An Intelligent and Efficient Workflow for Path-Oriented 3D Bioprinting of Tubular Scaffolds

Abstract

Modern 3D printing is a valuable tool for tissue engineering (TE), and the fabrication of complex geometries such as tubular scaffolds with adaptable structure, for example, as replacements for intestines, bronchi, esophagus, or vessels, could contribute to standardized procedures in the future of regenerative medicine. However, high-precision bioprinting of scaffolds for tubular TE applications remain a major challenge and is an arduous endeavor with currently available three-axis bioprinters, which are limited to planar, layer-by-layer printing processes. In this work, a novel, straightforward workflow for creating toolpaths and command sets for tubular scaffolds is presented. By combining a custom software application with commercial 3D design software, a comparatively large degree of design freedom was achieved while ensuring ease of use and extensibility for future research needs. As a hardware platform, two commercial 3D bioprinters were retrofitted with a rotary axis to accommodate cylindrical mandrels as print beds, overcoming the limitations of planar print beds. The printing process using the new method was evaluated in terms of the mechanical, actuation, and synchronization characteristics of the linear and rotating axes, as well as the stability of the printing process. In this context, it became clear that extrusion-based printing processes are very sensitive to positioning errors when used with small nozzles. Despite these technical difficulties, the new process can produce single-layer, multilayer, and multimaterial structures with a wide range of pore geometries. In addition, extrusion-based printing processes can be combined with melt electrowriting to produce durable scaffolds with features in the micrometer to millimeter range. Overall, the suitability of this setup for a wide range of TE applications has thus been demonstrated.

Introduction

A variety of tubular structures in the human body, such as trachea, intestine, or blood vessels, can suffer from disease or trauma-induced damage ultimately resulting in the need for appropriate segmental replacement. Among those, most frequently, tubular scaffolds for vascular structures are needed, and here conventional therapies often use autologous veins that come with limited availability, or are based-upon synthetic or xenogeneic/allogeneic vascular grafts with limited durability as well as adverse side effects such as risk of infection or the need for ongoing medication. For example, patients affected by atherosclerotic lesions or coronary heart disease are in need for peripheral bypasses (diameter 0.8 cm and above) or coronary bypasses (diameter ∼0.5–0.6 cm), and here 3D printed tubular scaffolds would present an attractive vascular substitute.¹

Generally, tissue engineering (TE) aims to provide improved treatments by developing biomimetic grafts that ideally integrate seamlessly into the patient's body. These grafts must meet several application specific requirements, including biocompatibility, biodegradability, mechanical stability as well as porous architectures with features ranging from the micrometer to the millimeter range.^2–4

This makes 3D printing a popular manufacturing technique in TE because of its ability to provide virtually unlimited design freedom and a precise and repeatable production process.⁵ However, printing of tubular structures poses a major problem because conventional 3D printers use a planar layer-by-layer printing approach. This leads to the necessity to print hollow and often overhanging structures that suffer deformation or fail to print completely due to insufficient support of the deposited material. A promising and versatile approach to solve these problems is the replacement of the planar print bed by a cylindrical or physiologically shaped mandrel.

When using a mandrel, the material deposition is often performed using melt electrowriting (MEW), melt or solution electrospinning (ES), or direct extrusion similar to fused deposition modeling. Gao et al developed a method to create tubular structures by spiraling extruded hydrogel fibers on a rotating mandrel demonstrating the superior print fidelity, especially important for soft materials, compared with a planar print process.^6,7 Similarly, the winding of nano- and microfibers around a mandrel, based on MEW or ES processes, proved to be a versatile manufacturing approach for tubular scaffolds that are well suited for cell cultivation and proliferation.^8–12 To improve the mechanical characteristics of MEW- or ES-based scaffolds, additional extrusion-based layers, for example made from biocompatible thermoplastics, can be used.^13–15

To this end, a major problem is the lack of suitable software that supports the path planning and command set generation for these kinds of nonplanar print processes. Often limiting the material deposition to simple helical paths and the resulting pores to either rhomboid or rectangular geometries. More sophisticated control solutions that allow greater design freedom for extrusion-based print processes have been developed, but are often based on custom programming for a limited number of specific pore geometries.^16,17 Coulter et al^18,19 presented an advanced, and to some extent, automated approach for extrusion of silicone on complexly shaped mandrels, creating sophisticated structures that can mimic the mechanical and hemodynamic properties of heart valves. By utilizing parametric 3D design software not only to design the structures but also to directly generate the needed command sets, he unlocked great design freedom and simplified command set generation.

