Fabrication and Characterisation of 3D-Printed Triamcinolone Acetonide-Loaded Polycaprolactone-Based Ocular Implants

Triamcinolone acetonide (TA) is a corticosteroid that has been used to treat posterior segment eye diseases. TA is injected intravitreally in the management of neovascular disorders; however, frequent intravitreal injections result in many potential side effects and poor patient compliance. In this work, a 3D bioprinter was used to prepare polycaprolactone (PCL) implants loaded with TA. Implants were manufactured with different shapes (filament-, rectangular-, and circle-shaped) and drug loadings (5, 10, and 20%). The characterisation results showed that TA was successfully mixed and incorporated within the PCL matrix without using solvents, and drug content reached almost 100% for all formulations. The drug release data demonstrate that the filament-shaped implants (SA/V ratio~7.3) showed the highest cumulative drug release amongst all implant shapes over 180 days, followed by rectangular- (SA/V ratio~3.7) and circle-shaped implants (SA/V ratio~2.80). Most implant drug release data best fit the Korsmeyer–Peppas model, indicating that diffusion was the prominent release mechanism. Additionally, a biocompatibility study was performed; the results showed >90% cell viability, thus proving that the TA-loaded PCL implants were safe for ocular application.


Introduction
Ocular health possesses a significant effect on people's well-being and economic growth. The World Report on Vision reported that at least 2.2 billion people had blindness or severe vision impairment [1]. Damage to the posterior segment of the eye is typically the cause of irreversible blindness, and the treatment of this disease remains a challenge for formulation scientists due to the unique characteristics of the eye [2].
Generally, posterior eye segment diseases can be treated via topical, systemic, or periocular routes [3]. However, all these routes have some limitations, resulting in an insufficient level of drug at the target site. Topical dosage forms often fail to deliver drugs to the target site of action because of the short residence time in the cul-de-sac, and consequently demonstrate poor bioavailability of the drug, usually less than 5% [4]. Systemic drug delivery requires high doses to be administered because of the bloodaqueous and blood-retinal barriers, as well as rapid drug clearance from the eye. Moreover, high doses of steroids can increase harmful side effects [5].

Characterisation of TA-Loaded PCL Implants
Chemical interactions between the components of TA-loaded PCL implants were evaluated using Spectrum Two FTIR (Perkin Elmer, Waltham, MA) at room temperature. The IR spectra were recorded in the range of 4000-600 cm −1 and analysed using Spectrum 10 software. The resolution and the number of scans to record IR spectra were 4 cm −1 and 32, respectively.
The surface morphology of TA-loaded PCL implants was evaluated using scanning electron microscopy (SEM) (Hitachi TM3030; Tokyo, Japan). The observation condition was set to EDX mode. Implants were sectioned accordingly and mounted on HITACHI SEM Cylinder Specimen Mounts.

Characterisation of TA-Loaded PCL Implants
Chemical interactions between the components of TA-loaded PCL implants were evaluated using Spectrum Two FTIR (Perkin Elmer, Waltham, MA) at room temperature. The IR spectra were recorded in the range of 4000-600 cm −1 and analysed using Spectrum 10 software. The resolution and the number of scans to record IR spectra were 4 cm −1 and 32, respectively.
The surface morphology of TA-loaded PCL implants was evaluated using scanning electron microscopy (SEM) (Hitachi TM3030; Tokyo, Japan). The observation condition was set to EDX mode. Implants were sectioned accordingly and mounted on HITACHI SEM Cylinder Specimen Mounts.
The thermal properties of the 3D-printed implants with and without TA were evaluated. For this purpose, thermogravimetric analysis (TGA) was performed to measure the weight loss of the 3D-printed ocular implants. TGA was performed using a Q50 Thermogravimetric analysis (TA instruments, Bellingham, WA, USA). Scans were run from 25 to 500 °C, at the heating rate of 20 °C/min under a nitrogen flow rate of 40 mL/min. Moreover, a Q20 differential scanning calorimeter (DSC) (TA instruments, Bellingham, USA) was used to establish if the TA was crystalline or amorphous within the performed 3Dprinted ocular implants. Scans were run from 20 °C to 300 °C at 20 °C/min under a nitrogen flow rate of 50 mL/min. The thermal properties of the 3D-printed implants with and without TA were evaluated. For this purpose, thermogravimetric analysis (TGA) was performed to measure the weight loss of the 3D-printed ocular implants. TGA was performed using a Q50 Thermogravimetric analysis (TA instruments, Bellingham, WA, USA). Scans were run from 25 to 500 • C, at the heating rate of 20 • C/min under a nitrogen flow rate of 40 mL/min. Moreover, a Q20 differential scanning calorimeter (DSC) (TA instruments, Bellingham, USA) was used to establish if the TA was crystalline or amorphous within the performed 3D-printed ocular implants. Scans were run from 20 • C to 300 • C at 20 • C/min under a nitrogen flow rate of 50 mL/min.

