Stereolithography 3D printed implants: a preliminary investigation as potential local drug delivery systems to the ear

The current study is a preliminary investigation on the use of stereolithography 3D printing technology in the field of personalized medicines and specifically for delivering drugs locally, which can for example usefully be applied to ear infections. The main aim is the development of drug-loaded implants for the treatment of ear diseases, to improve patient compliance and to overcome the limitations of current delivery approaches. Multiple prototypes of implant geometries have been created and printed using a flexible resin containing 0.5% w/v of Levofloxacin. Physicochemical characterization of the printed implants was carried out using a variety of techniques (e.g., microscopic, spectroscopic, and mechanical analysis). Finally, preliminary in vitro tests were performed to evaluate the release profile of Levofloxacin, the prototype implant ’ s stability, and their antimicrobial property. The results obtained show that there is no interaction between the resin and the drug, which is perfectly solubilized in the device. In addition, the results of the mechanical tests show that the material used resists compression without compromising the design itself, and the diffusion test has shown that the drug diffused through the matrix prototype at 50% over 3 weeks. The selected designs showed higher antimicrobial activity on E. coli than on S. aureus .


Introduction
The idea of developing new medical implants arises from the need to find valid alternatives to traditional treatments that for years are putting patient compliance in second place.For this reason, a new well-studied technique, 3D printing (3DP), combined with flexible materials has been investigated as a powerful platform for manufacturing medical implants with anatomical models.
Human ear has a complicated anatomical feature, as shown in the scheme of Fig. 1, which is compartmentalised in the external ear, middle ear, inner ear.The external ear, called the auricle, is the visible flap of skin on each side of the head with the external auditory meatus, an about 3 cm channel leading to tympanic membrane.The middle ear lays between the tympanic membrane and the oval window, and it is composed by: ear ossicles, namely the hammer (malleus), the anvil (incus) and the stirrup (stapes), and the Eustachian tube, which in turn communicate with the nasopharynx.Middle ear's main function is to transmit sound waves from the outer ear to the inner ear.The inner ear is located next to the middle ear in the skull bone, and it has two important roles in the functions of hearing and balance (Fig. 1).It comprises two main parts: cochlea, which is the hearing portion, and the semicircular canals, known as the ear labyrinth, which are the balance portion (Sundar, Chowdhury and Kamarthi 2021).
Middle ear infections comprise infection and inflammation of the tympanic cavity, which starts medially of the tympanic membrane.Acute otitis media (AOM) originates here and within 21 days a process of acute inflammation occurs in the tympanic cavity.It is the most frequent infection requiring medical care in children under 5 years of age.It is often related to an upper viral respiratory tract infection (known as common cold), which leads to problems with the eustachian tube.When a child is suffering from a respiratory tract infection caused by a virus, it is also common for the eustachian tube to become clogged with fluid or mucus.The first problem that occurs as a result of this is the fact that the liquid is trapped in the middle space and becomes infected (Paul and Moreno 2020).Every year, the cost to society of treating this disease amounts to billions of euros.A recent study identified Streptococcus pneumoniae and Haemophilus influenzae as the leading causes of Otitis Media in children, despite this population being mostly vaccinated with pneumococcal conjugate vaccine (Silva et al., 2022).
The most used therapeutic approach involves the use of antibiotics, whose concentration in the site of infection has to be higher than the minimum inhibitory concentration (MIC) for the pathogen concerned (Marchisio et al. 2019).Amoxicillin is the first choice thanks to its efficacy, safety, low cost, high palatability and antimicrobial spectrum.The addition of clavulanic acid allows the antibiotic to eliminate H. Influenzae and to have an excellent activity against S. Pneumoniae (Marchisio et al. 2019).
Levofloxacine (LFX) is an antibacterial agent, which belongs to the third generation of fluoroquinolone, and it has a wide spectrum antimicrobial activity.LFX is usually used in the therapy of ocular bacterial infection and applied as one drop every 1-2 h for 3 days and then every 4-5 h.In recent years, the use of LFX for AOM in children has been explored and accepted by the medical world (Abdelbary, Abd-Elsalam and Al-Mahallawi 2019).