In line with these efforts, the goal of this work was to develop a user-friendly and fast workflow for a versatile fabrication of tubular scaffolds by 3D printing without the need for highly complex software for nonplanar toolpath generation or digital geometry design. A custom software application allows the user to create multilayered scaffolds based on simple path geometries that can be generated in a commercial 3D design software. The software application further allows the setting of common printing parameters and ultimately generates the needed command set. This way the user only needs limited understanding of the technical implementation, yet has great design freedom for fast production and testing of scaffolds.

This approach is intended to demonstrate the feasibility of retrofitting commercial bioprinters with rotary axis and the difficulties involved. The versatility of the new workflow is demonstrated in terms of design freedom and feasibility of producing tubular scaffolds from multiple materials and in multiple processes. The process is evaluated in terms of quality and limitations.

Materials and Methods

Rotary axis hardware

Two commercial bioprinters, a Cellink BIO-X and a Regenhu R-GEN 200, were equipped with a rotary axis that allows the rotation and positioning of a mandrel with variable diameter. In this work, a stainless steel mandrel with an outer diameter of 8 mm was used, which could serve as a peripheral arterial bypass for the leg in a patient with peripheral occlusive disease. The rotary axis for the BIO-X bioprinter makes use of a stepper motor with a small full-step angle of 0.36° (Orientalmotor PKP544MN18A) and an additional belt drive transmission of 2.125 (Mädler HTD 3M), resulting in a high positional resolution of 0.169° or 11.83 μm on the surface of the mandrel (orthogonal to its longitudinal axis). A stiff aluminum construction and multiple adjustment screws ensured good alignment with the bioprinter's axes. A pneumatic system compensates most of the additional weight and prevents mechanical overloading of the print bed/Z-axis of the bioprinter.

The design of the rotary axis for the R-GEN 200 bioprinter was focused on allowing a constant rotation of the mandrel for MEW. Therefore, a simplified mechanical design was chosen and a stepper motor with a full-step angle of 1.8° (ACT Motor 2801K) directly drives the mandrel. Two plastic bushings made of polyoxymethylene hold the mandrel in place and ensure electrical isolation. A spring-loaded sliding contact is used to establish an electrical connection between the mandrel and the arc and short circuit-protected high-voltage power supply. Additional protective grounding ensures safe operation. Three-dimensionally printed brackets carry the stepper motor, bearings, and the sliding contact and can be mounted on the print bed flange of the bioprinter. Figure 1 shows the retrofitted bioprinters. The stepper motors were controlled with an Arduino Due and Arduino Nano, respectively.

FIG. 1.

Design and implementation of the rotary axis in (a) the BIO-X and (b) the R-GEN 200 bioprinter.

Path oriented scaffold geometry design workflow and software

The design of tubular scaffolds with porous and lattice-like structures was enabled by a path-oriented workflow. By arranging at least one continuous, non-self-intersecting path geometry around the mandrel, the structure of a given layer can be designed, as visualized in Figure 2a. The overall scaffold geometry then follows from the arrangement of one or more layers.

FIG. 2.

(a) Visualization of the path-oriented workflow. (b) Simplified flowchart of the command set generation and the subsequent print process.

To create a fast and intuitively usable workflow, an automated command set generation is required. For this purpose, an application featuring a graphical user interface has been written in python. Except for a straight path that runs parallel to the mandrel's longitudinal axis, the path geometries have to be imported as planar X-/Y- coordinate datasets. For each path geometry, several geometric parameters and printing parameters can be set. The printing parameters include the extrusion pressure, the nozzle velocity relative to the mandrel's surface, and the desired print head. The geometric parameters include the layer height relative to the mandrel's surface, the nozzle height relative to the layer height, the desired number of equally spaced path repetitions around the mandrel, and the rotational start position of the first path repetition relative to the mandrel.