Scaffold Dimensions Characterisation
The surface area and volume of the implants were calculated based on the dimension of the implants. The surface-area-to-volume (SA/V) ratio was calculated by dividing the surface area by the volume of the implants. This study was completed in triplicate (n = 3).

Drug Content Analysis
TA-loaded PCL implants from each formulation were randomly chosen and accurately weighed before sample preparation. Samples were dissolved in 1.0 mL of a solvent mixture (methanol:dichloromethane = 3:5) and vortexed until thoroughly dissolved. Subsequently, samples were placed in the fume hood until solvent was completely evaporated. 1 mL of methanol was added to samples, followed by centrifugation at 5000 rpm for 15 min. A hundred microliters of supernatant were taken and diluted with methanol until 1.0 mL. Finally, sample solutions were filtered through a 0.22 µm membrane filter before being injected into the HPLC system. The HPLC system was set in an isocratic system with acetonitrile; water (45:55% v/v) used as the mobile phase. The flow rate was set at 0.9 mL/min and the temperature was maintained at 35 • C. The separation was performed using an Infinity Poroshell 120 EC-C18 column (4.6 × 250 mm, 4 µm) (Agilent, Santa Clara, CA, USA). The chromatogram was recorded at a wavelength of 236 nm and the injection volume was 20 µL. All measurements were performed in triplicate. The percentage of drug loaded was determined with Equation (1):

In Vitro Drug Release Study
An in vitro drug release study was carried out in glass vials filled with 4.0 mL of 0.5% w/v sodium dodecyl sulfate (SDS) in PBS buffer pH 7.4 containing 0.05% w/v sodium azide, in order to maintain sink conditions. The glass vials were then stored in an incubator shaker that was shaken at 40 rpm and maintained at 37 • C. All of sample solutions were collected and replaced with 4.0 mL fresh media at predetermined times. Sample solution was filtered using 0.22 µm membrane filter and to be analysed using HPLC system as used in TA content determination. The cumulative percentage release of TA was plotted against time. Experiments were conducted in triplicate (n = 3).
The release kinetics of TA from each formulation was analysed using the Higuchi (Equation (2)) and Korsmeyer-Peppas (Equation (3)) mathematical models. Only the portion of the graph showing up to 60% drug release was fitted to the Korsmeyer-Peppas model.
where Q t /Q ∞ is the fraction of drug released at time t, K H is the Higuchi dissolution constant, and K KP is the Korsmeyer-Peppas constant. The value of the exponent "n" in the Korsmeyer-Peppas model can be used as an indication of the release mechanism [47,48]. If the obtained "n" value is around 0.5, this indicates that the drug is been released following Fickian diffusion [47,48]. On the other hand, if the obtained "n" value ranges between 0.5 and 1, it indicates anomalous (non-Fickian) transport [49][50][51]. Finally, when the "n" value is around 1, it indicates case II transport [47,48].

