3.1. Implant Design and Characterisation
A rod-shaped implant with a size of 2.5 × 40.0 mm was chosen because this shape and these dimensions are similar to dimensions that have already been shown to be acceptable in commercially available products and applicator devices have already been developed for an implant of these dimensions [32
]. Implants were loaded with MB (68.6 ± 5.1 mg), IS (68.1 ± 3.0 mg) or IA (72.3 ± 3.2 mg). Images of the produced implants are shown in Figure 3
A–H. These images give an appreciation of the actual geometry of the 3D printed “windows” in comparison to what was designed. Figure 3
C, E shows that, although the 1.0 × 1.0 mm has been printed to the correct size, they are more circular in shape than square like the design. This is because the resolution of FDM printers is not as high as that displayed by other types of 3D printing such as stereolithography [33
Implants A–E loaded with MB, Implants B and E loaded with IS and Implant B loaded with IA were tested using the agitated vessel release model. Implant E loaded with MB or IS was tested using the agarose gel release model. IA was not included in the agarose release model because of its poor solubility and the difficulties this would present to maintain sink conditions. These molecules were used due to their differing solubility values: MB 40 mg/mL [34
]; IS 100 mg/mL [35
] and IA 0.021 mg/mL [36
]. These three molecules cover a wide range of hydrophobicity. Therefore, they are good candidates to establish how this parameter affects drug release from the 3D printed implantable devices. The influence of the solubility on the release profiles can be used to anticipate the release kinetics of other drugs loaded within the implants described here.
The architecture and topology of the 3D-printed implants were analysed using a Bruker SkyScan 1172 system μCT (Figure 3
I–K). Cross-section reconstructions in the y
plane of an implant containing (Figure 3
I) MB and (Figure 3
J) IS were performed, and representative x
cross-section of a 3D-printed implant was used for quantitative analysis. These images (Figure 3
I,J) give an appreciation of the drug distribution within the cavity of the implant and show that the drug distribution is uniform for both MB and IS. The dimensional measurements calculated at different locations over the implant 3D volume for the core and shell of the samples are reported in Figure 3
K and show that there is no significant (p
> 0.05) difference in the size of the drug core for either drug. This indicates that the drugs were dispersed through the entire implant cavity and that the packing process did not damaged the implant structure.
Dissolution of the PVA “windows” in Implants B and E were visualised using OCT, digital microscopy and SEM and are shown in Figure 4
A,B, respectively. It can be seen that complete dissolution of the PVA “window” in Implant B occurred after 25 min (Figure 4
A1). Whereas, complete dissolution of the PVA “window” in Implant E took 35 min (Figure 4
B1). Despite the “window” in Implant B being significantly larger than the “window” in Implant E, it fully dissolved more quickly. This may be explained by the reduced surface area-to-volume ratio of the “window” in Implant E, reducing the rate of dissolution for this implant. Goyanes et al. investigated the effect that the surface area-to-volume ratio had on the dissolution of PVA tablets and reported that a higher surface area to volume ratio resulted in tablets that dissolved more quickly [37
]. It is important to note that the PVA “window” is designed to dissolve quickly to allow the drug to diffuse trough the generated “window”. The “window” material can be tailored to achieve a delayed drug release. Additionally, in the last section of the manuscript, an alternative coating approach was described to prepare implants allowing sustained drug release over months. It is important to note that a quick-dissolving commercial PVA filament was used for this study. PVA is a biocompatible polymer [38
], but commercial filaments can have potential excipients, such as plasticisers, that are not ideal for medical applications. However, the present work is a proof-of-concept study exploring the influence of the structure of the implant on the drug release kinetics. Accordingly, a commercial PVA was used as it was the quickest approach. However, future work will require the use of filaments prepared using pure biocompatible polymer. This approach opens the possibility of developing implants with delayed release by printing the implant windows with polymers with slower dissolution/disintegration kinetics such as cellulose derivatives [39
To predict robustness of the designed implants, their break strength and degree of flexibility were evaluated. A very rigid implant is likely to break during insertion or in situ; therefore, a degree of flexibility is required, as well as sufficient strength to withstand insertion and remain mechanically strong enough for the duration of drug release. If an implant breaks or cracks, it is likely to cause an increase or a burst in the rate of drug release which would, in turn, cause undesirable side effects in the patient. The maximum force required for breaking the implants and the angle of bending at the break point were calculated for each implant configuration and shown in Figure 5
It can be seen from Figure 5
that there is no significant difference in the breaking force of Implants B–E (PLA implants). A significantly (p
< 0.5) larger force was required to break Implant A (PVA), than was required for Implants B–E. This test was performed to evaluate if changing the design of the release “windows” from the implant has a direct influence on the mechanical properties of the resulting material. No mechanical tests directly comparable to those performed in this study have been performed on commercially available implantable drug delivery devices. However, mechanical testing of medical devices has been extensively reported. The results obtained here can be compared with the results reported by Horal et al. for 3D-printed PLA screws for orthopaedic applications [41
]. In this case, PLA screws were manufactured and a three-point bending test was performed. The dimensions of these implants were similar to the ones described here (1–2 mm), and the forces applied during the bending tests were lower than the ones reported here (ranging between 0.5 and 10 N). These screws where designed for bone healing applications. Higher forces will be applied to bone screws than to implants designed to be implanted in soft tissue. Therefore, the implants presented here showed fracture forces higher than the forces that will be expected for soft tissue implants. As PLA has a long degradation time, up to 2 years [42
], degradation of the implant structure would not be expected to have an effect on the mechanical properties during drug release or an effect on the release rate itself.