For two neighboring points of the coordinate dataset $\left(x_i,y_i;x_{i-1},y_{i-1}\right)$, the required velocity and travel distance is calculated for the X-axis of the bioprinter $\left(v_{x,i},\Delta x_i\right)$ and for the rotary axis$\left(v_{y,i},\Delta y_i\right)$ to perform a straight travel movement with the desired nozzle velocity $\left(v_E\right)$.

$$\Delta x_i=x_i-x_{i-1}$$ (1)

$$\Delta y_i=y_i-y_{i-1}$$ (2)

$$v_{x,i}=\frac{\Delta x_i\cdot v_E}{\sqrt{\Delta x_i^2+\Delta y_i^2}}$$ (3)

$$v_{y,i}=\frac{\Delta y_i\cdot v_E}{\sqrt{\Delta x_i^2+\Delta y_i^2}}$$ (4)

During the calculation, the travel distance perpendicular to the mandrel's longitudinal axis is transformed into an equivalent rotation of the mandrel $\left(\Delta {\phi }_i\right)$ with respect to the current distance of the nozzle from the center of the rotary axis $\left(d_{nozzle,center}\right)$.

$$\Delta {\phi }_i=\frac{\Delta y_i\cdot 360}{2\cdot d_{nozzle,center}\cdot \pi }$$ (5)

The velocity and distance/angle information are then translated into G-Code commands for the bioprinter and into a set of step and time information for the control of the rotary axis stepper motor, respectively. This procedure is demonstrated here for the BIO-X bioprinter but adaptations in the used G-Code commands can be made for the use with other bioprinters.

A stable extrusion and thus printing process is ensured by the addition of a stabilizing path section that runs parallel to the mandrel's longitudinal axis and that extends each path geometry on both ends, as seen in Figure 2a. The length of this section was set to 20 mm for all print processes in this work. After each movement of the print head along a path and the associated straight sections, the mandrel is turned to the rotational start position of the next path and printing continues. This process is repeated until the complete scaffold is manufactured and finally the print head is moved to a safe parking position. Figure 2b shows the principle of the command set generation and printing process.

Path generation

The Rhino 3D software was used for the design of the path geometries. To export the path geometry as a planar X-/Y- coordinate dataset and to simplify the parametric design of some basic path geometries (sinusoidal, rectangular etc.), the integrated Grasshopper programming environment was utilized. A fixed X-axis resolution can be specified and is maintained along the path geometry, thus defining the number of the coordinate data points. The software application described above translates the coordinates (data points) directly into a command set for the printer. Therefore, it is important to select an appropriate resolution to ensure smooth and uninterrupted execution of the command set. The resolution must be found experimentally and is limited by the processing speed of the bioprinter control. In this work, a resolution of 100 μm was used.

Axes synchronization

Due to the proprietary control systems of both bioprinters used in this work, a direct integration or synchronization of the printers' linear axes with the additional rotary axis was not possible. Therefore, an alternative synchronization approach was developed for the BIO-X bioprinter. Each time the print head reaches the start position for the next path and just before it begins to travel along it, the LED light of the printer is briefly switched on and off by a G-Code command. By monitoring the LED power supply with an opto-isolator (Vishay SFH610A-1) and Schmitt trigger circuit (STMicroelectronics LM393N), a synchronization impulse is created. This synchronization impulse is sensed by the Arduino Due and allows a synchronized start of the linear and rotary axes for each path along the mandrel. Since we use only a constant rotation of the rotary axis for all MEW experiments, no synchronization was implemented for the R-GEN 200 bioprinter.

Printing parameters and materials

Polycaprolactone (PCL) with an average molecular weight of 45.000 g/mol (Sigma-Aldrich; lot no. MKCD3536) was chosen as an exemplary thermoplastic material due to its broad application in bioprinting.^20,21 For the BIO-X bioprinter a high temperature pneumatic print head was used with a 0.25 mm nozzle to extrude the molten PCL onto the mandrel. The temperature of the print head was set to 100°C, the extrusion pressure to 300 kPa, the nozzle velocity to 8 mm/s, and the height of the nozzle above the mandrel or preceding layers to 0.35 mm. The parameters were experimentally determined to ensure good fusion of layers as well as mostly round cross-sections of the strands. For the demonstration of multimaterial print processes skin cream (Nivea; item no. 80105) mixed with red food dye was used as a hydrogel replacement material. This material was printed at room temperature using a pneumatic print head with a 1/4″25 G blunt needle. The extrusion pressure was set to 350 kPa, the nozzle velocity to 4.167 mm/s, and the height of the nozzle above the mandrel or preceding layers to 0.5 mm.