Implants Degradation
Degradation of PCL implants was evaluated by weighing the initial dry implants (W o ), immersing them in the release media, incubating at 37 • C, and shaking at 40 rpm. At the end of the degradation study, the implants were dried overnight under vacuum conditions at room temperature. Dried implants were then weighed and recorded as W t . The percentage weight remaining was calculated using Equation (4): SEM analysis was performed to observe the surface morphology of the implants at the end of the release study. Mixture F-12), modulated with HEPES/glutamine and enriched with 10% heat-inactivated foetal bovine serum and 1% penicillin/streptomycin stock solution (10,000 units/mL penicillin and 10 mg/mL streptomycin). Cells were then incubated in humified conditions (5% CO 2 /air) at 37 • C.

Cell Passage
ARPE-19 cells were cultured in T75 cm 2 media consisting of DMEM/F-12, modulated with HEPES/glutamine and enriched with 10% heat-inactivated foetal bovine serum and 1% penicillin/streptomycin stock solution (10,000 units/mL penicillin and 10 mg/mL streptomycin); complete media change was performed every 2-3 days. At the third passage, the cell pellet was seeded in a 96-well tissue culture plate with a seeding density of 2 × 10 4 cells/well. Finally, cells were incubated for 36 h before the biocompatibility study.

Sample Preparation
In this study, filament-shaped implants were used to evaluate the in vitro biocompatibility of the samples. The test was carried out using indirect and direct assay methods according to ISO 10993-5:2009-Biological evaluation of medical devices. For the indirect assay, implants were dipped in 70% ethanol for sterilisation purposes, immersed in 4 mL of DMEM/F12 containing 1.2% streptomycin/penicillin, and stored for 24 h in an incubator. Finally, 200 µL of sample solution was pipetted into pre-incubated TCPs and incubated for 24 h. For the direct assay, implants were placed on top of ARPE-19 cells in 96-well plates. Subsequently, 200 µL of fresh media was added and incubated for 24 h at 37 • C. After incubation, the 96-well plate was inverted to remove the implants.

Cell Viability Assay
Cell viability assay was performed based on resazurin reduction assay. Twenty microliters of resazurin sodium salt solution was added to 200 µL samples in a 96-well plate photometer and incubated for 4 h. Samples were then analysed using fluorescence at 545 nm, and cell viability was calculated based on the optical density (OD) by using Equation (5). DMEM/F12 and DMSO were used as negative and positive controls, respectively. All experiments were performed in triplicate (n = 3).

Statistical Analysis
Data were analysed with GraphPad Prism ® version 9 (GraphPad Software, San Diego, CA, USA) and Microsoft Excel. The results are presented as means ± standard deviation (SD). The comparative study was performed using a t-test and one-way variance analysis (ANOVA). Significant difference was accepted with a p-value < 0.05.

