Controlled-release systemic delivery - a new concept in cancer chemoprevention

Abstract

Many chemopreventive agents have encountered bioavailability issues in pre-clinical/clinical studies despite high oral doses. We report here a new concept utilizing polycaprolactone implants embedded with test compounds to obtain controlled systemic delivery, circumventing oral bioavailability issues and reducing the total administered dose. Compounds were released from the implants in vitro dose dependently and for long durations (months), which correlated with in vivo release. Polymeric implants of curcumin significantly inhibited tissue DNA adducts following the treatment of rats with benzo[a]pyrene, with the total administered dose being substantially lower than typical oral doses. A comparison of bioavailability of curcumin given by implants showed significantly higher levels of curcumin in the plasma, liver and brain 30 days after treatment compared with the dietary route. Withaferin A implants resulted in a nearly 60% inhibition of lung cancer A549 cell xenografts, but no inhibition occurred when the same total dose was administered intraperitoneally. More than 15 phytochemicals have been tested successfully by this formulation. Together, our data indicate that this novel implant-delivery system circumvents oral bioavailability issues, provides continuous delivery for long durations and lowers the total administered dose, eliciting both chemopreventive/chemotherapeutic activities. This would also allow the assessment of activity of minor constituents and synthetic metabolites, which otherwise remain uninvestigated in vivo.

Introduction

Botanicals have been a great source for complementary and alternative medicine (1). Numerous studies have been reported with isolated phytochemicals from the time the words chemoprophylaxis (2) and chemoprevention (3) were coined by Lee Wattenberg and Michael Sporn et al., respectively. Due to significant advancements in cancer screening and early detection, a large population is available for such interventions. Yet the global cancer incidence rate has barely declined for the majority of cancers (4). This is due to the lack of translation of pre-clinical studies.

Extensive pre-clinical screening has resulted in the identification of a multitude of agents (5,6). However, only limited agents were tested in vivo due to the requirement of large quantities for long-term studies. The majority of the compounds that have been tested failed to elicit expected results, with the central issue being bioavailability (7,8). The norm in pre-clinical studies has been to escalate the dose to achieve effectiveness. The high doses are not readily translatable due to the toxicity concerns, which are generally associated with such doses. In addition, despite spiking the doses, some compounds still demonstrated poor bioavailability. For instance, daily doses as high as 4–12g of curcumin still resulted in low bioavailability (9,10) and limited or no response (11,12).

There are several factors that play a role in limiting the bioavailability, including the solubility of the compound, stability in the gut influenced by gastric and colon pH, metabolism by gut microflora, absorption across the intestinal wall and active efflux mechanism in the intestine, and the liver first-pass metabolism effect and elimination from circulation. Many phenolics, including luteolin, baicalein and diaidzein, flavan-3-ols including tea catechins, and stilbenes like trans-resveratrol are substrates for the efflux transporters that limit their bioavailability (13). For compounds that are not substrates of efflux transporters, the bioavailability is mostly influenced by first-pass metabolism or rapid excretion. Some compounds have demonstrated increased bioavailability by co-administration with piperine, an inhibitor of glucuronidation. Piperine is reported to increase the bioavailability of epigallocatechin gallate by 40–60% in mice (14), enhance the bioavailability of curcumin by 134–154% in rats (15,16) and up to 2000% in humans (16). A caveat to this approach is the long-term effect of continuous use of such modulators on other harmful xenobiotics. Hence, alternative approaches for delivering chemopreventives that circumvent issues related to oral bioavailability are warranted.

Several delivery systems have been developed to overcome the bioavailability issue, including nanoparticles, liposomes, microparticles and implants (reviewed in (17,18)). Multiple factors limit the use of these alternate approaches to deliver the agents. For example, if the nanoparticles are larger than 100nm, they are stuck in the liver and may not reach the target site (19). Liposome delivery is limited to agents with high solubility in pharmaceutically suitable vehicles (20). Different types of implantable devices have been used, such as non-biodegradable silastic tubes, which have encountered issues with their removal after the end of treatment due to fibrous growth over them (21), and high melting-point polymers that limit their application to compounds with high thermal stability (22). Polycaprolactone (PCL) is a FDA-approved material, has a low melting point (~60°C), is biocompatible, biodegradable and can be easily molded to any required shape. It allows the homogenous entrapment of drugs in the polymeric matrix, can result in sustained delivery through drug diffusion into the extracellular matrix and exhibits a long half-life of degradation (23,24). The low melting point of the polymer is essential for implant formulation by the extrusion method in order to avoid the instability of compounds. Furthermore, a diffusion-based PCL matrix implant is an ideal choice for the long-term delivery of chemopreventive agents, unlike surface erosion-based matrix implants, which have a short half-life and may only be suited for short-term therapeutic applications (25).

In this study, we report a systemic controlled delivery system developed for chemopreventive and chemotherapeutic activities using PCL implants in vivo and detailed methodology on the preparation of these implants. The use of this FDA-approved material will provide faster translation from bench-to-bedside.