For the MEW processes on the R-GEN 200 bioprinter, the molten PCL was extruded through a 0.28 mm nozzle at a temperature of 100°C and with a pressure of 23 kPa. The distance between the nozzle and the mandrel was set to 3 mm and a voltage of 4 kV was applied to draw the fiber. The relative velocity of the nozzle to the mandrel surface was constant and set to 45 mm/s with a winding angle of 22.5, 45, or 67.5° relative to the mandrel's longitudinal axis. The rotational direction was kept constant during the printing process, whereas the nozzle repeatedly moved along the mandrel over a distance of 120 mm.

The strand and fiber width were measured perpendicular to the mandrel surface with an Olympus IX83 Microscope and its cellSens Dimension 2.3 software package.

Results and Discussion

Mechanical evaluation of the rotary axes

After the initial installation and alignment of the rotary axes, the mechanical and positioning characteristics were evaluated. Because mechanical deviations as well as the overall axes positioning performance have an impact on the overall positioning performance of the system, it is important to optimize both. The repeatability of the BIO-X axes and of the associated rotary axis was evaluated by repeatedly moving to specific points and measuring the relative position with a dial gauge (Käfer M2 Top) (Fig. 3a, c).

FIG. 3.

(a) Travel movements for the measurement of the repeatability of the BIO-X axes. Black dots indicate programmed points. (b) Measurement of the radial run-out (R), axial play (A), and tilting of the mandrel relative to the X-axis of the bioprinter (T). (c) Indirect measurement of the repeatability $\left(\phi \right)$ of the rotary axis by measuring the linear displacement $\left(\Delta a\right)$ orthogonal to the axis at a given distance. Red circles and arrows indicate measurement positions.

The results as well as the theoretical positioning resolutions are listed in Table 1. The results indicate that the drive system of the rotary axis is well suited for our application and that it is providing an appropriate positioning performance. These measurements were not performed for the rotary axis of the R-GEN 200 bioprinter since it is only used for a constant rotation.

Table 1.

Positioning Performance of the Different Axes of the BIO-X Bioprinter

Axis	Theoretical resolution (μm)/(°)	Repeatability (μm)/(°)
X	1–10a/—	5.2/—
Y	1–10a/—	9.73/—
Z	1–10a/—	3.35/—
Rotary	11.83b/0.169	0.33b/0.005

^a Smallest allowed step width of movement depending on firmware version, no exact specifications available.

^b Theoretical resolution and repeatability values in micrometers for a mandrel with an outer diameter of 8 mm.

For both bioprinters the radial runout at both ends of the mandrel as well as the tilting relative to the bioprinters' X-axis was measured (Fig. 3b). The measurements were taken three times with the dial gauge and with a thickness gauge after remounting and, in case of the BIO-X bioprinter, recentering of the mandrel. For the BIO-X bioprinter, it was found that the radial run-out on both sides is about 50–100 μm, with less variability on the left side due to the stiffer bearing. The tilting of the mandrel, expressed as the distance variability between the print head and the mandrel along its longitudinal axis, is on average about 12.5 μm. For the R-GEN 200 bioprinter, the run-out is about 200–300 μm, and the tilting results in a distance variability of about 50–100 μm. As expected, the much more elaborate mechanical design for the BIO-X rotary axis results in a better mechanical performance. Lastly, the axial play of the mandrel was measured with the dial gauge to be within 5 μm for the rotary axis of the BIO-X bioprinter, indicating a good axial fixation. Again, for the R-GEN 200 bioprinter, this measurement was not taken.