Noel et al., explored the role of LFX as treatment of infants and young children with recurrent or persistent AOM, focusing their attention on LFX versus amoxicillin/clavulanate.The authors based their trial on the assumption that fluoroquinolones may be safer and effective compared to the use of high doses of amoxicillin/clavunate in the treatment of children with AOM, as reported in the literature (Pichichero et al. 2005).In the trial performed by Noel, 1678 children aged 6 months to <5 years old with AOM caused by S. Pneuominae and H. Influenzae were treated with LFX at 10 mg/kg twice daily for 2-5 days.The rate of clinical success was higher in LFX-treated children (94%) than amoxicillin/ clavunate therapy (90.8%).At the end of trial, LFX was well tolerated by children, and the only problems concerned diarrhea and vomiting (Noel et al. 2008).
Hearing loss is the common endpoint of many inner ear disorders, which can also cause balance dysfunction (e.g., Meniere's disease), benign positional vertigo and labyrinthitis.First approach is systemic administration, and the most used drugs are streptomycin for vertigo and steroids for sudden sensorineural hearing loss (SSNHL).However, by this administration route, side effects were greater than benefits in several cases, (McCall et al. 2010).In example, systemic steroid therapy has been often associated with severe side effects such as hyperglycemia, hypertension and hypokalemia, peptide ulcer disease, osteoporosis and immunosuppression (McCall et al. 2010).
New administration routes have been investigated with the aim to reduce side effects associated with systemic administration and improve drug delivery to the inner ear.In the last years, new approaches have been studied, based on a combination of inner ear pharmacokinetics with new drug delivery technologies.New strategies include drugdevice combination, cell-specific targeting, and manipulation of gene expression or use of stem cells, viral vectors, and gene editing systems (Devare, Gubbels and Raphael 2018).An example of drug delivery devices for ear therapy are the ear cubes developed by Gehrke et al.The implant consists of two components: a cuboid dexamethasone reservoir that will be implanted in the middle ear and a cylinder that will be implanted into the oval window.The drug is released to the cochlea by diffusion through the cylindrical portion of the device and from the cuboid part into and through the oval window (Gehrke et al. 2016).
In recent years, attention has been focused on 3DP, which plays a key role for future improvements in personalized medicine.Indeed, it makes possible the design of drug delivery systems with complex geometries and shapes, and inherently to control the release of drugs (Jacob et al. 2020).This new technology is easier, cheaper, and more efficient than the traditional ones (Menguso Cui 2021).
One of the 3DP technologies used for drug delivery manufacturing is stereolithography (SLA) 3DP, which is based on controlled solidification of a resin by photopolymerization.The process consists in an UV-laser source which illuminates the surface of the resin to photopolymerized it according to a computer-aided-design (CAD), leading to the resin solidification.It is a layer-by-layer process, after the polymerization of the first layer, the platform moves away from the photopolymerizable resin to start polymerizing the next layer until the object is completed (Karakurt et al. 2020).The main advantages related to SLA are high surface finish and resolution, short print times, heat-free printing.Moreover, SLA offers the possibility to produce elaborate shapes that can help to tune controlled drug release by changing geometry (Kyobula et al. 2017).
Vivero-Lopez et al. employed digital light processing 3DP to develop hearing aids loaded with two antibiotics, ciprofloxacin and fluocinolone  acetonide.The devices allowed a sustained release of both drugs for more than a week.Moreover, they showed anti-biofilm properties against P. aeruginosa and S. aureus (Vivero-Lopez et al. 2021).
In a recent study, levofloxacin-loaded implants for inner ear therapy were 3D printed by fused deposition modelling (FDM) (Haddow, Mathew and Lamprou 2021).Three different designs were produced, with one of the 3 proved to be the best in terms of total release of levofloxacin and ease and time of fabrication.However, one of the main drawback of FDM printing technology is the inability to obtain prints with high resolution and good print quality.
This experimental work was a preliminary experimental trial carried out to analyse the potentialities of 3D-printed implants and to be able to evaluate how implant geometry can affect drug release.In the following work, the authors will fix drug release to fit ear infection therapy.3DP allows for the creation of implant geometries customisable to patient anatomy, allowing a more personalised approach to overcome some of the issues due to the complex anatomy of the ear.Levofloxacin is a synthetic broad-spectrum antibacterial agent of the fluoroquinolones class; it is commonly used to treat infections and inflammations of the ear.Kimitsuki et al., have treated 18 patients with 500 mg of LFX for middle ear disease.In only three patients no therapeutic effect was observed, due to the presence of LFX-resistant methicillin-resistant Staphylococcus aureus (MRSA) or Pseudomonas Aeruginosa, while the others 15 patients received therapeutic relief from the middle ear disease (Kimitsuki, Yoshida and Komune 2012).Based on this background, the current study was focused on the development of five different polymer implant geometries loaded with 0.5% w/v LFX.The optimal implant geometry was determined by evaluating printability, mechanical, thermal properties, and diffusion testing.