Materials and methods

Chemicals

The polymeric materials used were obtained from these sources: PCL, mol. wt. 65 000 (P65) and PCL, mol. wt. 15 000 (P15) were from Sigma–Aldrich (St. Louis, MO, USA), and polyethylene glycol, mol. wt. 8000 (PEG-8) from Fisher Scientific (Fair Lawn, NJ, USA). Pluronic^R F68 (F68) was a gift from BASF Corporation (Florham Park, NJ, USA). Silastic tubings of different diameters (2.0 and 3.2mm internal diameter) were purchased from Allied Biomedical (Ventura, CA, USA). Test chemopreventive agents were from the following sources: Curcumin and individual curcuminoids (Sabinsa Corp, East Windsor, NJ, USA; curcumin is typically a mixture of three curcuminoids - 75% curcumin (curcumin I); 20% demethoxycurcumin (curcumin II) and 5% bisdemethoxycurcumin (curcumin III)), resveratrol (98%; Biotivia Bioceuticals LLC, New York, NY), tanshinone II (>95%) and cucurbitacin B (>90%) (PhytoMyco Res. Corp., Greenville, NC, USA), luteolin (≥97%; LKT Laboratories, St. Paul, MN), oltipraz (99%; National Cancer Institute, Bethesda, MD, USA) and diindolylmethane (98%; Sigma–Aldrich, St. Louis, MO, USA). Polyphenone E [Green tea polyphenols (GTPs)] was generously provided by Pharma Foods International Co., LTD (99.8% polyphenols of which 89.1% is catechins; Kyoto, Japan). Punicalagin (99%) from pre-enriched Punica husk (28) and withaferin A (>98%) from Withania somnifera standardized extracts (Sabinsa Corp, East Windsor, NJ, USA) both were isolated in our laboratory. The phosphate-buffered saline (PBS) tablets were from Sigma–Aldrich (St. Louis, MO, USA). Bovine calf serum was from Hyclone (Logan, UT, USA). The dichloromethane (DCM) and absolute ethanol were from BDH chemicals (VWR, West Chester, PA) and Pharmco-AAPER (Louisville, KY, USA), respectively. All other chemicals were of analytical grade.

Formulation of polymeric implants

The lipophilic polymer P65 and hydrophilic polymer F68 (9:1) (2.7g) were dissolved in 6ml DCM. Curcumin (0.3g; 10%) was dissolved in an appropriate solvent that was completely miscible with DCM; the drug load for withaferin A was 3%. The drug load for withaferin A was reduced to 3% due to its much higher antiproliferative activity observed in cell culture studies. After mixing the two solutions in a glass beaker, solvent(s) was removed by initially heating in a water bath at 65–70°C with occasional stirring with a glass rod under a hood, followed by the complete removal of the solvent(s) under reduced pressure using a Savant Speed-Vac (Thermo-Savant, Holbrook, NY). The molten material was filled in a 5 ml plastic syringe attached to 6–10cm silastic tubing (3.2mm dia). After heating this assembly at 70°C in an incubator for 10–15min or more, the assembly was removed from the incubator and the material was extruded. After cooling at room temperature, the implant was removed by cutting the silastic tube longitudinally and the implants were excised into desired sizes. Sham implant formulation was prepared in essentially the same way, with the absence of a chemopreventive agent. Implants were stored in amber vials under argon at 4°C. For in vivo studies, the implants were prepared with a 20% drug load for curcumin and 3% for withaferin A. Structures of select test compounds and photographs of implants are shown in Figure 1.

Fig. 1.

The structures of selected chemopreventive agents investigated (A) and photographs of representative polymeric implants (B). Implants were prepared from polycaprolactone (P65): F68 (9:1) as described in text. Implant size: 2cm length, 3.2mm dia.

To determine that the implant formulation was free from any residual DCM, four 1 cm sham implants were dissolved in 1ml tetrahydrofuran (THF), followed by polymer precipitation by 3vol of methanol. The polymer was removed by centrifugation at 10 000×g for 10min and 1 µl of supernatant was then analyzed by gas chromatography–mass spectrometry (GC–MS) for DCM using THF as the internal standard. A standard curve was generated from 1.6 ppm to 100 ppm DCM spiked in a mixture of 1ml THF and 3ml of methanol. The detection limit of DCM by GC–MS was <1.6 ppm.

In vitro release

To determine the rate of release of test agents, 1 or 2 cm implants were placed in 10ml PBS containing 10% bovine serum in 20 ml amber vials. Penicillin–Streptomycin solution (Invitrogen, Carlsbad, CA) (1%, v/v) was added to the release medium to minimize the growth of microorganisms. The vials were incubated at 37°C with constant agitation in a water bath (Julabo SW 23, Seelback, Germany). The medium was changed every 24h, except wherever stated otherwise. Ethanol (10% final conc.) was added to the release medium to completely solubilize the compound, except for withaferin A, which was extracted using acetonitrile and chloroform. The release was measured spectrophotometrically and the concentration was calculated against the standard curve. The absorbance for curcumin and withaferin A was measured directly at 430 and 215nm, respectively. The absorbance for all other compounds listed in Table I was measured spectrophotometrically at their respective lambda-max. Calibration curves for each compound were generated by spiking PBS containing 10% bovine serum and 10% ethanol with known concentrations of the test compound.