Fabrication and Characterisation of TA-Loaded PCL Implants
PCL is one of the FDA-approved biodegradable polymers for developing drug delivery systems due to its biocompatibility and biodegradability [18][19][20][21]. TA-loaded PCL-based implants were successfully manufactured using semi-solid extrusion (SSE) 3D printing technology. Implants were printed in filament-like, rectangular, and circular shapes ( Figure 2) using all the formulation compositions listed in Table 1. printing technology. Implants were printed in filament-like, r shapes ( Figure 2) using all the formulation compositions listed in SEM was used to confirm the surface morphology of TA-l shown in Figure 3, the surface of PCL-based implants was smoo no visible drug crystals or aggregates were found on their surfa miscible and evenly distributed throughout the PCL matrix. Ad the blank PCL implants (T0L20H80 and T0L40H60) was also sm thus showing that both H-PCL and L-PCL were evenly mixed. O cate the capability of the SSE 3D printing technology to produc plants with a vast range of doses. SEM was used to confirm the surface morphology of TA-loaded PCL implants. As shown in Figure 3, the surface of PCL-based implants was smooth and homogenous, and no visible drug crystals or aggregates were found on their surfaces, indicating that TA is miscible and evenly distributed throughout the PCL matrix. Additionally, the surface of the blank PCL implants (T0L20H80 and T0L40H60) was also smooth and homogeneous, thus showing that both H-PCL and L-PCL were evenly mixed. Overall, these results indicate the capability of the SSE 3D printing technology to produce homogenous solid implants with a vast range of doses.
ATR-FTIR analysis was investigated to define the chemical interaction between the PCL and TA. The ATR-FTIR infrared spectra of TA and blank PCL matrix implants were compared with the TA-loaded PCL implants, as presented in Figure 4. TA exhibited a characteristic infrared absorption band at 3395 cm −1 and 2948 cm −1 that linked to the stretching vibration of the OH and C-H groups, respectively. Moreover, the absorption band at 1707 cm −1 represents the C=O stretching vibration of this drug. The band located at 1661 cm −1 and 1609 cm −1 corresponds to the stretching vibration of C=C bonds. Other specific infrared peaks of TA were located at 1171 cm −1 and 1056 cm −1 , and were attributed to the C-O-C bond in aliphatic esters and the stretching vibration of C-F, respectively. The bands of TA found in this study were comparable to those previously obtained in other works [6,26,52]. Blank PCL matrices are composed of two PCL polymers with varying molecular weights. Moreover, both blank PCL-based implants showed a strong carbonyl stretching vibration at 1723 cm −1 .
Moreover, the bands located at 2940 and 2865 cm −1 represent the stretching vibration of C-H bonds. The band located at 1180 cm −1 represents C=O=C stretching. Another characteristic PCL band can be found at 1293 cm −1 , associated with C=O and C-C stretching in the crystalline phase of PCL [38,[53][54][55]. All spectral characteristics of PCL were observed in TA-loaded implants, although TA peaks were not observed when the drug loading was low. This phenomenon has also been documented in several publications [56,57]. When the drug loading is low, the bulk polymer is anticipated to conceal or obscure the FTIR contributions of the drug at low loading levels [58]. However, by increasing the drug loading, peaks at 1661 cm −1 and 1609 cm −1 were identified, in addition to those of PCL. These characteristics could be used to determine the presence of TA in implants. ATR-FTIR analysis was investigated to define the chemical interaction betw PCL and TA. The ATR-FTIR infrared spectra of TA and blank PCL matrix impla