3.1. Tuning TAF Release Rates: Surface Area and Wall Thickness
These studies involve a reservoir-style PCL implant (Figure 1
), that can deliver TAF at sustained, zero-order release kinetics. Once inserted subcutaneously, biological fluid from the surrounding environment transports through the PCL membrane into the reservoir and can solubilize TAF. TAF can partition into the PCL and transport passively through the PCL membrane to exit the implant. Transport of a drug through the PCL material is dictated by many parameters, such as the diffusion coefficient and partition coefficient, as described elsewhere [53
]. As an aliphatic polyester, PCL undergoes bulk hydrolysis through random chain scission as water permeates through the polymer [37
]. However, biodegradation of PCL is slow and can require years (e.g., 1–2 years) for complete bioresorption [37
], depending on the starting MW. Because bulk erosion of PCL is slow, the faster process of drug delivery is decoupled from biodegradation, enabling zero-order release profiles of drug from the implant. At this zero-order release profile, the daily drug delivery rates are controlled by various parameters: surface area of the device, thickness of the device wall, polymer properties, and drug formulation. A digital camera image of RTI’s trocar-compatible reservoir-style implant is shown in Figure 1
To evaluate the relationship between release rates and the surface area of the extruded PCL tubes, implants were fabricated with three different surface areas, as generated by varying the implant length: 82 ± 1 mm2
, 311 ± 4 mm2,
and 543 ± 5 mm2
, with an average of 31, 124, and 216 mg of TAF loaded into the implant, respectively. All devices comprised Sigma-PCL with a wall thickness of 100 µm, an OD of 2.5 mm, and a formulation of 2:1 TAF:castor oil. The cumulative release of TAF from the implants were monitored for approximately 30 days, as shown in Figure 2
a. As expected for a membrane-controlled system, the higher surface area results in a higher release rate of TAF from the implant. Furthermore, the linear relationship between daily release rates and surface area supports the mechanism of membrane-controlled release from these implants (Figure 2
b). These results align with a previous report using reservoir-style devices with thinner walls of PCL (8.5 µm) fabricated via solvent film-casting [55
]. In the current study, devices were fabricated using PCL tubes prepared via melt extrusion, which produced thicker walled tubes (between 45–200 µm). Despite the thicker PCL wall and different fabrication approach, these devices also maintained membrane-controlled release in this range of wall thickness, demonstrating the robustness of the PCL-based drug delivery platform. For the remainder of this paper, the cylindrical geometry was fixed at 2.5 mm OD and 40 mm length to accommodate commercially available trocars utilized for contraceptive implants [56
], and the release rates were normalized to the surface area of 314 mm2
The thickness of the implant walls was another attribute that affected release rates of drug. Figure 3
shows the daily release rates of TAF from implants comprising PCL of different wall thicknesses and containing a formulation of 2:1 TAF:castor oil excipient. The release rates of TAF inversely correlated with thickness of the PCL walls: 0.91 ± 0.23 mg/day (45 µm), 0.61 ± 0.09 mg/day (70 µm), 0.29 ± 0.05 mg/day (100 µm), 0.19 ± 0.04 mg/day (150 µm), and 0.15 ± 0.03 mg/day (200 µm). As the wall thickness increased from 45 to 200 µm, the release rates approach a plateau wherein the release rates of TAF show minimal change. Importantly, the daily release rates were calculated over the first 35 days of TAF release from the implants, which included a burst release that is more pronounced in thinner walled implants that results in a higher standard deviation (e.g., 45 µm walled implant). We speculate the burst release may arise from transport of drug into the PCL. This inverse relationship between the thickness of the PCL walls and the release rates of TAF was also demonstrated for thin-walled PCL implants fabricated by a solvent casting approach despite considerable differences in the device processing technique [55
]. To reserve adequate volume in the reservoir for drug load, this study only investigated wall thickness up to 200 µm. Overall, these experiments demonstrate the ability to employ two parameters, surface area or wall thickness, to tailor the release rates of TAF from a reservoir-style implant fabricated with extruded PCL tubes.