Animal study 1 (In vivo release and stability)

In vivo release. To determine the rate of release of test agents in vivo, 5- to 6-week-old female August-Copenhagen Irish (ACI) rats (Harlan Laboratories, Indianapolis, IN) were acclimated for a week and then grafted subcutaneously with one 2 cm implant (200mg) of curcumin containing a 20% drug load (40mg of test agent). Control rats received sham implants. Implants were grafted at the back of the animals as described elsewhere (29). Animals from both sham- and test agent-treated groups (n = 5) were euthanized by CO₂ asphyxiation at various intervals (1, 2, 3, 5, 8, 18, 36, 42 and 70 weeks). The implants were then recovered, dried under vacuum and stored in amber vials under argon at 4°C until analysis.

Analysis of curcumin in implants. Initial and residual contents of the test agent in implants removed from the rats (n = 3) were measured to determine the average weekly release and cumulative release, as well as to determine their stability. Three implants selected randomly were analyzed to determine the residual amount and stability in the implants. Implants were dissolved in 5ml DCM, diluted with ethanol (1:1), and the solution was directly used for the measurement of curcumin following further dilution with ethanol:DCM (80:20). The absorbance for curcumin was measured spectrophotometrically at 430nm. The concentration of the solution was calculated against the standard curve. A calibration curve was generated by spiking blank solvent mixture with known concentrations of test compound. The amount of daily or weekly release was calculated as follows:

(Initial amount – Residual amount)

Rate of release = ---------------------------------------------

Time in days or weeks

Analysis of stability of compounds in the implants and extraction efficiency. To determine the stability of curcumin in the implants, the compound was extracted from the implants as described above and analyzed along with reference compounds using the Shimadzu HPLC system (Kyoto, Japan) equipped with two LC-10ADvp pumps, an autosampler, fluorescent and diode array detectors and a ShimPack reverse phase column (Shimadzu; 250×4.6mm, 5 μm) operated by class VP (ver 7.4 SP3). An aliquot of the extracted curcumin was analyzed by HPLC using acetonitrile and 1% citric acid (adjusted to pH 2.5) at a flow rate of 1ml/min with a linear gradient elution in which the acetonitrile concentration was increased from 0 to 30% for the first 5min, followed by an increase to 45% from 5 to 20min; the latter ratio was then maintained till 36min. The three curcuminoids were detected at 411 and 500nm as excitation and emission maxima, respectively, in the fluorescence detector and the total curcumin concentration was calculated from standard curves of individual curcuminoids.

The extraction efficiency of curcumin from the implants was tested by extracting an implant with a known weight and a 20% drug load with the solvents as described above and quantitating the amount by both spectrophotometry and HPLC.

Animal study 2 (Bioavailability and tissue distribution)

To determine the bioavailability and tissue distribution of curcumin as a representative compound, 5- to 6-week-old female ACI rats were acclimated for a week and divided into two groups (n = 4). One group received curcumin via the diet at 1000 ppm mixed in AIN-93M diet. The other group received four 2 cm implants (200mg, 20% drug load, 40mg curcumin per implant), two on either side of the back of the rat placed subcutaneously as described previously (29). The high curcumin dose via the implants was given to ascertain that plasma and tissue levels were present at measurable levels. All rats received food and water ad libitum. After 30 days, the rats were euthanized, and liver, brain and blood were collected along with the implants. Plasma was analyzed after pooling (750 µl each) from two rats, thus generating two pools. After adding 200 µl of 0.5M sodium acetate, pH 5.0, the plasma was extracted three times with 3ml ethyl acetate. The pooled extract was dried under a vacuum and the residue was reconstituted in 100 µl of acetonitrile. Liver and brain tissues (400–500mg) from each animal were homogenized in 1.5ml PBS (pH 7.4) containing 200 µl of 0.5M sodium acetate. The homogenate was extracted three times with 2vol of ethyl acetate. After the evaporation of the pooled extracts under the vacuum, the residue was reconstituted in 100 µl acetonitrile. One half of the extract from plasma, liver and brain was analyzed by HPLC using conditions described above. The extraction efficiency for curcumin from plasma and tissue was determined by spiking a known amount followed by extraction as described above and quantitated by HPLC.

Animal study 3 (Influence on DNA adducts)

To determine the effect of curcumin via the implant delivery on benzo[a]pyrene (BP)-induced DNA adducts, 5- to 6-week-old female Sprague–Dawley rats (n = 4) (Harlan Laboratories, Indianapolis, IN) were acclimated and then grafted subcutaneously with sham or curcumin implants (one 1 cm implant, 100mg/implant, drug load 20%). Two weeks later, animals were challenged with a low-dose, continuous release BP implant (1 cm implant, 100mg containing 10% BP). BP implants were prepared as described elsewhere (30). Animals were euthanized 30 days after BP treatment and selected tissues and implants were collected. Curcumin implants were removed and stored under argon at 4°C until analysis. Tissues were stored at −80°C until analysis.