Materials
Flexible 80A Resin was purchased from Formlabs Inc. (Somerville, Massachusetts, USA), with the following properties: elongation at break 120%, tensile strength (MPa) 8.9, shore hardness 80A, tear strength: 24 kN/m.The resin is a type of UV-cured 3DP resin that mimics the properties of rubber, biocompatible and indicated by the manufacturer for applications in the biomedical field, especially for soft matter 3D printing.
Staphylococcus Aureus NCTC 10788 (S. aureus) and Escherichia Coli NCTC 10418 (E.coli) were kept in cryopreservative beads in 10% glycerol at − 80 • C. The bacterial cultures were performed in Mueller-Hinton broth (MHB) at 37 • C. For microbiology studies, Mueller-Hinton Agar (MHA) plates and Mueller Hinton Soft agar were used.

Preparation of 3D printed implants
The computer-aided-design softwareTinkercad® was used to generate 5 different designs (Fig. 2).The software was chosen due to both its simplicity, ease of use and large number of figures it offers for the designs.
The implants were printed using Form-2 printer from Formlabs (Somerville, Massachusetts, USA).All designs were exported in stereolithography (.stl) format from Tinkercad to the printer's Preform software 3.18 and sent to the printer.The devices were printed with supports on the build platform, selecting "Flexible" as the material setting.Once the printing process was completed, all the printed implant prototypes underwent a post-treatment phase.Supports were removed from the implant using a side cutter, then the implants were placed in a beaker with 2-Propanol, sonicated for 5 min, and finally placed under UV (Form Cure, Formlabs; wavelength 405 nm) for 10 min at 60 • C. The surface area (SA), volume (V), and surface area to volume (SA/V) ratio of final designs were summarized in Table 1.
The empty implants were printed by using the resin directly from the automatic filling tank.For the drug loaded implants, a solution of 150 mL of resin and 0.75 g of LFX (0.5% w/v) was prepared and left under stirring overnight at room temperature, and then introduced to the tank.Open-Mode option was activated to allow disabling the automatic filling of the tank, which may alter the resin-drug ratios.

Table 1
Surface area (SA), volume (V), and surface area to volume (SA/V) ratio of the designs.

Scanning Electron Microscopic (SEM)
The placebo implants, and the LFX loaded ones were analysed using Scanning Electron Microscopic (SEM, Hitachi TM3030 SEM, Tokyo, Japan) in the Energy Dispersive X-Ray (EDX) condition.Images were taken at magnifications of 50X and 100X; analyses were focused on prototype morphology and presence of LFX residues on implant surfaces.

Fourier Transform Infrared (FT-IR) spectroscopy
Fourier Transform Infrared (FT-IR) spectroscopy analysis was performed on LFX powder and physical mixture (150 mL of resin and 0.75 g of LFX, 0.5% w/v).For each spectrum, 64 scans were scanned, between 4000 and 400 cm − 1 at a resolution of 4 cm − 1 .The instrument used was the Nicolet iS50 from Thermo Fisher Scientific (Waltham, Massachusetts, USA).

Differential Scanning Calorimetry (DSC) and Thermogravimetric analysis (TGA)
Differential Scanning Calorimetry (DSC) was used to investigate how resin, LFX and resin + LFX physical mixtures behave when subjected to thermal stress, and whether the two materials can undergo changes in their solid affecting the final product.The analysis was performed using a DSC 214 Polyma from NETZSCH-Gerätebau GmbH (Wolverhampton, UK).
Thermogravimetric Analysis (TGA) was used to investigate the percentage of mass lost in relation to temperature changes and how materials behave once they lose weight.TGA was performed using a Q500 Thermogravimetric analysis (TA instruments, Bellingham, WA, USA).
In both analyses, the scans were run from room temperature to 500 • C, at a speed rate of 10 • C min − 1 under two nitrogen flow rates of 40 and 60 mL min − 1 .

Mechanical testing
This test is performed with Mecmesin MultiTest 25-dV from Mecmesin (Slinfold, West Sussex, UK).Break point test was performed to compare placebo implants to those loaded with LFX.The down-speed employed was 20 mm/s and the up-speed 100 mm/s.Each test was performed with five replicates.The force/displacement curves were recorded for each run.

