug content. These effects can be explained by the influence of these parameters on the overall weight of the IVRs, which in turn governs the total mass of drug incorporated. As illustrated in Fig. 9(b), a linear relationship was observed between the weight of the IVR and its corresponding drug content. This behavior supports our hypothesis that β-estradiol distribution within the polymeric matrix was uniform and reproducible under the chosen processing conditions.

Fig. 9.

(a) Plot of the effects of printing parameters on drug content (mg of β-estradiol), (b) Plot of drug content as a function of IVR weight.

3.7. In vitro drug release

The results in Fig. 10 show the cumulative release of β-estradiol from the IVRs under accelerated conditions of temperature and agitation in simulated vaginal fluid as the release media. Release profiles are shown for IVRs with low DAR (1.18 and 1.20) values (Fig. 10(a)), DAR = 1.25 (Fig. 10(b)), and high DAR (1.3 and 1.31) values (Fig. 10(c)). These cumulative percentages were determined at 7, 21, 28, 35, 42, and 49 days. Regardless DAR values, higher cumulative release percentages were printed using the lower ID values, while those with lower release percentages were printed with the high ID values. Specifically, IVRs from runs R6 (ID = 44.26, DAR = 1.25, DR = 59) and R13 (ID = 50, DAR = 1.3, DR = 56) exhibited higher release percentages of approximately 73.8 (2) % and 72.3 (1.5) %, respectively, by day 49. In contrast, IVRs from runs R4 (ID = 90, DAR = 1.2, DR = 56) and R5 (ID = 90, DAR = 1.2, DR = 62) showed lower release percentages of around 35.2 (3) % and 31.7 (4) %, respectively, within the same period. It may be noteworthy that these differences in the β-estradiol rates are attributed to the variation in the internal microstructure of IVRs produced by the employed printing parameters, even though all IVRs were manufactured using the same printing materials and formulation.

Fig. 10.

Release profiles of 3D-printed IVRs as a function of DAR and ID. Release profiles are shown for IVRs with (a) low DAR values, (b) medium DAR values, and (c) high DAR values. Red, green, and yellow symbols represent experimental conditions at 50 % ID, 70 % ID, and 90 % ID, respectively. (d) Plot of the effects of printing parameters on drug release at 21 and 49 days.

The release mechanism of β-estradiol from IVRs can be described by the following stages. Upon immersion in the release medium, the outer surface of the IVR is initially hydrated, dissolving surface-allocated drug molecules into the surrounding medium. Once hydrated, drug release is then primarily governed by the transport of the medium through the TPU matrix. Within this stage, the medium diffuses initially into the matrix and permeates through the interconnected open pore structure, which forms flow channels that facilitate further penetration. Over time, a liquid-induced erosion degrades the closed pores, allowing the medium to infiltrate into inaccessible regions within the IVR structure (Liu et al., 2023). As the drug dissolves within both open and closed pores, the drug gradually diffuses out of the matrix. The diffusion rate and medium permeation are influenced by the free volume available within the IVR, which is determined by the structural characteristics of these IVRs (Sharma et al., 2017).

Regression models were established based on the CCD to predict the cumulative percent of β-estradiol released from the IVRs at each time point. No significant relationship was found between the percentage of β-estradiol released at 7 days and the printing parameters (p-value > 0.05) due to the low variability observed in the β-estradiol release at this early stage. Since the initial stage of the release primarily occurs at the surface, all IVRs exhibited similar release characteristics at 7 days. The effect plots in Fig. 10(d) show the impact of ID, DAR, and DR on the percentage of β-estradiol released at 21 and 49 days. No significant interactions were observed (p-value > 0.05); however, the trends observed at 28, 35 and 42 days mirrored those seen at 21 days. The results in Fig. 10(d) demonstrated that the increase in ID consistently reduced the percentage of β-estradiol release across all time points. This effect can be attributed to the increase in the connectivity between adjacent droplet chains by increasing the ID, which decreases the porosity of IVRs and limits the diffusion rate and permeation of the release medium into the matrix, thereby slowing the dissolution and release of β-estradiol from the IVRs. The results in Fig. 11(a) and (b) confirm further that the correlations between the total porosity of IVRs and the amount of β-estradiol released at 21 (R² = 0.79) and 49 days (R² = 0.76), respectively, are evident. These correlations demonstrated an increase in the β-estradiol released with an increase in the total porosity of IVRs. It is noteworthy that pores in the IVRs are created during the printing process according to the geometric design and printing parameters; hence controlling the β-estradiol release from the IVRs.

Fig. 11.