Animal study 4 (Antitumor activity)

To determine if agents delivered by implants elicit a therapeutic response, 5- to 6-week-old female nu/nu mice (Harlan, Indianapolis, IN) were inoculated with human lung cancer A549 cells (2.5×10⁶ cells in 100 µl matrigel) subcutaneously on the right flank. When the tumors reached ~30mm³, the mice were randomly divided into five groups (n = 8 per group). One group received withaferin A (4mg/kg) intraperitoneally on alternate days. Another group received tricaprylin as a vehicle intraperitoneally. The third group received three 1.5 cm implants (2.4mm dia, 60mg, 3% drug load) of withaferin A at three different sites (two on the left and right shoulder blades, and a third on the left flank). The fourth group received three sham implants as described for withaferin A implants. The fifth group received no treatment. The tumor size was measured weekly with a digital caliper, and when the tumor volume reached nearly 400mm³ in the control group (6 weeks), the animals were euthanized and the tumor tissue along with other organs and blood were then collected. The implants were removed for measurement of the residual withaferin A in the implant and for stability studies.

Statistical analyses

Statistical analyses were performed using a one-way analysis of variance or Student’s t-test where a P-value of <0.05 was considered significant. All analyses were carried out using the GraphPad Prizm statistical software, ver. 4.3 (La Jolla, CA).

Results

Preparation of polymeric implants

Of the several polymers or copolymers surveyed, PCL was identified as most appropriate, as it is known to be biocompatible and biodegradable as well as melts at a relatively low temperature (60°C). PCL is FDA approved for human use for drug delivery devices as well as tissue scaffolding and suture, due to its biodegradable properties. DCM was found to be the best solvent to dissolve the polymeric materials used. Curcumin was dissolved in ethanol prior to mixing with the polymer solution. THF can also be used since it readily dissolves many lipophilic chemopreventives, and it is readily miscible with DCM. Once the molten polymeric material was extruded into silastic tubing, the implants were removed from the mold by cutting the silastic tube longitudinally with a scalpel and pulling the implant out. After cutting into their desired lengths (0.5–3cm), implants were weighed and slightly trimmed to equalize their weights (±2%). P65 implants were highly plastic, unlike P15 implants, which were relatively less plastic, indicating that lower mol. wt. implants are more fragile. Polymeric implants were prepared with different thicknesses, with 3.2mm dia for the rat studies whereas thinner implants (2.0mm dia) were used for mice studies. Implants can be prepared with many diameters as silastic tubing is available in varying sizes (1.0, 1.6, 2.0, 2.6 and 3.4mm internal diameter). Once mixed, solutions of polymers and test agents can also be freed from most of the solvent(s) by simply pouring the solution in a glass Petri dish and keeping it under a closed hood (instead of removing the solvent in a water bath at 70°C), followed by the removal of the residual solvents under vacuum. Addition of water-soluble F68 or PEG polymer to the PCL facilitates the formation of micro pores for the extracellular fluid after the water-soluble polymer dissipates in a few days after implantation. Since DCM is a chlorinated solvent, we measured the residual amount of the solvent in the implants by GC–MS. From the four 1 cm implants measured, we did not detect any residual DCM, with our detection limit of <1.6 ppm. Thus, from the safety point of view, we are well below the FDA limit for chlorinated solvents (30 ppm), even if multiple implants were to be used.

In vitro release from polymeric implants

The compounds were embedded in a highly lipophilic polymer matrix, P65, containing 10% F68, a hydrophilic polymer. The presence of F68 in the matrix reduced the overall viscosity thus easing extrusion of the matrix to form cylindrical implants. When 2 cm curcumin implants were shaken in PBS at 37°C, little or no release was observed. However, the addition of a 10% serum in the release medium to simulate the in vivo extracellular fluid conditions greatly facilitated the release. In fact, the release of curcumin was evident within 1h of incubation, and when the medium was replaced after 1, 4, 6, 12 and 24h, a total of 1.6% of curcumin was released (Figure 2A). Subsequent releases were measured every 24h. The release kinetics followed a two-stage process - an initial burst release followed by a more steady release. Data presented in Figure 2A also show that withaferin A was released continuously. However, significant differences were found in the relative rates of daily release, particularly in the initial phase. For example, the percent release on day 1 was high for withaferin A (>7%) compared with curcumin (1.6%). It may be noted that despite its only very slight aqueous solubility, the maximum release of withaferin A was during the initial phase. We have not yet determined if this relatively high release of withaferin A is due to its lower drug load compared with the other compounds (3% versus 10%). Withaferin A was used at a lower drug load due to its high antiproliferative activity observed in cell culture (data not shown). Furthermore, both withaferin A and curcumin are lipophilic but the cumulative release of withaferin A was several fold higher than curcumin (Figure 2A).