In vitro LFX diffusion testing
The drug depots should be fitted into the outer ear, like a common earplug or earpiece.Since the study is a preliminary one, the authors carried out diffusion testing in standard conditions, not considering the peculiar in vivo conditions of the ear as an application site.This was done to evaluate the capability of LFX to diffuse through polymeric prototypes.
The test was performed in triplicate to investigate the release profile of LFX from the different implant prototypes.Each LFX loaded implant was placed in a vial containing 4 mL of PBS at pH 7.4 or pH 5.5, this last was used for mimicking an infection.All the vials were stored in static conditions at 37 • C. At scheduled times, 0.5 mL of PBS was withdrawn and replaced with a fresh buffer at the same pH.The withdrawals were analysed at 292 nm wavelength (Domínguez-Robles et al. 2020) using a 7205 UV/Visible Spectrophotometer from Jenway Instruments (Staffordshire, UK).All absorbance values are converted to concentrations using a calibration curve.

Mass loss evaluation
The studies were performed under the same experimental conditions as the drug release studies in section 2.7.Each implant was placed in a vial containing 4 mL of PBS at pH 7.4.At scheduled times (e.g., 2 h, 4 h, 1 d, 2 d, 3 d, 1 w, 2 w and 3 w), the entire PBS was withdrawn and replaced with a fresh buffer and the sample was dripped, dried and weighted.All weights are recorded and compared with the weight at t0 to investigate if there were any changes in weight that would suggest a degradation status.The test was carried out with triplicate samples for each design.

In vitro microbiological tests
All the implants were tested for their inhibitory effect on bacterial cultures of Staphylococcus Aureus NCTC10788 (Gram+) and Escherichia Coli NSM59 (Gram -).Both categories of bacteria are the most common found in ear infections.The analyses were performed following the protocol set from Domìnguez-Robles et al with small modifications (Domínguez-Robles et al. 2020).First, printed scaffolds were sterilised with 70% ethanol and allowed air dry.A saturated culture of S. aureus/ E. coli was prepared by adding the bacteria into flasks containing 100 mL of Mueller-Hinton broth (MHB) and incubated at 37 • C for 24hrs.After 24hrs, 1 mL of a saturated culture of S. aureus/ E. coli was added in 9 mL Mueller Hinton Soft agar tubes and thereafter poured onto a MHA agar plate.Once soft agar was set, the sterilised implant was placed in the centre of the agar plate and left to incubate for 24hrs at 37 • C. Five implants of each design were tested.After 24hrs incubation, plates were removed, and zones of inhibition recorded in mm.For each design one plate without bacteria and implants with or without drug were also made as a blank as well as one plate with each bacterial type and no printed implant as a positive control.

Implants fabrication and characterization
Both placebo implants and implants loaded with 0.5% w/v LFX were successfully printed (Fig. 3), in accordance with the Tinkercad designs (Fig. 2).Due to the resin properties, the implants resulted to be flexible, smooth, and homogeneous.
All implants were designed considering ear anatomy, ear disease or drug release aspect.Fungo, FungoX4 and Rocket were designed to achieve a good stability in the ear, the protruding parts allow keeping the device in the site, avoiding falling out of the ear.Pod and Screw were designed following the ear anatomy with the same aim to achieve stability but using a different approach.If the patient's compliance is considered, it is very important to consider the level of pain the ear disease causes.In the case that the patient's pain is tolerable, Fungo, FungoX4 and Rocket are probably the best choice due to their rigidity and structure, otherwise, when the pain level is very high, Pod or Screw could be the most suitable choice having a more flexible and conformable structure.Moreover, Fungo, FungoX4 and Rocket have a hole that allows patients to hear.Considering the LFX release, for the Pod and Screw designs, the drug released from their outer surface structure (based to SEM evidence) and will come into immediate contact with the ear surface, while for the other three designs (Fungo, FungoX4, and Rocket) there is the risk that the drug released inside the prototype reaches the ear tissue with more difficulty affecting bioavailability.Future studies will be needed to prove this.

SEM analysis
SEM analysis was carried out to provide microscopic analysis of prototype surfaces (Fig. 4).It is observed that all designs have smooth surfaces, and the few residues of LFX were found linked to the resin.LFX was dissolved completely in the resin solution and homogeneously dispersed into printed prototypes.