Plot of drug release as a function of total porosity at (a) 21 days and (b) 49 days, and as a function of IVR weight at (c) 21 days and (d) 49 days.

An increase in DAR was found to enhance the percentage of β-estradiol released at 21, 28, 35 and 42 days. This increased drug release may be linked to the less densification and more porous structure of the IVRs as DAR increases, as shown in Fig. 2. A more porous structure promotes the diffusion rate and permeation of the release medium into the matrix, thereby increasing the dissolution and release of β-estradiol from the IVRs. Previous studies have shown that an increase in the density of polyurethane polymers reduced their water absorption rate (Wang et al., 2024). These results were confirmed by the inverse linear relationship observed between drug release at 21 and 49 days and the weight of IVRs (R² > 0.75), as shown in Fig. 11(c) and (d). At 49 days, DAR also increased the drug release; however, this effect was not statistically significant (p-value > 0.05). On the other hand, an increase in DR led to a decrease in the percentage of β-estradiol released for all the time points, as shown in Fig. 10(d), but these effects were also not statistically significant (p-value > 0.05).

Higuchi’s model was applied to characterize the release kinetic of IVRs fabricated under different printing conditions. This model assumes a porous matrix with solvent-accessible channels (i.e., open pores) through which drug release is primarily governed by Fickian diffusion (Higuchi, 1963). While some deviation from the model may be expected due to the presence of closed pores within the printed structures, μCT analysis revealed that the proportion of closed porosity was minimal—less than 2.5 % across all parameter combinations. This finding supports the assumption that open porosity is the dominant contributor to mass transport. The regression analysis shown in Fig. 12(a) illustrates Higuchi’s model fitting for representative manufacturing runs R4, R10, R13 and R14. The R² values for all 17 experimental runs were consistently above 0.97, indicating that the drug release follows a diffusion-controlled mechanism. The results in Fig. 12(b) show the release rates from all manufacturing runs, which come from the slopes of the regressions. Variations in the printing parameters were found to influence the release rate of 3D-printed IVRs, which ranged from 8.98 %/day^0.5 (R5) to 23.92 %/day^0.5 (R6). Highest release rate was observed in manufacturing run R6 (ID = 44.26, DAR = 1.25, DR = 59), while lowest release rate was noted in R5 (ID = 90, DAR = 1.2, DR = 62). It was found that lower ID values corresponded to higher release rates, whereas higher ID values resulted in lower release rates.

Fig. 12.

(a) Square root time plot of representative drug release profiles; (b) Higuchi release rates for IVRs from all manufacturing runs. The error bars represent the standard deviation of the slope values (n = 3 per IVR). (c) Contour plot showing the Higuchi release rate as a function of Shore M hardness and total porosity.

The CCD model demonstrated that ID was the only factor examined that impacted the release rate, with no significant influence from other factors or interaction effects under the conditions of this study. These results indicated an inverse relationship between ID and the release rate, consistent with findings reported in previous studies in DDM 3D-printed dosages forms (Ebrahimi et al., 2024; McDonagh et al., 2022b; Welsh et al., 2019; Zhang et al., 2021). As previously discussed, an increase in ID increases the area of contact during droplet deposition, creating more intricate structures with reduced porosity and improved mechanical properties. This denser matrix structure impedes medium permeation pathways, consequently retarding β-estradiol release kinetics. The contour plot in Fig. 12(c) shows the release rate as a function of total porosity and Shore M hardness. A higher total porosity corresponded to an increased release rate, whereas greater Shore M hardness resulted in a reduced release rate. These findings suggest that maximizing the mechanical properties of the IVR contributes to a slower release of β-estradiol.

The release rate is a key parameter for the personalization of dosage forms, enabling tailored drug delivery to meet individual patient needs. These results demonstrated that the β-estradiol release rate from 3D-printed IVRs manufactured using the DDM process may be customized by inducing deliberate changes to the printing parameters investigated in this study. This approach offers a flexible and precise method for manufacturing IVRs with customizable release profiles, providing advantages over traditional manufacturing methods, which may be less adaptable. Furthermore, the ability to fine-tune these printing parameters opens the door to more effective, patient-specific drug delivery systems that can improve both the efficacy and safety of treatments.