Fig. 2A.

Average in vitro daily release and cumulative release of curcumin (A1) (1cm length, 3.2mm dia) and withaferin A (A2) (1cm length, 2.4mm dia) from polycaprolactone (P65):F68 (9:1) implants. The drug load was 20% for curcumin and 3% for withaferin A. The release was measured by incubating implants in a shaker incubator in PBS supplemented with 10% bovine serum and the medium was changed daily for withaferin A and after 1, 4, 6, 12 and 24h, and daily thereafter for curcumin. Data represent an average of three implants. SD was 5–10% (not shown). Detailed release conditions are described in Materials and methods.

Other chemopreventive agents tested for implant formulation and in vitro release include GTPs, punicalagin, resveratrol, tanshinone II, cucurbitacin B, luteolin, oltipraz and diindolylmethane. Table I describes their average daily release and cumulative release for up to 20 days. These agents vary greatly in their lipophilicity. As noted for withaferin A, the other two terpenoids, tanshinone II (≈25% released after 10 days) and cucurbitacin B (59% released after 10 days) also showed high cumulative release. On the other hand, oltipraz, which contains a thione group, showed a much lower release (≈6% after 20 days). It is noteworthy that resveratrol, which is known for its low oral bioavailability and limited chemopreventive efficacy (31), was released in an appreciable quantity (>40% in 20 days). The data presented in Table I represent the average daily release from three implants processed in parallel. The SD was calculated from the daily release measured on different days over the period of analysis indicated in the table.

Implant formulation for several other compounds was made successfully. Of these, lycopene, tocopherols and tocotrienols are the most lipophilic compounds with long aliphatic chains. The preliminary in vitro release with lycopene implants showed essentially no release initially, but then showed a cumulative release of about 6% after 25 days. Over half of the agents listed in Table 1 were of the same diameter (3.2mm) although the lengths varied 1 or 2cm. The remaining agents (withaferin A, tanshinone II, cucurbitacin B and luteolin), being more potent in cell culture studies, were formulated using thinner molds (1.9 or 2.4mm dia) and were intended for use in a mouse model for their therapeutic activity. The amount of release of agents increases proportionately with the surface area of the implant, as shown for curcumin implants (26). However, the percent release is largely independent of the surface area. So, the large differences observed for various agents in Table I are due to the chemical nature of the compounds. Compounds were measured at the lambda-max specific for each compound, confirmed by a spectral scan and quantitated against the standard curve from standards dissolved in the same media. All compounds tested were stable under the formulation conditions (70°C) based on a spectrophotometric scan from compounds extracted from unused implants. Heat-labile compounds that degrade at 70°C cannot be formulated by the method described. Alternate methods are under development to accommodate such compounds.

Table I.

Chemopreventive agents tested for implant formulation and in vitro release

Compound	Dimensions (L×I.D; cm × mm)	Implant weight (mg)	Drug (mg)	Drug load (%)	Cumulative release (%)	Avg. daily release (µg ± SD a)
Curcumin	1×3.2	100	20	20	09.8 % in 20 d	098±2
Green tea polyphenols (Poly E)	2×3.2	100	20	20	31.2 % in 20 d	623±8
Punicalagin	1×3.2	100	20	20	12.0 % in 20 d	120±27
Resveratrol	2×3.2	200	40	20	42.1 % in 20 d	842±73
Withaferin A	1×2.4	50	1.5	3	34.7 % in 20 d	026±2
Tanshinone II	1×1.9	35	3.5	10	25.7 % in 10 d	090±2
Cucurbitacin B	1×1.9	35	0.7	2	47.1 % in 10 d	033±5
Luteolin	1×1.9	35	3.5	10	11.4 % in 20 d	020±2
Oltipraz	2×3.2	200	20	10	05.9 % in 20 d	059±0.4
Diindolylmethane	1×3.2	100	20	20	13.4 % in 10 d	268±13

aThe average daily release was from three implants. The SD was calculated from the daily release measured on different days over the period of analysis indicated. Other compounds for which implant formulations have been made successfully are the following: β-carotene, co-Q10, curcumin I, demethoxycurcumin, bisdemethoxycurcumin, diadzein, ellagic acid, equol, genistein, indole-3-carbinol, lupeol, lycopene, plumbagin, tocopherols and tocotrienols; with the exception of lycopene, these compounds are not yet tested for their release in vitro or in vivo.

