Radiofrequency-Triggered Release for On-Demand Delivery of Therapeutics From Titania Nanotube Drug-Eluting Implants

Manpreet Bariana; Moom Sinn Aw; Eli Moore; Nicolas H Voelcker; Dusan Losic


Nanomedicine. 2014;9(8):1263-1275. 

In This Article

Results & Discussion

Characterization of TNT Structures

Typical structures of prepared TNT arrays fabricated on a Ti foil used as the drug-releasing implant are shown in the SEM micrographs in Figure 3. Figure 3A shows the cross-sectional view of the TNT layer with the thickness of the TNT (nanotube length) measured at approximately 50 µm. A complete TNT–Ti sample prepared on Ti foil (1.2 × 1.2 cm) is shown in the inset of Figure 3A. Figure 3B & C shows the top surface of TNTs with open pores and diameters measuring 120 ± 20 nm, and are uniform, well-defined, highly ordered and equally distributed on the surface of the whole TNT. Figure 3D presents the bottom layer of the nanotubes upon their detachment from the flat Ti foil, showing that the nanotubes are closed at this end. Interestingly, the image contains an individual broken nanotube from the bottom part, and reveals the slightly conical shape of nanotube structures with a significant reduction of pore diameters (<30 nm) at their bottom compared with their top part. This morphology is explained by an extensive dissolution of the oxide layer on the top part during the anodization process. These presented SEM images of prepared TNT structures confirmed that the nanotubes were self-assembled, vertically aligned in a geometrically close-packed and highly organized formation, showing TNT as an ideal platform for the storage and release of drugs and drug carriers.[14] Additionally, the dimension, (i.e., the thickness of the TNT), can be precisely controlled by selecting the appropriate voltage and anodization time, depending on the amount of drug or drug carrier required for loading. It is also worth noting that the TNT layers are not restricted to flat substrates, because they can also be prepared on curved or 3D Ti and Ti alloy surfaces, including existing orthopedic implants, plates, wires, needles, stents and screws.

Figure 3.

Scanning electron micrograph images of prepared titania nanotube samples. Scanning electron micrograph images of prepared titania nanotube arrays fabricated on titanium foil by electrochemical anodization. (A) The cross section (actual whole sample is shown in inset); (B & C) the top surface with open pores; and (D) the bottom surface showing closed nanotubes.

Characterization of TPGS Micelles & AuNPs

The hydrodynamic size of TPGS micelles determined by DLS before and after indomethacin loading is shown in Figure 4A. The average diameter of the prepared micelles was approximately 15 ± 5 nm. It was slightly increased to 20 ± 3 nm when indomethacin was encapsulated within their hydrophobic core. Micelle cores fully occupied with the drug were separated via dialysis, and hence the mean particle size for the micelles was largely independent of the micelles concentration or drug to micelle ratio.[46] The size of prepared drug-encapsulated polymer micelles was monitored over 3 months (stored at 1°C) and their size was found to be stable, confirming their good drug carrier capability. A decrease in free energy of the polymeric colloidal system due to the removal of hydrophobic molecular components (i.e., excessive drugs) from the aqueous surroundings, together with the formation of a micelle core stabilized with hydrophilic blocks (micellar outer shell) exposed to water is the major driving force behind the self-association and stability of the micellar structure.[47] TPGS micelles have a negative ζ-potential of -20 mV; indomethacin = -15 mV, while indomethacin-loaded TPGS = -22.5 mV. Hydrophobicity of water-insoluble drugs, such as indomethacin in this case, is determined by surface free energy measured by ζ-potential, which could reflect drug concentration change during drug elution. In addition, it is useful to determine surface electric charge of TNT based on intermolecular interactions in order to assess drug adsorption between the material surface and drugs, and polymers and drugs via measurement of the ζ-potential.

Figure 4.

Size distribution characterization. (A) Size distribution curves of TPGS micelles with and without indomethacin drug loading; and (B) size distribution curve of AuNPs obtained by dynamic laser light scattering measurements.
AuNP: Gold nanoparticle; TPGS: D-α-tocopheryl PEG succinate 1000.