3.8. Personalization of IVRs

Multiple qualitative studies based on surveys have been reported in literature, evaluating the acceptability of IVRs among women across diverse demographic groups, including variations in ages, ethnicity, race, education, marital status, and prior experience with vaginal products (Karakatsani and Ayres, 2025; Morrow Guthrie et al., 2015; Rosen et al., 2015; Vargas et al., 2019; Zhao et al., 2022). These qualitative studies highlight that the acceptability of IVR therapy is influenced by multiple factors, including device dimensions, surface texture, ease of handling, insertion and removal without damage, dose accuracy, long-term use, and user discretion. Despite considerable variations in women’s preferences, currently marketed IVRs are designed and manufactured using a one-size-fits-all approach, limiting their adaptability to individual patient needs. The current study demonstrated the feasibility of using 3D printing technology for the manufacturing of personalized IVRs with different characteristics within a single manufacturing batch.

Previous research suggest that women generally perceive smaller IVRs as more comfortable and easier to insert, while larger IVRs are more noticeable during clinical use (Morrow Guthrie et al., 2015; Zhao et al., 2022). In 3D-printing, IVR dimensions are determined by the computer-aided design, enabling customization of IVR geometry to suit specific anatomical and comfort requirement. The surface texture also plays a role in user perception and device performance (Vargas et al., 2019); while some women perceive textured IVRs as rough, other associate their porous surface with enhanced drug delivery potential. The current study demonstrated that ID, DAR, and DR may be adjusted to customize IVRs with different internal structures, offering additional personalization options. Additionally, the user’s ability to handle, insert or remove an IVR without damage is influenced by the device’s mechanical properties (Morrow Guthrie et al., 2015; Rosen et al., 2015). Excessively rigid IVRs may be difficult to insert and uncomfortable during use, while overly flexible IVRs may lack retention, potentially leading to involuntary expulsion. The 3D-printed IVRs manufactured using DDM in this study exhibited a wide range of mechanical properties, including values comparable to those reported in the literature for commercially available IVRs (McCoy et al., 2019; Promadej-Lanier et al., 2009; Sierra-Vega et al., 2024; Welsh et al., 2019). The fine-tuning process of ID, DAR and DR enhances the capacity to produce IVRs with precisely tailored mechanical properties to meet diverse and specific healthcare needs of individual women.

Dose accuracy and long-term therapeutic delivery are both patient-centered preferences and essential clinical requirements to ensure safety and efficacy of a drug delivery system. Fixed-dose IVRs may not be suitable for all populations and may increase the risk of adverse effects in women of different age, varying racial background, or with coexisting conditions. For example, fixed-dose estrogen in menopause treatment was associated with an increased risk of venous thromboembolism and stroke (Fu et al., 2018). The current study demonstrated that ID, DAR and DR played important roles in controlling drug loading or release kinetic of IVRs, offering a route toward individualized dosing.

Fig. 13 illustrates a theoretical framework for the smart personalization of IVRs, integrating patient-specific data with advanced manufacturing technologies under regulatory oversight. The process may begin with patient assessment and digital health data collection, capturing individualized physiological and clinical parameters. This information informs a personalized e-prescription and decision support system, which generates device specifications tailored to the patient’s needs. These parameters are then used in 3D CAD and process simulation to model the device geometry and optimize drug release profiles. The process may then begin with patient assessment and clinical data collection, followed by the generation of a personalized e-prescription. Based on these inputs, the IVR is digitally designed and simulated for tailoring the IVR geometry and drug release characteristics to the specified therapeutic needs. The device is then fabricated using automated 3D printing with in-process controls over the ID, DAR, and DR. Finally, the personalized IVR is dispensed and administered under clinical oversight, completing the closed-loop system. This approach is expected to align and overseen with regulatory compliance pathways, underscores the potential for data-driven, scalable, patient-specific IVR therapy.

Fig. 13.

Smart personalization of IVRs.

Taken together, the findings from the current study highlights the potential of DDM-based 3D printing technology as a flexible and scalable platform for the manufacturing of patient-specific drug delivery systems. The total time required to print 18 IVRs was about 5 h (~16 min per IVR), depending on selected printing parameters. The throughput rate may be further increased by scaling the DDM process through the addition of print heads or the use of parallel printing units (Sierra-Vega et al., 2025). Notably, only a limited number of IVRs may be required per patient to cover a year of therapy, suggesting that large-scale manufacturing may not be necessary in distributed manufacturing environments for these personalized IVRs.

4. Conclusion

This study demonstrated the feasibility of using DDM to manufacture IVRs with varied internal structures, mechanical characteristics, drug loading, and release kinetics within a single production cycle. The adaptability of DDM technology underscores its potential for creating tailored drug delivery systems, thereby enhancing patient-specific therapeutic interventions. This methodology presents a paradigm shift from conventional drug delivery approaches towards more individualized solutions.

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijpharm.2025.126382.

Data availability

Data will be made available on request.