compared with the TA-loaded PCL implants, as presented in Figure 4. TA ex characteristic infrared absorption band at 3395 cm −1 and 2948 cm −1 that linke stretching vibration of the OH and C-H groups, respectively. Moreover, the ab  Moreover, the bands located at 2940 and 2865 cm −1 represent the stretching vibration of C-H bonds. The band located at 1180 cm −1 represents C=O=C stretching. Another characteristic PCL band can be found at 1293 cm −1 , associated with C=O and C-C stretching in the crystalline phase of PCL [38,[53][54][55]. All spectral characteristics of PCL were observed in TA-loaded implants, although TA peaks were not observed when the drug loading was low. This phenomenon has also been documented in several publications [56,57]. When the drug loading is low, the bulk polymer is anticipated to conceal or obscure the FTIR contributions of the drug at low loading levels [58]. However, by increasing the drug loading, peaks at 1661 cm −1 and 1609 cm −1 were identified, in addition to those of PCL. These characteristics could be used to determine the presence of TA in implants. Moreover, no new peaks were observed, suggesting no chemical reactions occurred during the 3Dprinting process.
DSC analysis of the pure TA, blank PCL implants, and TA-loaded PCL implants was performed to understand TA incorporation into the polymer ( Figure 5A). Both blank PCL implants (T0L20H80 and T0L40H60) showed an endothermic peak between 54 to 59 °C [18,34,55]. Moreover, a sharp endothermic peak at 290 o C corresponds to the melting temperature of TA, as previously reported [21]. This endothermic melting peak of TA was not observed in any of the TA-loaded PCL implants, thus indicating that the crystalline form of TA was converted to the amorphous form after being combined with the PCL matrix (H-PCL and L-PCL). Similar outcomes have already been reported for different drugs such as dipyridamole (DIP) [31] or acetylsalicylic acid (ASA) [33], as well as for other types of polymeric matrix such as thermoplastic polyurethane (TPU) [32]. DSC analysis of the pure TA, blank PCL implants, and TA-loaded PCL implants was performed to understand TA incorporation into the polymer ( Figure 5A). Both blank PCL implants (T0L20H80 and T0L40H60) showed an endothermic peak between 54 to 59 • C [18,34,55]. Moreover, a sharp endothermic peak at 290 • C corresponds to the melting temperature of TA, as previously reported [21]. This endothermic melting peak of TA was not observed in any of the TA-loaded PCL implants, thus indicating that the crystalline form of TA was converted to the amorphous form after being combined with the PCL matrix (H-PCL and L-PCL). Similar outcomes have already been reported for different drugs such as dipyridamole (DIP) [31] or acetylsalicylic acid (ASA) [33], as well as for other types of polymeric matrix such as thermoplastic polyurethane (TPU) [32]. TGA data ( Figure 5B) were obtained to assess the thermal stability and degradation profiles of pure TA, blank PCL implants, and TA-loaded PCL implants. TGA results showed that TA started degrading at temperatures slightly above 300 °C. Moreover, as TGA data ( Figure 5B) were obtained to assess the thermal stability and degradation profiles of pure TA, blank PCL implants, and TA-loaded PCL implants. TGA results showed that TA started degrading at temperatures slightly above 300 • C. Moreover, as expected, the PCL blank implant containing the highest percentage of L-PCL (T0L40H60) started degrading at temperatures around 300 • C. In contrast, the PCL blank implant containing the highest percentage of H-PCL (T0L20H80) showed a higher T onset (around 400 • C). Therefore, the T onset of the TA-loaded PCL implants was shifted lower when a higher percentage of L-PCL and/or TA was added.