Stability of the extrusion on the retrofitted BIO-X bioprinter

The stability of the PCL extrusion on the mandrel was evaluated by the measurement and statistical evaluation of the strand width, which is a good indicator for the overall strand and extrusion process quality. A higher strand width indicates increased flattening, whereas a smaller strand width indicates thinning and/or lift off or bad fusion to the underlaying surface or layer. The strand width measurements were taken on the left and right side of the mandrel shortly before the path transitions into the stabilizing path sections. Subsequently, the mean value and the standard deviation (σ) was calculated. Two path geometries were investigated. First, a path running parallel to the mandrel's longitudinal axis and second a path running at a 45° angle to the said axis. In each print process, 10 paths were arranged with an equal spacing around the mandrel and the print process was repeated three times for each path geometry.

The results are shown in Figure 4; it was found that the strand width was comparable between the two path geometries and thus the printing parameters did not have to be adapted to the chosen path geometry. Further effects of the earlier discussed radial run-out and of the distance variability between the print head and the mandrel are visible in the measurements. The lower variability in the radial run-out on the left side translates to a reduced standard deviation of the strand width compared with the right side (Δσ_straight = 3.34 μm, Δσ_angled = 8.59 μm). On the other hand, the slightly smaller distance between the nozzle and the mandrel surface on the right side, due to the slight tilting of the mandrel, causes a slight flattening and thus, on average, bigger strand width on the right side.

FIG. 4.

Analysis of extruded strand width. (a) For a straight path along the mandrel's longitudinal axis. (b) For a straight path running at a 45° angle to the said axis. Boxplot is supplemented with mean value (dotted orange) and standard deviation (dotted purple). (c) Cross-section of a strand printed in a straight path along the mandrel's longitudinal axis. The contact area between strand and mandrel is visible as a flat surface at the bottom.

Stability of the MEW process on the retrofitted R-GEN 200 bioprinter

By depositing fibers during 20 transitions of the nozzle along the mandrel and subsequently measuring the fiber width and the fiber angle relative to the mandrel's longitudinal axis at 10 random locations, the fiber properties were evaluated. This was repeated for three printing processes and the combined mean fiber width, winding angle, and their standard deviations were calculated (Table 2).

Table 2.

Mean Values and Standard Deviations of the Measured Melt Electrowriting Fiber Orientation and Width for Three Different Winding Angles

Winding angle target (°)	Winding angle mean (°)	Winding angle standard deviation (°)	Fiber width mean (μm)	Fiber width standard deviation (μm)
22.5	21.71	0.7	11.63	1.42
45	42.72	1.12	12.55	1.85
67.5	64.3	3.59	12.53	1.49

Due to the observed, mostly round fiber profile, the fiber diameter is about the same as the measured fiber width. By visual inspection, the winding angle showed only slight variations, as could be detected under the difficult measuring conditions. Overall, the fiber deposition can be described as stable and reproducible with good tolerance toward changing nozzle-to-mandrel distances caused by the simple mechanical design. By adjusting the winding angle, the overall pore geometry can be chosen. However, due to the unsynchronized axes, the pore sizes show great variability and precise fiber deposition/stacking is not possible. Furthermore, it was observed that with increasing transitions of the print head, and thus a denser fiber deposition, the occurrence of uncontrollable fiber stacking and of interrupted fibers also increased. Based on the work done by Ding et al²² and by Wunner et al,²³ we assume that these effects are probably caused by residual charges of the fibers.

Synchronization errors between the linear and rotary axes

Since the BIO-X bioprinter axes and the rotary axis movements are only synchronized once at the beginning of each path, it is possible that the synchronization progressively deteriorates along its printing path. These synchronization errors (SE) were evaluated qualitatively by visual comparison of the ideal path with the actual path described by the print head. This was done by attaching printouts of the ideal path to the mandrel, which were then traced by a pencil that was attached to the print head, eliminating the variability of an extrusion process. The influence of the path geometry was investigated by using sinusoidal and rectangular path geometries with similar sizes and periodicities.

As shown in Figure 5, the larger path geometries were followed well while only the sinusoidal geometry was reproduced with only minimal deviations on a smaller scale. The smaller rectangular geometry shows that alongside the process a progressively growing mismatch between the axes movements appears, which leads to a significant deviation between the ideal and the real path. Thus, it was concluded that internal processing of the command set by the bioprinter causes the SE. During this processing for example, acceleration limits are enforced, which then lead to slower-than-expected movements of the bioprinter axes. Since these limits are not known nor changeable, it was not possible to take them into account in the software application that generates the command set. As a result, path geometries with many sharp edges and/or short sections between them could not be reproduced properly.