In vivo release and stability of compounds in the implant formulation

Unlike the in vitro release, which is based on spectrophotometric measurements of the release medium, the rate of in vivo release was based on the measurement of the residual amount in the implant. The extraction efficiency of the compounds from implants was determined from unused implants. The compounds were extracted after dissolving the polymeric implant and solvent extraction. Curcumin was found to be extracted at ~95% efficiency from the implants whereas withaferin A was extracted with an efficiency of ~91%. To determine the rate of release of curcumin, female ACI rats, grafted with one 2 cm implant of curcumin (20% load), were euthanized at different intervals (1, 2, 3, 5, 8, 18, 36, 42 and 70 weeks). The measurement of the residual amount in the respective implants showed that the weekly release after 1 week was 11.7% and 18.4% after 2 weeks, but later on the weekly release decreased to 3.0–5.2% by 3–8 weeks. Curcumin continued to be released at a weekly rate of 0.7–0.9% during 18–36 weeks. Curcumin was still found to be present in the implants after 70 weeks. The weekly release in vivo was calculated by subtracting the residual amount in the implants from the original amount divided by the number of weeks. The cumulative release at the end of 70 weeks was 72.1%, indicating that the compound can be released for an even longer period of time (2 years or more).

To determine the stability of the compounds during the formulation, as well as in vivo, compounds extracted from implants collected at various intervals were analyzed by HPLC along with reference compounds. The chromatograms presented in Figure 3 indicate that curcumin was completely stable during the formulation as well as after 70 weeks of grafting in the rats. The relative levels of the three curcuminoids—curcumin I (curcumin, 75%), curcumin II (demethoxycurcumin; 20%) and curcumin III (bisdemethoxycurcumin; 5%)—are evident from the HPLC-UV profile. However, the relative peak heights in the fluorescence mode (Figure 3A - inset) are distinct from the UV mode since curcumin III is most fluorescent, followed by curcumin II and curcumin I. The comparison of the fluorescence profiles of reference curcumin versus the one extracted from implants after 70 weeks (Figure 3A versus C) showed that curcumin III may have been released at higher rates compared with curcumin I and curcumin II. In contrast to curcumin, implants of the other two compounds tested, GTPs and punicalagin, showed additional peaks 4 months after implantation, indicating the partial conversion to degradation products. The main additional peak from punicalagin was identified as ellagic acid, a known bioactive metabolite; the additional peaks observed from GTP implants remain unidentified (data not shown).

Fig. 3.

The stability of curcumin during the implant formulation and in vivo studies. Shown are HPLC chromatograms of reference curcumin (A) and curcumin extracted from polycaprolactone (P65): F68 (9:1) implants at time zero (B) and 36 weeks (C) after grafting in ACI rats. The main trace is from the UV detector (430nm) and the insets represent chromatogram from fluorescence detector (Ex. 411nm; Em. 500nm). Peaks I, II and III represent curcumin I, curcumin II and curcumin III, respectively. Experimental conditions were described in Materials and methods.

To determine the effect of the storage of implants, curcumin implants were stored at 4°C, 22°C and 40°C and analyzed at different time intervals for the stability of curcumin. When analyzed by HPLC, extracts of these implants showed no degradation of curcumin at any of the temperatures tested, even after 1 year of storage (data not shown).

Enhanced tissue and plasma levels of curcumin administered via the implants versus the diet

To demonstrate that the implant-delivery system can enhance bioavailability and reduce the total administered dose, female ACI rats were administered curcumin via the diet (1000 ppm) or P65:PEG-8 (65:35) implants (four 2 cm implants, 200mg/implant, 20% drug load, 40mg curcumin/implant). After 30 days, selected tissues (liver and brain) and blood were collected along with the implants. The analysis of curcumin in the liver, brain and plasma by HPLC in conjunction with fluorescence spectroscopy showed significantly higher curcumin levels in both the liver (Figure 4A) and brain (Figure 4B) in the implant group compared with the diet group. In the plasma, curcumin was readily quantifiable in the implant group, but it was below the limit of quantification in the diet group (Figure 4C). Total curcumin delivered from the implants, as determined by measuring the residual amount from all four implants, was found to be substantially lower (36.6 versus ≈300mg) than the total dietary curcumin based on the daily diet intake (10g diet @1000 ppm curcumin). The extraction efficiency of curcumin from plasma was ~90%, whereas it was ~70% from tissues.

Fig. 4.

Polymeric implants of curcumin (A) inhibit benzo[a]pyrene (BP)-induced lung DNA adducts in female Sprague–Dawley rats. Animals received sham or curcumin implant (one 1 cm implant, 10% load) 2 weeks prior to challenge with BP for 30 days. BP was administered by a polymeric implant (1 cm, 10%). DNA adducts were analyzed by ³²P-post-labeling assay. Data represent an average of 4 or 5 animals ± SD.

Curcumin implants diminish BP-induced tissue DNA adducts

To determine if curcumin administered via the implants will inhibit tissue DNA adducts, female Sprague–Dawley rats were pre-treated with sham or curcumin implants (one 1 cm implant, 100mg/implant, 20% curcumin) for two weeks and then challenged with BP and euthanized after 30 days. The analysis of lung DNA adducts by ³²P-post-labeling showed a qualitatively identical pattern of two major adducts. These adducts have previously been identified as dG derivatives of anti-BP-7,8-diol-9,10-epoxide (32) and 9-hydroxy-BP-3,4-epoxide (32,33). However, animals treated with curcumin implants showed a significant reduction of total adducts by 61.5% (P < 0.00047) (Figure 5). The measurement of residual curcumin in the implants showed that a total of 5.31mg curcumin was delivered over the entire duration, i.e. an average 118 µg daily. This dose is over 30-fold lower than the typical dietary administration of 400 ppm curcumin (34,35). This is the first demonstration indicating that curcumin can inhibit BP-induced DNA adducts in the lung.

