3.2. Micro Injection Molding of Implant Prototypes
Figure 7 shows the finished and assembled 3D printed halves of the mold. The parts fit closely. There is minimal clearance between the upper and lower half before they are used for the µIM injection molding process (see
Figure 7 A). The edges of the parts are flush with each other. After the µIM process (single use,
Figure 7 B), the clearance between the upper and the lower half is decreased. The upper half of the mold shows a crack as a sign of wear.
Figure 8 shows a separated mold after a finished µIM process. The mixture of the silicone elastomer (MED-4244) and the glucocorticoid DEX (10 wt%) was successfully injected and cured. Despite the decreased clearance between the parts (compare
Figure 7 A and B), the two halves were not significantly merged and the separation of the halves was easy. However, there are significant signs of wear all over the surface of both parts of the mold. Nevertheless, the separation of the GP-RNI from the mold worked well and did not cause any damage to the µIM implant.
Low thermal conductivity, low heat resistance and, consequently, low durability and high cycle times are known limitations of molds manufactured from available photopolymeric resins [
29,
31]. In this study, the combination of thermal and mechanical stress during the µIM process is most likely the reason for the wear of the DLP 3D printed molds, as these are significant factors [
29]. The glass transition temperature (T
G) of the photopolymer PlasGRAY V2, which was used for DLP 3D printing of the molds, is T
G = 84 °C [
32]. The mold temperature during the performed µIM process was T
mold = 160 °C. When the mold temperature is similar to T
G the polymer material may become brittle and prone to failure [
29]. Zink et al. recommend keeping the mold temperature below the glass transition temperature of the mold material [
36]. Following Zink et al., the parameter of the mold temperature significantly affects the applicability of 3D printed polymeric molds. The mold material's mechanical properties are impaired as a function of temperature, especially when using temperatures above T
G. In another study by Martinho et al, the authors recommend keeping the mold temperature below 15 °C above T
G of the 3D printed photopolymer resin [
37]. In our study, we used a relatively high T
mold = 160 °C for a mold material with a T
G = 84 °C. The manufactured GP-RNIs were of proper quality. However, the molds showed significant wear because of thermal stress. Because of that, we used a single mold only once for our µIM process.
Due to economical manufacturing criteria, 3D-printed polymer molds can be well suited for low and medium-volume injection molding production [
29]. Commonly, for a medical implant, it is desirable to meet a patient's individual needs. Implants such as the described RNI benefit from high customization. Low durability of the molds and relatively high cycle times of an injection molding process might be acceptable. As a result, the described rapid tooling process chain based on 3D printed molds and µIM has proven to be a promising manufacturing technique for the production of highly individualized single parts. Furthermore, there are promising strategies to overcome the limitations of heat transport and heat resistance, such as composite materials [
29] and innovative cooling channels in the 3D printed molds [
36,
38]. Even relatively simple methods such as tight fitting metal mold holders can help to deal with low heat resistance of 3D printed molds, as we showed in [
39].
Figure 9 A shows the top view and
Figure 9 B shows the bottom view of two GP-RNI. The implants are homogeneously colored and show no failures such as burns, black spots, short shots or deformations. There are a few flash failures around the contour of the implant, where the mold halves met (
Figure 9 C). Moreover, the implant’s surface show stair casing (
Figure 9 C) because the built resolution of the implant is limited to the DLP 3D printing resolution of z = 10 µm per layer, which was used for mold manufacturing. Nevertheless, a layer height of z = 10 µm is a relatively low value compared to other, especially non-photopolymerzing 3D printing methods [
31]. Consequently, the utilized DLP technique allows µIM of implants with a relative high resolution. An alternative photopolymerizing 3D printing technique that enables a higher resolution and a lower staircase effect, could be two-photon polymerization (2PP) [
19]. However, in comparison to DLP, the use of 2PP would most likely lead to significantly higher 3D printing times. With our current state of knowledge, it is not clear which resolution is needed for therapeutically effective RNIs. In [
40] we reported good fitting accuracy of prototypes of human RNIs with a z = 100 µm per layer, which were implanted in human cadaver RWN. We further found that a higher resolution, respectively a lower z-value per layer, leads to an increased contour accuracy. The highest possible contour accuracy might be desirable. An increased contour accuracy most likely may lead to a better interface between the implant and the RWM. Consequently, a more effective drug transport into the inner ear by drug diffusion through the RWN might be achieved. Further investigations are however, needed. At this point, the xy-resolution of 32 µm and the z-resolution of z = 10 µm (layer height) used here enable 3D printing of a relatively high resolution and a high contour accuracy.
We used the silicone elastomer MED-4244 since it is a soft and stretchable material [
33]. Materials with such mechanical behavior are favorable for RNIs because of the beneficial tactile feedback and handling during implantation while also minimizing the likelihood of traumatizing sensitive structures such as the RWM during insertion [
40]. Moreover, there is high potential to adapt our rapid tooling based µIM manufacturing process to other medical grade materials and biodegradable materials, as such materials are established for IM processes [
23,
24].
3.3. Drug Release
The release of DEX from the GP-RNI shows a two-phase progression with a burst release at the beginning, followed by a slower release (
Figure 10 A,B). The burst release occurred within the first 13 hours (
Figure 10 C,D), during which a relatively large amount of DEX (about 0.7 µg in total) is released. The release of DEX from the GP-RNIs showed a diffusion-controlled mechanism, behaving like a matrix system [
41,
42]. As diffusion is dependent on the concentration gradient between the drug-releasing implant and the release medium, a faster diffusion results with a higher concentration gradient. The gradient remains maximal at the beginning of the release. A slower release phase follows with a linear slope. A release rate of about 1 µg/week is observed from the second week onwards. At the time of evaluation, this phase had not yet ended. After 6 weeks, about 10% of the DEX (8.2±0.6µg) had been released.