To enhance therapeutic effectiveness in vivo, the ζ-potential model is able to show the effects of drug attachment and internalization of the drug carrier (i.e., micelles in this case) to estimate different cell endocytosis capacities via changes in the surface charge of cells. Effects of solvation interactions or drug uptake from micelles on TNT relates to changes in ζ-potential. For AuNPs, particle size analysis is shown in Figure 4B. The average particle size was found to be approximately 25 ± 5 nm in diameter, which is similar to the size of drug carrier, but significantly higher compared with the free drug (~2 nm for indomethacin). The larger sized AuNPs are used, assuming they can transmit more heat and more effectively remove drug and drug carriers from the nanotubes.

Drug Loading Characterization of TNTs by TGA

To quantify the amount of drug and drug-loaded micelles with and without AuNPs inside the TNT–Ti implants, TGA showed a single stepwise decrease of weight loss for both drug in TNT and drug-loaded micelles in TNT samples (data not shown). The decomposition range/peak for both TPGS micelles and indomethacin was between 200 and 375°C, with TPGS showing a wider decomposition range and slightly quicker weight loss than pure indomethacin. For the indomethacin-loaded TPGS sample, owing to overlapping of the vaporization temperatures between the polymer and drug, they could not be detected separately. The loading was determined to be 25 mg for the drug and 20 mg for the drug-loaded TPGS micelles, respectively (Table 1), confirming 21–24 wt% of loading capacity for the TNTs, which indicates a great deal of space in the TNT structures to accommodate them. This high loading value was attributed to the high aspect ratio of the TNT and very high surface area:volume ratio. The weight loss graphs for bare TNTs (without any loaded drug/micelles) and AuNP-loaded TNTs showed no change in the TGA curves over the temperature range in the experiment. This proves the high inertness and robustness of these nanomaterials, and stability of the AuNPs as their boiling temperature is extremely high and greatly beyond the TGA range (>800°C).

RF-triggered Release of Drug & Drug Carriers From TNTs

Figure 5A & B shows the in vitro release graphs of indomethacin and indomethacin-loaded TPGS micelles from TNT–Ti samples previously loaded with AuNPs using RF-triggered release compared with the non-RF-triggered condition (free diffusion; control sample). These graphs present both mass (mg; left, y-axis) and cumulative percentage (%; right, y-axis) release with respect to time from the TNT, under 5 min RF exposure performed at the initial stage of drug release. This experiment was specifically conducted to probe the efficiency of RF as a triggering source, alongside AuNPs as an effective energy transducer, to examine the effect of this noninvasive stimulation on the release rate and its impact compared with normal release by diffusion (natural, no trigger) that is not related to RF. To distinguish and compare these release profiles, we performed a control experiment showing the usual burst release graphs without any RF. We also included another control experiment using drug and drug carriers in TNT–Ti without AuNPs to demonstrate release that was caused by RF heating of the TNT–Ti.

Figure 5.

Drug and drug carrier-releasing profiles using radiofrequency. (A & B) Cumulative release graphs (% and mg) of RF-triggered release (5 min) of drug (indomethacin) and drug carrier (polymer micelles, D-α-tocopheryl PEG succinate 1000, loaded with indomethacin) from titania nanotube-titanium compared with nontriggered release (free diffusion) and AuNP-free samples used as control. (C) Bar graphs showing comparison between cumulative release for drug and drug carriers using a RF trigger and no RF trigger. (D) The summarized release rate (μg/min) versus time graphs for different conditions (with or without AuNP, and with or without RF trigger) and samples (drug or drug-loaded micelles). Results are presented as the mean ± standard deviation of three independent repeats.
AuNP: Gold nanoparticle; RF: Radiofrequency.