Fig. 5.

The liver, brain and plasma levels of curcumin in female ACI rats treated with curcumin via polymeric implants (four 2 cm implants; 20% curcumin load) or diet (1000 ppm) for 30 days. Samples were analyzed by HPLC as described in Materials and methods. Data represent an average of 4 samples ± SD, except for plasma, which was analyzed by pooling samples from two animals, and the data represent the average of two pools.

Withaferin A implants provide significant inhibition of lung tumor xenograft

In this study, we investigated the therapeutic potential of withaferin A by both the implant and i. p. routes against a human lung cancer cell line (A549) using a mouse xenograft model. When P65:F68 implants of withaferin A (3% drug load) prepared by the extrusion method were grafted in a pilot study, the implantation site was found to show significant local reaction, presumably due to the initial burst release of the compound as found in vitro (Figure 2). Therefore, we prepared withaferin A implants by coating P65:F68 blank implants (1.4mm dia) with a solution containing 10% P65 and 0.3% withaferin A. Repeated coating with intermittent drying resulted in ≈2.4mm dia “coated” implants; sham implants were prepared similarly with the absence of the drug. Following inoculation with A549 cells, nu/nu mice were treated with sham or withaferin A implants (three 1.5 cm implants, 2.4mm dia., 60mg/implant, 3% load) subcutaneously. Additional groups were treated intraperitoneally on alternate days with the vehicle (tricaprylin) alone or withaferin A (4mg/kg). The measurement of the tumor volume showed no antitumor activity of withaferin A when given by an i. p. route. However, animals receiving withaferin A via the implants elicited significant inhibition of nearly 60% at the end of 6 weeks (Figure 6). The measurement of the residual compound in the implants showed that the total dose administered via the implants (2.09±0.11mg) was similar to the total amount administered intraperitoneally (2.1mg total). To confirm that the biological effects of withaferin A are not related to any local reaction or systemic effects coming from the implants or implantation process, we had grafted sham implants in the control animals with xenografts. No local toxicity reaction was observed in sham implant-treated animals, and no reduction in xenograft growth was observed when compared with the untreated control. This rules out the possibility of any systemic effects from implants per se.

Fig. 6.

Polymeric implants of withaferin A inhibit human lung cancer A549 cell xenograft in nude mice. Following inoculation with human lung cancer A549 cells (2.5×10⁶ cells), when tumor xenografts grew to over 30mm³, animals were grafted with polymeric implants (three 1.5 cm implants, 2.4mm dia; 3% drug load) (A). Two other groups, following inoculation with tumor cells, were treated intraperitoneally on alternate days with the vehicle or withaferin A (4mg/kg) two days after inoculation (B). Withaferin A implants were prepared by coating >25 layers of 10% PCL and 0.3% withaferin A in dichloromethane with intermittent drying. Control groups received respective sham treatments. Data represent an average ± SD (n = 8).

Discussion

Cancer prevention can be categorized into active and passive prevention. For example, the avoidance of smoking, charred meats and unhealthy foods with increasing fruit and vegetable servings, including healthy activities, is considered passive prevention. Active prevention of cancer by therapeutic prevention, preventive therapy or risk reduction is to actively intervene in the cancer development process before the transformation of a normal cell to tumor cell by using risk-reducing agents (36). Active prevention has been expanded and diversified to three sequential levels, depending on the time of intervention: primary prevention involving healthy high-risk individuals, secondary prevention involving subjects with dysplasia or pre-cancer and tertiary prevention involving patients after therapy (37). The implant strategy is promising in that it is applicable to all stages of active prevention. One time implantation is less cumbersome and will encourage clinicians and patients alike. The adaptation of this concept to clinical studies is not expected to be a major issue, considering that more than a million women have received subcutaneous implants for long-acting reversible contraception (38). The subcutaneous implants were also used for the delivery of testosterone (39) and histrelin, a gonadotropin hormone-releasing hormone agonist for prostate cancer patients (40). Norplant implants—six 3-cm silastic tubing implants—have been used in humans to deliver hormones for various purposes including contraception; these implants were grafted subcutaneously under one arm (41). Thus, it seems possible to use at least six polycaprolactone implants (3 cm or 300mg per implant) under one arm for systemic delivery; if necessary, additional sets of implants could be grafted at multiple sites. The goals and intentions of this article are to demonstrate the systemic release of chemopreventive agents in a controlled manner for long durations, and present a viable alternative to high oral dosing. Over time, the method can be refined to overcome many of its obstacles before the approach transforms from bench-to-bedside in patients or high-risk individuals.