FT-IR spectroscopy
FT-IR was performed to investigate any potential interactions between the resin and LFX.Spectra were obtained for resin loaded with 0.5% LFX, for LFX powder and for resin (Fig. 5).The range in which most peaks identified are 1800 -500 cm − 1 (Table 2), with no significant differences between the three graphs.The characteristic peaks of levofloxacin (Fig. 5C) were found at 2848 and 1620 cm − 1 corresponding to C-H stretching for alkanes -CH3 and aromatic rings respectively.The characteristic peak for C--O stretching vibration of the COOH group can be seen at 1724 cm − 1 .At 1290 cm − 1 was found the C-F stretching peak.Flexible 80A resin is mainly composed of acrylate monomers and urethane dimethacrylate (Formlabs).At 1722 cm − 1 (Fig. 5A) strong peak of the C--O stretching of α, β-unsaturated esters can be observed.The peaks between 1210 and 1263 cm − 1 , can be attributed to the esters C-O stretching and between 3000 and 2840 cm − 1 to C -H stretching for alkanes.When the resin is loaded with 0.5% LFX (Fig. 5B) the strong peak at 1620 cm − 1 could be attributed to the aromatic ring observed in Fig. 5C for LFX, which disappears in the presence of resin.This may be due to the opening of the aromatic ring.
FT-IR results suggest that there are no significant chemical interactions between LFX and resin probably due to the low concentration of drug used for preparation, as it is supposed by Domínguez-Robles who mixed TPU and 1% LFX (Domínguez-Robles et al. 2020).This is the first time that it has obtained FT-IR spectra from Formlabs Flexible 80A Resin; therefore, there are no previous data in the literature for comparison.

Thermal analysis
The results of DSC analyses performed on the three different compositions: resin loaded with LFX, LFX powder and resin as such are reported in Fig. 6.The DSC curve of pure LFX (Fig. 6B) showed one endothermic peak at 125 • C that could be attributed to its dehydration and one exothermic peak at 229.8 • C representing its melting point.The obtained DSC curves for LFX were consistent with curves reported in the literature (Jalvandi et al. 2016).The DSC curves (Fig. 6) obtained from the resin provided information about resin thermal behaviour.The exothermic peak in curve C, at 176.0 • C, refers to a melting point of crystalline material while the two endothermic peaks show a recrystallization phenomenon (Sickler and Data, 2021).Flexible 80A resin is mainly composed of acrylate monomers and urethane dimethacrylate (Formlabs).An exothermic peak like the one we obtained for Flexible 80A has been previously reported for urethane dimethacrylate alone (Alarcon et al., 2022).The DSC curve of the resin loaded with 0.5% LFX shows a sharp exothermic peak at 183.3 • C. Furthermore, two endothermic peaks are observed at around 360 • C and 425 • C. The exothermic peak observed at 229.8 • C for pure LFX cannot be observed probably due to the small concentration of the drug in the resin.
If TGA is concerned, there are no significant differences between curves A and curve C (Fig. 7).The weight loss remains in a range between 100% and 60% until 400 • C, followed by a vertical decrease from 60% to 0%, in accordance also with the work of Voet et al (Voet et al. 2018).

Mechanical testing
The force applied in Pod (Fig. 8A), grows steadily throughout the duration of the test until it reaches the limit imposed by the test (120 N).Rocket design (Fig. 8B) and Fungo (Fig. 8C), show similar behaviour, after a small peak, the force reaches a plateau phase before growing during the last second of the test reaching the limit (120 N).Screw (Fig. 8D) shows a different trend, there is a plateau phase during the entire duration of the test, with the last stage being similar to the other designs.The only design with completely different behaviour is Fun-goX4 (Fig. 8E); it is possible to see that after a big peak, the force returns to 0 before growing at the end of the test.This behaviour has been explained considering the low flexibility and excessive hardness, provided by the design structure, for this reason this design was discarded.
Fig. 9 shows that there are no significant differences between placebo and drug-loaded prototypes.The load and displacement profile are superimposable; therefore, it can be speculated that LFX addition does not change mechanical properties of resin, confirming the hypotheses (low concentration of drug used) already advanced in the paragraph 3.3 (FT-IR Spectroscopy).