FIG. 5.

Comparison between the ideal path (black) and the path described by the print head (pink). For rectangular and sinusoidal path geometries with (a) an amplitude of 2.25 mm and period of 9 mm and (b) with an amplitude of 0.5 mm and period of 2 mm. Blue arrows indicate the travel direction of the print head.

Tubular scaffolds with great topological variability

Using the path-oriented workflow, it was possible to demonstrate the fast and reproducible fabrication of tubular scaffolds with a comparatively large degree of freedom regarding the pore geometry design (Fig. 6a–d). Due to the earlier discussed SE, scaffolds with small pores are currently limited to simple paths and thus simple pore geometries. Furthermore, as a result of SE, significant variability in the pore sizes and geometries (Fig. 6f), as well as scaffold integrity was observed due to shifting of strands. This shifting proved to be especially problematic in areas where strands of neighboring paths were supposed to fuse together (Fig. 6g, h). In these locations, a shift of the strands can lead to decreased fusing or even gaps in the scaffold.

FIG. 6.

(a–d) Examples for different pore geometries that can be produced by adapting the path geometry. (e) Necking of the strand leading to inconsistent strand diameters. Red arrow indicates the travel direction of the print head. (f) Pore geometries and sizes for a two-layered scaffold. Each layer consists of 35 strands following a path running at a ±45° angle to the mandrel's longitudinal axis. Theoretical expected pore size is 326 by 346 μm. (g) Low SE leading to good fusing between adjacent strands. (h) High SE leading to displacement of strands and gaps in the scaffold. SE, synchronization errors.

Another problem resulting from the fusion of adjacent strands was that the later extruded strand showed necking when leaving the fusion area (Fig. 6e). We suspect that a change in the forces acting on the extruded but not yet fully solidified material is the cause. Similar defects were observed by van Kampen et al in a comparable printing process.¹⁶ It should be possible to alleviate this effect with adapted and spatially resolved printing parameters; this is the object of further ongoing work.

Multimaterial and combined MEW and extrusion-based tubular scaffolds

Since many tubular tissues, such as blood vessel walls, comprise multiple layers with different biological and mechanical properties as well as distinct cell compositions, we demonstrated multimaterial and multiprocess manufacturing of tubular scaffolds that could help to recreate and mimic such structures in the future.

First, a multilayered scaffold comprising PCL and a hydrogel replacement material was printed as a mock-up vascular graft structure. The scaffold consists of seven layers forming different pore geometries (Fig. 7).

FIG. 7.

(a) Multimaterial mock-up vascular graft scaffold. (b) Dissected mock-up scaffold showing the different layers (L1–L7). Layers L2–L4 are difficult to differentiate due to their similarity and strong bonding.

Table 3 lists the printed layers, their properties, and their contribution to the overall structure. The orientation of the pore geometries and strands was inspired by the prevailing cell orientation in the Tunica intima, Tunica media, and Tunica externa of a vascular wall, respectively.²⁴ This design freedom might be used in future cell cultivation experiments to positively influence cell alignment.²⁵ The ability to deposit hydrogel-like materials into specific layers might be used to include cell-laden materials into specific areas of the scaffold. Overall, this scaffold demonstrates how the path-oriented workflow could be used to fulfill tissue- and cell-specific requirements.

Table 3.

Properties, Composition, and Functional Contribution of the Different Layers of the Mock-Up Vascular Graft Scaffold

Layer	Number of strands	Path geometry and its orientation relative to the mandrel's longitudinal axis	Material	Contribution to overall structure
1	46	Parallel	PCL	Tunica intima
2	15	Angled, 67.5° ↑	PCL	Tunica intima, Tunica media
3	12 + 4	Angled, 67.5° ↑	Hydrogel replacement material + PCL	Tunica media, demonstration of hydrogel incorporation, PCL for intralayer bonding
4	15	Angled, 67.5° ↑	PCL	Tunica media, Tunica externa
5	46	Parallel	PCL	Tunica externa
6	10	Angled, 45° ↑	PCL	Structural support
7	10	Angled, 45° ↓	PCL

PCL, polycaprolactone.