TA Content Analysis
Ensuring that the chosen fabrication method of TA-loaded PCL implants is capable of producing samples with uniform drug distribution throughout the sample is critical to ensure controlled drug release [6]. Implants with different shapes from each formulation were randomly selected and analysed using the optimised HPLC method. The mean value of relative drug content from tested implants was 99.98 ± 1.60%, with a p-value > 0.5236. The results of the uniformity content confirmed that the TA was distributed uniformly throughout the polymer matrix ( Figure 6). The results indicate that the 3D printing implant manufacturing technique is reproducible and produces a uniform drug distribution in the PCL matrix.
started degrading at temperatures around 300 °C. In contrast, t taining the highest percentage of H-PCL (T0L20H80) showed °C). Therefore, the Tonset of the TA-loaded PCL implants was sh percentage of L-PCL and/or TA was added.

TA Content Analysis
Ensuring that the chosen fabrication method of TA-loade of producing samples with uniform drug distribution through ensure controlled drug release [6]. Implants with different sha were randomly selected and analysed using the optimised HPL of relative drug content from tested implants was 99.98 ± 1. 60 The results of the uniformity content confirmed that the TA throughout the polymer matrix ( Figure 6). The results indica plant manufacturing technique is reproducible and produces a in the PCL matrix.

In Vitro Release Study
In vitro release studies were conducted using PBS buffer (pH 7.4) containing 0.05% w/v sodium azide and 0.5% w/v SDS. Sodium azide was added to the media to prevent microbial growth during the in vitro release studies [59][60][61][62]. In addition, SDS can increase TA solubility by up to 14 times [63]. Figure 7 shows the cumulative drug release of all formulations over 180 days. As expected, formulations with a higher ratio of L-PCL (L40H60) released TA faster than those with a higher H-PCL ratio (L20H80). Therefore, implants containing a higher ratio of L-PCL exhibited two different release stages, and this effect was much less pronounced (or even non-existent) when implants contained a higher ratio of H-PCL. H-PCL has a longer chain length than L-PCL; thus, it is more hydrophobic than L-PCL. Additionally, a higher ratio of L-PCL in formulations may provide better interactions with release media so that TA can be quickly released from the implants. this effect was much less pronounced (or even non-existent) when implants contain higher ratio of H-PCL. H-PCL has a longer chain length than L-PCL; thus, it is more drophobic than L-PCL. Additionally, a higher ratio of L-PCL in formulations may pro better interactions with release media so that TA can be quickly released from the plants. The surface-area-to-volume ratio effect on drug release was also evaluated using ferent implant shapes ( Table 2). The filament implant has the highest surface-area-toume ratio (SA/V~7.30), followed by the rectangular shape (SA/V~3.70), and, finally circle shape (SA/V~2.80). The dissolution tests show that the highest SA/V ratio co sponds to the highest cumulative release, which is crucial in defining drug release prof The surface-area-to-volume ratio effect on drug release was also evaluated using different implant shapes ( Table 2). The filament implant has the highest surface-area-to-volume ratio (SA/V~7.30), followed by the rectangular shape (SA/V~3.70), and, finally, the circle shape (SA/V~2.80). The dissolution tests show that the highest SA/V ratio corresponds to the highest cumulative release, which is crucial in defining drug release profiles. These results follow the studies by Goyanes et al. and Reynolds et al. [64,65]. Furthermore, statistical analysis was performed to see any significant effect of the SA/V ratio on the cumulative drug release. A one-way ANOVA study showed significant differences in cumulative release amongst implant shapes in each formulation (p < 0.0001). The percentage cumulative release data were fitted to the Higuchi and Korsmeyer-Peppas models to assess drug release mechanism (Table 3). Most samples show high regression coefficient (R 2 ) values for Korsmeyer-Peppas, with different n-values. Rectangularshaped implants showed an n-value less than 0.5, indicating that the release mechanism is based on Fickian diffusion. In the Fickian diffusion mechanism, the solvent diffusion rate determines the drug release, rather than the polymer relaxation rate. These results agreed with previous reports [51,66,67]. On the other hand, the filament-and circle-shaped showed n-value lower than 0.5, suggesting a pseudo-Fickian diffusion mechanism [68]. Interestingly, in some cases, the n-values were slightly higher than 0.5, indicating an anomalous transport mechanism, which is the release is not only governed by diffusion but also influencing by the relaxation of polymeric chains [49][50][51]. This behaviour was found only in some L40H60 samples. This suggests that the presence of a higher content of low molecular weight PCL can contribute to matrix relaxation. Despite these small differences between the implants, all the n-values obtained were close to 0.5, thus suggesting that TA relies heavily on Fickian diffusion to be released from the implants. Considering that implants were made of PCL, which degrades slower, it is not surprising that the drug is released predominantly by diffusion, as has been reported in the past for PCL-based implants [29,69].

Implant Degradation
Implant degradation, quantified by percentage mass loss, was also investigated during in vitro release. Table 4 depicts the mass loss of all implant formulations over 180-days, with a higher mass loss observed from implants containing a higher ratio of L-PCL (L40H60). Further statistical analysis was performed to observe the parameters (drug loading or polymer ratio) that governed the mass loss of PCL implants. The statistical analysis results showed that different polymer ratios affect TA release from the implants, with the p-value (0.0027) < 0.5 for all drug loadings; whereas different drug loadings have no significant effect on mass loss (p value = 0.1357). The degradation rate of PCL not only depends on the morphological and structural formation, but is also influenced by the surface-area-to-volume ratio [70,71]. A higher SA/V ratio improves the penetration of release media into the implants and initiates the chain scission rate [72]. Therefore, a higher mass loss was observed from filament-shaped implants, which have the highest SA/V ratio. The changes in the surface morphology of PCL implants after in vitro release was also observed using SEM (Figure 8). Unlike TA-loaded PCL implants on day 0 (Figure 3), TA-loaded PCL implants showed a more porous structure after 180 days. Accordingly, we investigated pore size measurements for all implants after the drug release study (Table S1, Supplementary Data).