In vitro LFX diffusion testing
In vitro diffusion testing was analysed using a UV spectrophotometer at wavelength 292 nm, at two different pH conditions: pH 7.4 (Fig. 10A) and pH 5.5 (Fig. 10B).An initial diffusion of about 20% was detected in the first week at both pH.The initial diffusion percentages may be due to the presence of LFX drug on the surface of the prints.Starting from the second week, the diffusion percentage reached a plateau for Fungo, Pod and Rocket in both pH and Screw at pH 5.5.The diffusion behaviour in both cases is similar, and this suggests that either prototype structure and pH value doesn't affect the LFX diffusion process.These data are consistent with results of in vitro release study performed by Chen R et al., which analysed the release profile of LFX-hydrochloride from different gels at two pH without noticing any difference in the percentage amount released (Chen et al. 2021).
The diffusion percentages at pH 7.4 were different only for Screw design, which diffused 45% of LFX loaded after three weeks, and the diffusion appears to be steadily increasing, so this suggests that at physiological pH this type of design could continue to diffuse the drug until fully depleted.This diffusion testing has been carried out to evaluate how LFX diffuses through polymer matrix; in order to assess the efficiency of 3D prototypes as local drug delivery system an in vitro release study will be carried out simulating in vivo physiological conditions, mucous and serous middle ear effusions will be used to investigate how the mucous and serous effusions composition affects the LFX release profile.Mucus is thick and viscous fluid, whereas serous is more watery and less thick and contains amylase enzymes whereas mucus contains little or no enzymes (Val et al. 2018).In vitro release profiles will be performed using a diffusion cell and a synthetic membrane to simulate in vivo physiological conditions.The concentration of LFX in the acceptor compartment and entrapped into the synthetic membrane will be measured by HPLC.

Mass loss evaluation
The aim of this study was to demonstrate the device stability simulating their conditions of use.The study was performed soaking the implants in static condition (37 • C) in PBS at pH 7.4.All the designs  show the same behaviour (Fig. 11), after an increase in weight in the first hours, the implant weight stabilizes at an intermediate value, and keeps stable until the end of test.

In vitro microbiological study
Based on the Mechanical Test (Section 3.5) and in vitro Drug Release (Section 3.6) studies, only the Fungo and Screw were chosen for the final studies.Screw design was the most flexible and promoted the fastest drug release.The Fungo design was the most rigid implant.
As shown in Fig. 12, there was no major difference between the zones of inhibition of S. aureus determined by the two different designs.Therefore, it can be inferred that a modification of the design did not have a significant impact on the zones of inhibition produced.However, it can be observed that the diameter of the zones of inhibition of E. coli were higher than those produced in S. aureus for both the designs.

Conclusions and future perspectives
This study demonstrated the importance and ability of SLA 3DP technology in the field of personalized medicine.This technology is a key factor for future improvements in personalized medicine, and medicine has the potential to both design and control different dosage forms with complex geometries and shapes (Jacob et al. 2020), trying to find the best one for the patient.This study demonstrates the potential of using SLA 3DP as a method to create medical devices loaded with drugs    for local ear therapy.Five different designs (Fungo, FungoX4, Pod, Rocket, and Screw) were produced by CAD software.In this study, it has been decided to use Flexible 80A Resin loaded with 0.5% Levofloxacin (LFX), which is a good antimicrobial drug against bacteria which cause ear disease.LFX has been solubilized completely in the device as demonstrated by the images of SEM and it doesn't affect the properties of the resin thanks to the results from TGA, DSC and Mechanical tests.The aim of the project was to achieve the diffusion of LFX from 3D prototypes in a wide time window to be used as potential dosage form, and it was possible to demonstrate that a good amount of LFX, between 35 and 50%, diffused in 3 weeks from all the designs.The method described can be applied to various medical devices linked to different pathologies loaded with different drugs, and the goal is that personalized medicine, thanks to these techniques, can become common practice in the clinical field.These data will be the starting point for further investigation more specifically addressed to an in vitro model using cell diffusion apparatus combined with a synthetic tympanic membrane, and eventually in vivo model.
As future work could also be interesting the results that we obtained with SLA-3DP with same devices printed to be compared with another type of vat-photopolymerization 3DP technology, such as digital light processing (DLP) where a digital micro mirror device guides the polymerization of an entire layer of photosensitive resin at time instead of being restricted to a single spot like SLA.DLP technology could allow a faster printing and a better corresponding of the final products to the CAD models (Zhang et al. 2020).Finally, photo-curable bio-resins can also be tested as substitutive to the commercially available resins (Ng et al. 2020).

Availability of data and materials
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

A
.Triacca et al.

Fig. 9 .
Fig. 9. Mechanical test performed on placebo (white) and drug-loaded (grey) prototypes.The graphs considered Force and Displacement in relation to the examined design.

Fig. 11 .
Fig. 11.Mass loss of the four design placebo implants (Fungo, Pod, Rocket and Screw) in static conditions.Standard deviation is <0.14.

Fig. 12 .
Fig. 12. Diameter of the zone of inhibition of S. aureus (A) and E. coli (B) for Fungo and Screw prototypes containing 0.5% w/w levofloxacin.