Second, to demonstrate the integration of very small fibers and pore structures, a scaffold comprising of an inner functional MEW layer and an outer structural extrusion-based grid was printed (Fig. 8). In a first step, the inner layer was printed on the R-GEN 200 bioprinter with a winding angle of 67.5° and 1000 repetitions of the print head along the mandrel. In a second step, the mandrel was transferred to the BIO-X bioprinter for printing of the outer structural grid. This extrusion-based grid comprises of two printed layers with 35 strands each and an inverse 45° orientation of the strands relative to the longitudinal axis of the mandrel. The MEW and extrusion-based structures were generally well connected. The fine MEW structure was not destroyed or significantly altered; however, the fine pores were filled out by the extruded strands. Releasing the scaffolds from the mandrel was generally problematic due to the strong adherence of the MEW fibers to the mandrel's surface. The outer structural grid was vital to prevent destruction of the MEW layer, and by cooling the mandrel for 10 min in an ice bath the adherence could be reduced to a level that no damage, but only small fiber displacements, occurred.

FIG. 8.

(a) Combined MEW and extrusion-based scaffold, cut open for better visualization. (b) Microscopic image showing the integrity and bonding of the MEW and extrusion-based layers. MEW, melt electrowriting.

Conclusion

In summary, a new path-oriented 3D (bio-)printing workflow was developed to allow a fast and user-friendly creation of many different tubular scaffold designs and their fabrication on a rotating mandrel. By avoiding the traditional 3D printing workflows and software solutions, including complex 3D models and slicing of those, a more intuitive, versatile, and with intermediate coding knowledge implementable approach, was made available.

The workflow is based on the arrangement of two-dimensional, non-self-intersecting but otherwise freely designable path geometries on and around a mandrel in multiple layers. With this approach, it was possible to demonstrate single-layered, multilayered, and multimaterial extrusion-based print processes. Furthermore, it was demonstrated that defined extrusion-based structures can be printed on top of MEW structures, forming well-connected structures with pores and strands/fibers ranging from the micrometer to the millimeter range.

Defects and deviations in the produced scaffolds were mainly a result of SE between the linear axes and the retrofitted rotary axis or of the mechanical inaccuracies of the setup. To reduce these errors, a more sophisticated integration of the additional axis or, ideally, a fully custom printing platform will be needed in the future. This includes more precise mechanical implementations, fully integrated and synchronous control systems for all axes, as well as error compensation by adapted print bed-leveling techniques.

With these improvements, we expect to achieve better accuracy and reproducibility, which will enable the reduction of extrusion-based pore geometry sizes even for complex pore shapes and smaller mandrel diameters. Additionally, this will be a first step toward unlocking precise fiber deposition and stacking during MEW. However, this will probably also require more elaborate path planning, similar to the work done by McColl et al²⁶ and Hrynevich et al,²⁷ to account for the complex MEW process characteristics. Future path planning should also investigate the implementation of spatially resolved printing parameters that might be beneficial to improve the fusion of neighboring strands in extrusion-based print processes as well as the process stability and, therefore, scaffold integrity in general.

Finally, evaluation of the print quality, spatial accuracy, and scaffold integrity might be improved by deployment of, for example, computer vision-enabled measurements that allow the comparison of the printed and designed morphologies by using appropriate fidelity metrics, similar to work done by Armstrong et al²⁸ This will not only improve and accelerate the further development of printing techniques for tubular scaffolds but also serve as a valuable tool for the quality control of the produced structures.

Beyond the mock-up vascular graft demonstrated in this article, we propose that the versatility of our workflow makes it well suited for a wide range of tubular scaffolds and applications, therefore, adding a valuable pathway for future 3D (bio-)printing-based development and production of tubular structures.

Authors' Contributions

T.B.: writing original draft, software, visualization, methodology, investigation, and formal analysis. S.L.: supervision, validation, and writing—review and editing. H.H.: supervision, validation, and writing—review and editing. F.C.: supervision. H.B.: funding acquisition, project administration, and writing—review and editing. C.B.: funding acquisition, project administration, and writing—review and editing.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work has been carried out within the framework of the SMART BIOTECS alliance between the Technical University of Braunschweig and the Leibniz University of Hannover. This initiative is financially supported by the Ministry of Science and Culture (MWK) of Lower Saxony, Germany.