3.2. Effects of PCL Properties on Implant Performance
PCL is a semi-crystalline, hydrophobic polymer with biodegradation kinetics that depend on the initial MW, typically occurring in the order of 1–2 years [38
], which supports a LA PrEP implant. In these studies, PCL starting material with two different MWs were selected to potentially support an implant with target duration of 6–12 months: Sigma-PCL (Mn
of 103 kDa) and PC-12 PCL (Mn
of 51 kDa). PCL tubes of different wall thicknesses (70, 100, 200 µm) were extruded with either Sigma-PCL or PC-12 and subsequently filled with a formulation of 2:1 TAF:castor oil. Evaluation of these implants using in-vitro release assays revealed two important concepts (Figure 4
). First, the release rates of drug from the implant depended on the selection of PCL; TAF releases at a higher rate from implants comprising Sigma-PCL as compared to implants comprising PC-12. Interestingly, the influence of PCL type on TAF release rates is minimal in tubes with thicker walls (e.g., 200 µm) versus thinner walls (e.g., 70 µm). Second, Figure 4
also shows that irrespective of the PCL type used to fabricate the implant, the release rates of TAF still scales inversely with wall thickness, as also shown in Figure 3
It is possible that differences in crystallinity between the two types of PCL starting materials could affect the release rates of TAF from the implant. Therefore, to further understand the effect of polymer properties on release rates of drug, extruded tubes comprising PC-12 or Sigma-PCL were evaluated with DSC and XRD. Analysis by DSC showed that all PCL tubes exhibit a melting endotherm with a peak near 60 °C (Figure 5
A, Figure S1
), the characteristic melting temperature (Tm
) of PCL [58
]. However, notable differences in the melting endotherms were also evident, such as a narrower melt transition of PC-12 compared to Sigma-PCL and the presence of a small shoulder peak around 50 °C in Sigma-PCL which was absent in PC-12. Quantitatively, the specific Tm
values also differed; Sigma-PCL showed a slightly higher Tm
compared to PC-12 for all thicknesses of the tube walls (Table 1
and Figure S1
). For each sample, Equation (1) was used to calculate the mass % crystallinity and Equation (2) (Thompson–Gibbs equation) was used to calculate the crystallite sizes. Results in Table 1
show that irrespective of the wall thickness, the crystallite size of PC-12 was slightly lower than the crystallite size of Sigma-PCL. Moreover, the crystallite size of Sigma-PCL slightly varied with different tube thicknesses, whereas PC-12 remained consistent. The % crystallinity was slightly higher in certain cases for PC-12 compared to Sigma-PCL, showing statistically significant differences for extruded tubes with 70 and 200 µm wall thicknesses.
XRD analysis was also performed to further examine the crystallite size of PCL extruded tubes using the Scherrer equation (Equation (3)). Extruded tubes (100 µm wall thickness) fabricated from Sigma-PCL and PC-12 showed similar diffraction patterns that include intense Bragg peaks at 2θ near 21.3° and 23.7°, correlating to diffraction of the (110) and (200) planes of the PCL crystallite, respectively (Figure 5
]. Results from XRD analysis (Table 2
) show that the crystallite sizes of PC-12 were slightly smaller than Sigma-PCL, where Sigma-PCL total crystallite sizes was 25 nm (14.2 + 10.8) and PCL-12 was 23.4 nm (13.2 + 10.2), which also agrees with DSC data. Both techniques used to measure crystallite size indicate a similar order of magnitude from the two PCL types, therefore it is unlikely that crystal size alone was responsible for the differences in drug diffusion kinetics from the materials considered in this study, however the observation that crystallite size increased with tube thickness for Sigma-PCL (as measured by DSC) may play a role in release kinetics.
Taken together, these data highlight the importance in considering properties of drug transport in products comprising semi-crystalline polymers, which contain both amorphous regions amenable to drug transport when above the polymer glass transition temperature (Tg
), and crystalline regions which pose a diffusive mass transport barrier. These data indicate that PCL is an ideal polymer suited for membrane-controlled drug diffusion applications given its material properties and semi-crystalline nature. For example, PCL has a Tg
of −60 °C which allows for drug transport at physiological conditions (37 °C) where the amorphous regions exhibit adequate free volume for passive diffusion of small molecules and fluid driven by concentration gradients. Concurrently, PCL crystals impart structural integrity to the implant and act as a transport barrier which modulate drug diffusion and allow for sustained release of TAF. The DSC and XRD results presented here suggest that crystallite size, quantity of crystallinity, and ultimately polymer free volume within PCL will impact transport properties of TAF through the polymer, as also supported by studies with other systems [62
]. Our results show that extruded tubes with lower MW (PC-12) contain smaller sizes of crystals and slightly higher % crystallinity (statistically significant for 70 and 200 µm tubes, p
= 0.008 and p
= 0.007, respectively) as compared to PCL with higher MW (Sigma-PCL). This suggests that higher degree of crystallinity and smaller crystallites could create a more tortuous path for diffusion of the drug, leading to a lower release rate from the implant. At 37 °C, TAF likely diffuses through the amorphous regions of PCL, where the polymer exhibits greater segmental mobility to facilitate passage of small molecules. The size and quantity of the crystal regions would affect the spatial arrangement and quantity of these amorphous regions, ultimately affecting transport kinetics. These findings are supported by the mathematical relationship between membrane flux through a given area which is inversely proportional to distance traveled (wall thickness) by the constant of mass diffusivity, i.e., Fick’s first law of diffusion. The diffusion constant is a function of temperature, molecular size, and viscosity. For polymers, the viscosity term describes polymer free volume, which is impacted by crystallinity, hence the differences in material properties and resultant release rates were observed here.