The drug release behavior found is promising, as the GP-RNI allows prolonged drug release over several weeks to months. Such release behavior can be beneficial compared to inner ear therapy methods [
4]. Nevertheless, the drug release from GP-RNIs has to be tested in a more realistic scenario in vivo. By intratympanic application, the drug has to pass the RWM. This affects the drug concentration reached in the inner ear [
43]. Further research on diffusion-based drug transport from the implant through the RWM is needed.
The thermal stress from our µIM process might increase the risk of the degradation of the drug load in the processed material [
43,
44]. In our process, we used a mold temperature of T
mold = 160 °C and a crosslinking time of t = 2 minutes. These parameters should not lead to significant degradation of the drug-load in the processed µIM material as DEX has a melting temperature of T = 262.4 °C. There is a rapid decomposition at higher temperatures [
45], but process temperatures below are considered to be suitable. Farto-Vaamonde et al. successfully processed a DEX-loaded filament via extrusion-based 3D printing at extrusion temperatures of T = 220 °C [
45]. In the work of Li et al., DEX was exposed to a temperature of T = 185 °C for a period of 5 minutes during hot melt extrusion, but the authors reported no significant drug degradation [
46]. However, as Farto el al. recommend in [
45], unnecessarily long heating periods at relatively high temperatures should be avoided as much as possible to prevent stability problems with the drug. The risk of drug degradation due to thermal stress should be taken into account, especially when using further, thermal sensitive drugs.
3.4. Biocompatibility and Bio-Efficacy
Compared to the blank (100%) the cell viability of the PC, including the cytotoxic agent DMSO, was significantly reduced (12 ± 9%; p < 0.001), proving the successful experimental setup. The cell viability of all GP-RNI-supernatant samples (mean ± SD; day 1: 114 ± 19%; day 3: 80 ± 13%; day 7: 94 ± 16%; day 10: 85 ± 22%; day 14: 99 ± 9%; day 21: 102 ± 9%; day 28: 104 ± 13%) did not differ significantly from that of the blank (
Figure 11). The lowest cell viability was 80.44 ± 13.30%. It was found when the supernatant of sampling day 3 was applied. It is still clearly above the 70% of the blank, which is the mark for indicating cytotoxic potential.
Cells without stress (NC) showed a very low basic level of TNF-α-production (11.58 ± 10.44 pg/ml), while this level increased significantly when LPS was added (PC, 776.2 ± 106 pg/ml). Compared to the PC all tested supernatant reduced the TNF-α amount in the DC-cell-LPS-stress test significantly (
Figure 12). This anti-inflammatory effect was highest on the 10th day (188.2 ± 136.5 pg/ml) and lowest on the 3rd day (318.9 ± 186.3 pg/ml). During the sampling period, the anti-inflammatory effect of the eluate varied. Data and results of statistical analyses are shown in the appendix (
Table A1).
Our results show neither the usage of photopolymeric molds for µIM nor the drug load of DEX affect the biocompatibility of the used RNI material silicone elastomer MED-4244 critically. The exact amount of DEX released in the supernatant, which was used for biocompatibility investigations, is unknown at this point. Toxic effects of DEX are reported in literature even for relatively low concentrations of 3 µM (0.00118 mg/ml, DEX: 392.46 g/mol) [
47]. The authors report the start of toxic effects on outer hair cells by that drug concentration in vitro. However, as we reported previously [
16], the findings in literature concerning critical drug concentrations in cochlear pharmacotherapy are not consistent, widespread and hard to compare as different individual experimental parameters must be considered. With an MTT assay as we used here in this work, we found no significant toxic effects for DEX concentrations up to 2000 µM (0.784 mg/ml) [
16]. With regard to the slow drug release behavior of the tested GP-RNI samples in isotonic saline, the DEX concentration in the supernatant should be far below 2000 µM. This supports our findings that the GP-RNIs containing DEX and which are made from 3D printed molds are biocompatible.
We found significant anti-inflammatory effects during the whole 28 day course of investigation. The anti-inflammatory effects of DEX are well known and there are further potentials in terms of the protection of CI patients from hearing loss, fibrotic CI encapsulation and spiral ganglion degeneration [
16]. In literature, effective concentrations for DEX are found from 0.00118 mg/ml [
47] to 24 mg/ml [
48]. However, because of a wide variety of experimental parameters and treatment protocols, findings from literature are hard to compare and there is a large variability between concentrations being toxic in vivo and those having a beneficial effect [
16]. Further in vivo investigations need to show what DEX concentrations are needed in RNIs to receive specific therapeutic effects for inner ear therapy. Many studies highlight the therapeutic potentials of DEX for inner ear diseases, such as [
15,
49,
50]. In addition to pure DEX, as we used here, there are other drug formulations such as dexamethasone dihydrogen phosphate disodium (DPS). In a previous study, we found a slight tendency for DPS to be more effective in reducing TNF-α-production than other DEX formulations [
16]. Moreover, other glucocorticoids such as prednisone and hydrocortisone (cortisol) offer therapeutic potential for inner ear therapy [
51] and are promising for inclusion in further investigations.