3.2. In Vitro Drug Release
MB has some inherent antibacterial activity; therefore, bacterial growth in the release media was not anticipated to be an issue for these implants [43
]. However, SA was added to IS and IA release media to prevent microbial growth [23
] over the course of the release experiment. The release profiles of MB from each of the five implant designs are shown in Figure 6
. Implants made entirely from PVA (Implant A) had the most rapid drug release, with 100% of drug releasing within 24 h. As expected, Implants B and C showed significantly extended release profiles in comparison with Implant A, with release time being extended to over six days. Although, Implants B and C took the same time to reach 100% release, Implant C showed a more sustained release profile, which showed less variation. Implants D and E showed an extended release profile in comparison to the other implants and show that reducing the size and number of “windows” effectively prolongs release from this type of implant. The release profiles of MB from each of the PVA “window” implants were compared using similarity and difference factor (F1
), and the results are shown in Table 1
. Implant A had a significantly different release profile to Implants B and C as the F1
values were higher than 15 and the F2
values were lower than 50. Implants B and C and Implants D and E also showed significantly different release profiles to each other. These results indicate that implant design has potential to modify the release profile of a loaded molecule by simply changing the design of the implant. Interestingly, implants with 1.0 × 1.0 mm “windows” were capable of providing drug release over 25 days. A sustained release profile like this can be useful for local antimicrobial therapy or for pain management after surgery [46
]. In these cases, a prolonged release over a period of a few weeks can be extremely beneficial to prevent infections or for pain management. However, for prolonged applications alternative approaches need to be evaluated. For this purpose, coated implants were evaluated. This approach is described in Section 3.3
of the present manuscript.
The effect of drug properties on release from the designed implants was investigated by comparing the release profiles of MB (solubility 40 mg/mL [34
]), with IS (solubility 100 mg/mL [35
]) and IA (solubility 0.021 mg/mL [36
]). The release profiles of IS from Implants B and E are shown in Figure 7
A. The release rate of IS from Implant B was significantly increased in comparison to MB from the same implant. Complete IS release was achieved after just 80 min, whereas, 100% MB release took seven days. A similar increase in release rate is seen for Implant E, with 100% IS release achieved after six days and MB release after 25 days. These results show that obviously the implant design is not the only factor that contributes to change the release profile. The physicochemical properties of the drug loaded are important too. All in vitro releases were carried out under sink conditions; therefore, it is the dissolution rate of each of the drugs rather than solubility that is having an impact on drug release from the implant. Accordingly, changing the nature of the loaded molecule or including a formulation with a slower dissolution rate will provide an extra degree of control over the release profile.
IA release from Implant B is shown in Figure 7
C. The release of this compound is significantly extended in comparison with MB and IS release from the same implant design, with release taking ten days in comparison to six days and 80 min for MB and IS, respectively. As mentioned previously, the release rate of this drug is slower due to its slower dissolution kinetics, confirming that the nature of the drug loaded need to be carefully considered for each application type.