When RF was applied (for 5 min) to the TNT samples loaded with drug and drug carriers in the presence of AuNPs, an immediate and near-complete release (80–100%) was achieved within 2–4 h after exposure. Figure 5C shows comparative graphs with no triggered release for both cases, confirming the efficacy of RF-triggered release induced by the heating of the AuNPs loaded at the bottom of the TNT structures. The energy coupling between emitted RF and AuNPs can explain the observed Joule heating effect. This, in turn, rapidly transmits thermal energy into the surrounding solution, increasing the temperature of the solution in the entire nanotube. Therefore, the increase in the thermal energy within the system results in increased flux of drug and drug carrier from the nanocavities of the TNT to the bulk. The RF-induced heating generates temperature gradients that lead to convective fluid flows or convection phenomena and cause drug molecules to be displaced from the TNT. These results confirm that the AuNPs can transfer energy from RF to stimulate drug release by heating and cause faster drug release by convection. A difference in release was observed for the drug, which showed faster release (64 ± 4% after 1 h) compared with drug-loaded micelles (48 ± 5% after 1 h) with slower RF-triggered release. This is expected, considering the difference in their size (2 vs 20 nm), and the resulting difference in their mobility and surface chemistry. Control samples without AuNPs and without a RF trigger showed considerably lower release, as compared with the RF-induced samples, as they did not show any acceleration in drug release. Their unforced release process occurred via the mechanism of diffusion from the top TNT surface at the initial stage of release (burst). This was followed by sustained release over a longer period of approximately 2 weeks (data not shown). This is a consequence of the extremely long and deep nanotubes acting as reservoirs. Very small increases in the rate of drug and drug carrier release in the system without AuNPs under RF exposure were observed, which can be explained by slight RF heating of the TNT–Ti substrate.

Release rate graphs (µg/min) for all samples are summarized in Figure 5D to present the magnitude of drug and drug carriers released over the respective time frames using different conditions. The decline in release rate ranging from 180–250 to 50–100 µg/min over 3 h was observed after 5 min of RF triggering. This trend is evidently the result of faster release from the drug and drug carriers located on the top of the TNT surface than those residing inside the nanotubes, which is a reason for the slight decrease in rate until 100% release was finally achieved. From the cumulative release and release rate graphs, it is important to conclude that a very high local concentration of drug dosing in a short time can be achieved by RF triggering, and that RF generates a fast response to effectively induce AuNPs of triggered release for both drugs and drug carriers. It is worth noting that this concentration is considerably higher than the local therapeutic concentration that can be reached by oral drug administration in the form of tablets (e.g., due to high transit time, nonspecific absorbtion and metabolism by intestinal wall and liver).[48] In addition, achieving this release rate and high local concentration is particularly important for cancer treatments, whereby many other types of water-insoluble cancer drugs can be applied in this DDS, because these concentrations of toxic anticancer drugs cannot be applied in systemic drug delivery without substantial side effects or, in the most severe cases, death. Therefore, this local DDS provides many advantages for the treatment of localized tumors, which can be achieved by using TNT–Ti wires located around cancer cells, as proposed in our previous work.[15,49]

To explore the influence of RF exposure time on the release pattern and kinetics, we applied RF signals for three different durations (2, 5 and 10 min). Figure 6A & B summarizes these results by presenting them as release rate (µg/min) versus time graphs for both the drug and drug carriers, respectively. The cumulative release (%) at the end of the first and second hour of release for different RF exposure times is depicted in Figure 6C. Results show a slight increase in release by increasing the time of RF exposure from 2 to 10 min, but the differences among the three curves is very small. The use of a short RF exposure of 2–5 min is sufficient to achieve up to 70–90% of loaded drug and drug carrier release during the first few hours after the trigger. To achieve 100% release, an increased RF exposure time (10–20 min) can be applied. By contrast, longer release times (3–4 h) are possible using shorter RF exposures (2–5 min). The RF switched-on time and other parameters, including frequency, power, distance and temperature are influencing factors that are expected to provide a broad scope of possibilities for this DDS to offer precise control of therapeutic dosage from implants. This tunability could present multiple options for pharmacists, clinicians, doctors and patients for dynamically changing the specific level of drug release in a short timeframe for emergencies, or to address acute conditions, as well as on-demand targeted local drug administration for tailoring specific applications.

Figure 6.