Biocompatibility Study
Filament-shaped implants were used for the biocompatibility study because this shape is more feasible for further application. An indirect biocompatibility assay was completed by calculating the viability of ARPE-19 cells after exposure to the release media. In

Biocompatibility Study
Filament-shaped implants were used for the biocompatibility study because this shape is more feasible for further application. An indirect biocompatibility assay was completed by calculating the viability of ARPE-19 cells after exposure to the release media. In contrast, the direct assay involved the placement of the implant on top of the ARPE-19 cells. According to ISO standards, a cytotoxic effect is considered when the percentage of cell viability is below 70% [73]. The results of the cell viability study are shown in Figure 9. The resazurin assay showed that the cells had more than 90% viability, indicating that implants were non-cytotoxic. Statistical analysis showed no significant differences between the negative control and all formulations for both indirect assay (p value = 0.1940) and direct assay (p value = 0.1901). Moreover, no significant differences were found when all formulations were compared with each other (p value > 0.05). According to the biocompatibility result, TA-loaded PCL implants are safe for ocular application. However, further in vivo studies are required to ensure the safety of the implants.

Biocompatibility Study
Filament-shaped implants were used for the biocompatibility study because th shape is more feasible for further application. An indirect biocompatibility assay was co pleted by calculating the viability of ARPE-19 cells after exposure to the release media. contrast, the direct assay involved the placement of the implant on top of the ARPEcells. According to ISO standards, a cytotoxic effect is considered when the percentage cell viability is below 70% [73]. The results of the cell viability study are shown in Figu  9. The resazurin assay showed that the cells had more than 90% viability, indicating th implants were non-cytotoxic. Statistical analysis showed no significant differences b tween the negative control and all formulations for both indirect assay (p value = 0.194 and direct assay (p value = 0.1901). Moreover, no significant differences were found wh all formulations were compared with each other (p value > 0.05). According to the b compatibility result, TA-loaded PCL implants are safe for ocular application. Howev further in vivo studies are required to ensure the safety of the implants.

Conclusions
This study revealed that TA-loaded PCL implants were successfully manufactured SSE using 3D printing technology. TA was selected as a drug candidate because it is an effective and safe treatment for posterior segment diseases. TA was properly incorporated within the PCL matrix (H-PCL/L-PCL), which was confirmed by SEM. The fabrication technique of TA-loaded PCL implants ensured that the drug was distributed uniformly throughout the system. The FTIR data revealed no chemical reaction between TA and PCL, which is also supported by the thermal analysis results of TGA and DSC. The development of biodegradable ocular implants capable of providing a sustained release of a corticosteroid drug, such as TA, for at least 6 months, marks a major step forward in the fight against posterior segment ocular diseases such as diabetic retinopathy. In addition, the biocompatibility study showed that the TA-loaded PCL implants were safe for ocular application, since all the formulations exhibited percentages of cell viability above 90%. Thus, TA in combination with H-PCL and L-PCL did not compromise the cell viability of ARPE-19 cells. These TA-loaded PCL implants are a valuable alternative to frequent intravitreal injections for treating posterior segment eye disease (PSED). These results thus suggest that SSE 3D printing technology can be successfully used to manufacture the abovementioned ocular implants and has great potential to be transferred to clinical applications. Furthermore, this work has shown that 3D printing technology can be used to precisely produce implants with modifications in shape and dose, allowing ocular implants to be personalised to the individual needs of each patient.