shows the release profiles of MB and IS from Implant E into an agarose gel release model. Release is expected to be slower in the agarose gel when compared to the agitated vessel release model. Within the agitated vessel model, convection rapidly homogenises the drug within the release media, thus, maintaining the drug concentration gradient at the interface of the implant with the release media. However, living tissues exhibit different conditions than those applied in the in vitro agitated vessel method. The extracellular matrix that these formulations are likely to be in contact with after implantation behave more like a gel than like a bulk fluid [48
]. Despite the existence of a large number of biorelevant media for simulating physiological fluids, there is still not an accepted standard for simulation of subcutaneous environment [48
]. Agarose gels form a 3D structure linked by hydrogen bonds with pore sizes similar to those encountered in physiological tissue and have been suggested as a more realistic in vitro release model than bulk fluid [29
]. Moreover, multiple research works have reported the suitability of agarose hydrogel as a good release medium simulating soft tissues [50
Both drugs demonstrated progressive drug release over a prolonged period. Figure 8
A,B shows the release obtained for MB-loaded implants. These results showed that the closest region (1.5 cm) to the implant reached a plateau in MB levels after seven days. However, in further regions the MB concentration increased over time up to 40 days for the further regions (4.5 cm). This shows that MB was continuously delivered over 40 days. This MB concentration increase is not due only to MB diffusion through the agarose gel, as the concentration always increased. This suggests that there was a constant MB release that took place over time. After 40 days no significant differences were found in the release obtained at different distances from the implant (p
> 0.05). This indicated that MB concentration all over the agarose gel was equivalent and that there was no concentration gradient that will drive more release. Similar behaviour was observed for IS (Figure 8
C,D) over a period of 21 days. These results confirm that the testing conditions had a substantial influence on the release results. Moreover, this set of results suggest that the selected implants can be used to provide drug release over periods of several weeks. Similarly, Hoang et al. investigated releases of ciprofloxacin hydrochloride and vancomycin hydrochloride from bone implants over 48 and 96 h, respectively and showed that release into an agarose model was extended when compared to release of the same drugs from the same implants into an agitated vial [29
The releases achieved in this work range from just 80 min to over 25 days in an agitated vessel and over 40 days in an agarose gel model and show promise as drug delivery systems for prolonged drug delivery. The use of local anaesthetics (commonly, bupivacaine, lidocaine and procaine) to treat localised pain has many advantages when compared with the systemic administration of opioids [14
]. Work has been carried out to optimise the drug delivery of these agents to achieve localised delivery and limit peripheral side effects. An implantable device that could locally deliver anaesthetic over days or weeks could be of benefit for delivery of these drugs. Currently, the majority of chemotherapeutic agents are delivery systemically. This allows the drug to distribute throughout the entire body, including to healthy tissues, causing adverse side effects [54
]. Polymeric devices aiming to locally deliver cancer drugs have been investigated and aim to improve the delivery of these drugs by providing localised sustained delivery and, therefore, reduce the effect on healthy tissue. Salmoria et al. investigated the use of polymeric implant to locally deliver fluorouracil and showed a desirable release rate over 45 days [54
]. Localised delivery of antibiotics may offer advantages over conventional oral delivery for localised conditions. Gimeno et al. showed promising delivery of antibiotics which could be tailored by changing the implant design, from rapid drug release within 20 h to longer release times around 200 h for the potential prevention of orthopaedic-implant-associated infections [15
]. These examples highlight instances where the implants developed in this work could be used.
3.3. In Vitro Drug Release from Coated Implants
The results described in previous sections show that these implants can be used for sustained drug delivery over periods of several weeks. The treatment of some medical conditions, especially chronic conditions, can be improved significantly with drug delivery devices capable of providing drug release over prolonged periods of time. These periods of time range from months up to years for potent compounds such as hormones. Examples of this will be the treatment of chronic conditions or even pre-exposure prophylaxis of human immunodeficiency disease (HIV).
A good alternative to obtain implants with prolonged drug release profiles is to coat them with a membrane capable of sustaining drug release [55
]. Accordingly, a simple dip coating procedure can be used to prepare implants with prolonged drug release profiles. Accordingly, a thin film covers the surface of the implant acting as a rate controlling membrane [9
]. Figure 9
shows the release profile of MB from implants (Implant B) coated with a PCL-based formulation. It can be seen that the PCL rate controlling membrane is capable of providing sustained drug releases over periods of 300 days. Interestingly, non-coated equivalent implants showed MB release profiles extended over only four days (Figure 6
). These results suggest that PCL coating could be an ideal approach for applications that require drug release over longer periods of time. PCL has been described previously as a good candidate to prepare rate controlling membranes for drug delivery applications [9
]. PEG membranes have been used before to release tenofovir alafenamide for HIV pre-exposure prophylaxis [56
]. These systems achieved prolonged releases between 100 and 200 days. Considering that tenofovir alafenamide shows a lower water solubility than MB, the system described has great potential for sustaining the release of hydrophilic molecules as MB showed up to 300 days of release.