In this report, we demonstrated that small molecules when embedded in the biocompatible, biodegradable polymer, PCL, can be administered with controlled, continuous delivery (“24/7”) for long durations (months to years). The PCL:F68 implants released curcumin even after 16 months, with nearly 28% of the original amount still present in the implants, suggesting that release could continue for many more months or even a year. The amount of curcumin release can be enhanced by increasing the composition of soluble polymer in the implant formulation, i.e. from PCL:F68 (9:1) to PCL:PEG (6.5:3.5) at the expense of a reduced length of release. The extent of release depends upon the nature of the compound. For example, implants of terpenoids (withaferin A, tanshinone II and cucurbitacin B) and resveratrol, all of which showed extensive release (40–60%) initially, may provide release for only a few months. Molecules such as curcumin, GTPs and resveratrol, which have been frequently cited for their low oral bioavailability, can now be administered to achieve substantially higher plasma and tissue levels by manipulating the implant system. Our in vitro sink condition mimics to a great extent the in vivo release kinetics, hence the drug of interest can be tested in vitro and extrapolated for in vivo studies

The PCL implant undergoes bulk erosion and the release profile followed Higuchi-mediated diffusion kinetics (42,43) where extracellular fluid (or release medium) diffuses into the polymer matrix, dissolves the drug and takes it out. Once the compound is systemic, it reaches different tissues and can remain sequestered for different lengths of time depending upon the lipophilicity of the compounds. However, the rate of release is different for different compounds, which are dependent on their partition co-efficient in the release medium. The data derived from a wide variety of compounds, which vary in their lipophilicity, molecular weights and functional groups, reveal that essentially any compound can be loaded in the PCL implants.

Release kinetics follows a two-step process, the initial burst release of the surface-bound drug followed by the slow steady release of the inner matrix-bound drug. This burst release can be controlled by coating the implant with a blank polymer but this release profile can be advantageous since it mimics the loading dose/maintenance dose scheduling of drug kinetics. The initial burst release achieves a higher/required circulating level, which is maintained by the second phase of controlled release. Furthermore, we also hypothesize that the site (tissue) of implantation can accumulate the drug at a higher concentration and act as a “drug depot,” which can also play a role in the controlled release.

Previously, we had demonstrated the effect of polymer molecular weights, drug load, the surface area of implants and the effect of various additives on the in vitro release kinetics of curcumin (26). We had also demonstrated the uniform embedding of curcumin in the polymer matrix by scanning electron microscopy, X-ray diffraction and differential scanning calorimeter studies (26), and provided evidence on the in vivo biocompatibility and safety of these implants (27). While this work was in progress and presented in preliminary forms at the annual conferences, Desai et al. reported PLGA implants embedded with N-acetyl cysteine (44) and 2-methoxy estradiol (45) and showed that these compounds were released in vitro and that the release lasted for more than 28 days. These investigators also showed that black raspberry extract can be embedded in PLGA implants and released in vitro and in vivo 46.

Since the subcutaneous polymeric implants deliver agents systemically and require only small quantities, it opens several new avenues, including the flexibility to embed any small molecule for efficacy assessment. This will lead to the identification of new agents at a rapid pace and allow the efficacy assessment of minor plant constituents or synthetic metabolites, which otherwise remain uninvestigated in vivo as they are unavailable in the large quantities needed for oral administration. The implant system can also be adapted to deliver a multitude of agents tuned to deliver different levels for individual compounds. This is important, since the combination chemoprevention targeting multiple pathways with different agents is essential for the success (47).

Conclusion

Our data clearly support the working hypothesis that compounds delivered by subcutaneous PCL implants enhance bioavailability and tissue levels, lower the total administered dose compared with traditional oral delivery and elicit biological responses.

Fig. 2B.

Rate of in vivo release of curcumin from polycaprolactone (P65): F68 (9:1) implants (2 cm, 20% drug load) grafted subcutaneously in female ACI rats. Animals were euthanized after 1, 2, 3, 5, 8, 18, 36, 42 and 70 weeks and the implants recovered were solvent extracted to measure the residual amount as described in text. The average weekly release was calculated by subtracting the residual amount from the original amount divided by the time in weeks. Data represent an average of three implants (SD ≤ 10%) and are shown as percent weekly/cumulative release.

Glossary

Abbreviations:

ACI

August-Copenhagen Irish rats;

BP

Benzo[a]pyrene

BRCA 1 and 2

Breast cancer type 1 and 2 susceptibility gene;

DCM

Dichloromethane

d; G

Deoxyguanosine

DSC

Differential Scanning Calorimeter;

Ex

Excitation wavelength

Em

Emission wavelength;

F68

Pluronic^R F68

GC

–

MS

Gas Chromatography–Mass Spectroscopy;

GTPs

Green tea polyphenols;

HPLC

High Performance Liquid Chromatography

i.p.

Intraperitoneal;

P15

Polycaprolactone, mol. wt. 15 000

; P65

Polycaprolactone, mol. wt. 65 000

; PBS

Phosphate-buffered saline;

PCL

ε-Polycaprolactone

; PEG-8

Polyethylene glycol, mol. wt. 8000;

THF

Tetrahydrofuran.