In addition to the quantitative differences in polymer physicochemical properties observed in this study, three important qualitative findings are also of note. First, Sigma-PCL exhibited irregularities in the melting endotherm as evidenced by a small but apparent shoulder peak prior to the melt whereas PC-12 did not. It was likely a result of thermal history incurred in Sigma-PCL processing from the manufacturer. Given this consideration, the shoulder peak likely did not represent the PCL crystalline phase and may have contributed to an over-estimation of the crystalline content of the Sigma-PCL. A second difference was noted when comparing the width of the melt transition where PC-12 exhibited a narrower melt endotherm compared to Sigma-PCL, suggesting a tighter distribution of polymer molecular weight comprising the crystalline phase. Finally, another difference between these two grades of material was the crystallite size as a function of tube thicknesses. While PC-12 demonstrated more consistency in crystallite size, Sigma-PCL crystallite size increased with tube thickness, indicating a lack of control on the final properties during processing and perhaps explaining the large variability in release rates at low thicknesses and more consistent release rates at higher thicknesses. Sigma grade PCL also exhibited a steeper decline in release rate with respect to wall thickness, while PC-12 demonstrated a more gradual decline in release rate as a function of thickness. It was hypothesized that the concomitant increase in wall thickness and crystallite size observed with Sigma PCL was responsible for the attenuation in release rate values for the two grades of PCL at the higher wall thicknesses studied here. Taken together, these observations highlight the importance of material choices in the design of drug delivery devices from an engineering and quality control perspective.
3.3. Performance and Fabrication of a LA PCL Implant for Delivery of TAF
The duration of this reservoir-style implant for TAF is dictated by two parameters: the drug quantity within the reservoir and the rate of drug release from the implant. Using selected implant dimensions (2.5 mm OD, 40 mm length), TAF payload within the reservoir for the 2:1 TAF:castor oil formulation was approximately 115 mg for an implant with wall thicknesses of 100 µm. Within these constraints of drug payload, the duration of a single TAF implant for PrEP ultimately depends on the daily drug release required for protection as administered via the subcutaneous route, which is currently unknown. In this manuscript, in-vitro release rates from prototype implants were tailored to the range between approximately 0.2 and 0.8 mg/day from a single device, as suggested from previous animal studies and in-silico modelling set with TFV-DP target concentrations of >48 fmol/106
cells among 500 virtual healthly women [31
Using these dimensional parameters, a batchwise process was developed to fabricate TAF implants from extruded PCL tubes, which entails loading the drug formulation into the cavity of a PCL tube and sealing the ends. Given the low melting temperature of PCL (60 °C), the implant was readily end-sealed by controllably heating PCL into the desired geometry using an in-house customized polymer extruder. Care was given to avoid contamination of the interior walls of the tube near the sealing points, which could hinder the formation of end seals during the melt sealing. This fabrication process was used to generate implants for a six-month in-vitro study to assess the release of TAF. As shown in Figure 6
, implants released TAF at a rate of 0.28 ± 0.06 mg/day over the course of 180 days. After 180 days, approximately 68 mg of TAF remained within the implant, with a chromatographic purity of 89.2 ± 0.8% (Table S1
). The trend of decreased TAF stability over time results from ingress of water into the implant as drug depletes, which, in turn, facilitates hydrolytic degradation of TAF [64
]. We are currently evaluating other formulations of TAF that help optimize stability. The implant maintained structural integrity throughout the 180-day release period in simulated physiological conditions.
To support in vivo use of the implant, gamma irradiation was used to sterilize the implant after fabrication. Since gamma irradiation has been shown to affect the chemical and physical properties of PCL [65
], studies were performed to evaluate its potential effects on the implant performance. Table S2
shows the GPC analysis of PCL, including samples of PCL raw material used for the extrusion process and extruded PCL tubes before and after gamma irradiation. Both PCL types (Sigma-PCL and PC-12) showed a slight decrease in Mn
after gamma irradiation, as expected, but the extrusion process minimally affected the Mn
of PCL. To further ascertain release rates of TAF from the implant after gamma irradiation, in vitro release assays were performed on implants with and without gamma irradiation at dosages between 18–24 kGy. As shown in Table 3
, the release rates were comparable irrespective of treatment with gamma irradiation and the difference in release rates were not statistically significant when comparing non-irradiated and gamma irradiated release rates (p
= 0.27 and p
= 0.42 for Sigma-PCL at 70 and 100 µm, respectively; p
= 0.11 and p
> 0.99 for PC-12 at 70 and 100 µm, respectively).
In summary, the in vitro studies presented in this manuscript demonstrate a new fabrication method to produce reservoir-style PCL implants using a batchwise process of polymer tube extrusion, formulation filling, and implant sealing via melt procedures. Although this fabrication process is not currently conducive to high-throughput manufacture, incremental steps detailed here support future production and implementation efforts. For instance, use of melt extrusion to produce PCL tubes presents an improvement over previously reported thin-film casting techniques, in terms of throughput of tube manufacturing and ultimate robustness of the implant. By increasing the implant wall thickness (e.g., from 70 to 200 µm) using an extrusion method, the implants were more aesthetically pleasing, exhibited greater mechanical sturdiness during handling and therapeutic use, while simultaneously retaining sustained release kinetics of drug. Moreover, the injection end-sealing method presented here produces an implant suitable for commercially available trocars, and is loaded with enough of the drug to last longer than six months; all considerations that were deemed preferable by end-users [67
]. With an eye towards product development needs for clinical implementation, compatibility with applicators already clinically employed offers options to adopt existing procedures to support LA-PrEP rollout.