Drug and drug–micelle release rate, and cumulative release over time with radiofrequency trigger. Release rate (µg/min) profiles under different radiofrequency exposure times, for (A) drug and (B) drugs loaded in a drug carrier. (C) Summarized cumulative release (%) for 1 and 2 h obtained using different durations for radiofrequency exposure (2, 5 and 10 min). Results are presented as the mean ± standard deviation of three independent repeats.

Finally, to demonstrate that RF-triggered release can be performed on-demand at arbitrary time points, we performed a 5-min-long RF-triggered release at day 5 of immersion of TNT–Ti in the PBS solution. Figure 7A & B shows the cumulative release graphs (mass and wt.%) compared with the controls (unforced release) of indomethacin and indomethacin encapsulated in TPGS micelles as a function of time. After 5 days of release, approximately 84% of drug and 45% of drug carriers were released from the TNTs before the RF signal was turned on. When a 5-min-long RF was applied in both cases, 100% release of indomethacin was achieved within 25 min. For drug carriers, 100% release was achieved after 105 min (Figure 7A & B). Again, the release of drug compared with drug carriers was faster as a result of the larger polymer micelle size relative to the smaller drug molecules. These results show that RF can be applied at arbitrary and unsolicited times, even under circumstances when extensive therapy is required during any fixed time point of the life of a drug-eluting implant. Interestingly, UV–Vis spectroscopy and DLS of polymer micelles after release from TNT showed that the micelle structures were still intact with encapsulated drug inside the micellar core. The storage life of this system is not specifically determined in this work. However, the stability of drug-loaded carriers in TNT was determined in our previous work, showing their preserved integrity and no significant changes in size during 2 months of storage. The storage should provide conditions to prevent any potential passive diffusion of drug carriers from nanotubes.[25] The preservation of structural integrity of the polymer micelles after release is important for anticancer therapy when the drug is only released after micellar interaction with cancer cells. Nevertheless, the stimuli-responsive behavior in the case of drug (indomethacin) release may not be useful for biomedical applications that require a low drug dosage, as >30% drug was released during the first 3 h. Under conditions where there is no RF, almost 70% was released in the first day (Figure 7). In oncology, this could strongly limit its application due to the high toxicity of the drugs used. Thus, it is necessary to mainly use drug-loaded carriers, preferably over drug-only loading. In this case, the stimuli-responsive behavior is clear and could be of great value for clinical use. It is important to state that AuNPs are considered as biocompatible and not cytotoxic, although there are some concerning studies about their accumulations and potential toxicity.[33,34,35,38] The potential toxicity issue of AuNPs is outside of the scope of this work and not explored in this study. Considering that less than 10% (compared with the total amount of drug/drug carrier) of AuNPs are loaded inside TNT implants, after release their elimination from the body should not be critical.

Figure 7.

Drug and drug–micelle release profiles using radiofrequency. (A & B) Cumulative release graphs (% and mg) of RF-triggered release (5 min) after 5 days of free-diffusive release of the drug (indomethacin) and drug carrier (D-α-tocopheryl PEG succinate 1000 micelles with indomethacin) from titania nanotube-titania implant compared with nontriggered release by RF. The insets show (zoomed-in) cumulative release graphs with a time scale in minutes after RF exposure. Results are presented as the mean ± standard deviation of three independent repeats.
RF: Radiofrequency.

From the presented results, we can clearly conclude that AuNP-assisted RF stimulation can be successfully used for on-demand and noninvasive triggered release of drug carriers from TNT–Ti implants. The RF-induced AuNP heating is effective in causing accelerated transport of drug carriers from TNT. This localized heating of AuNPs occurs in a manner similar to the phenomenon observed in RF-induced hyperthermia or regional ablation of cancer cells.[38] The power from the RF generator transferred to our samples in solution was merely 20 W, which is considerably lower when compared with 100–800 W radiation generally used in targeted hyperthermia for cancers. Other influencing factors for RF-triggered release, such as RF frequency, amplitude, distance, and the direction and orientation of the trigger source (e.g., copper holder to position samples in our work) can be further optimized to make this process more controllable and efficient. This strategy is particularly valuable for drug-eluting implants in orthopedic, anticancer, wound healing and cardiovascular therapies, where triggered release can be made possible for urgent/critical, immediate and intensive treatment.