DEVELOPMENT OF PEDIATRIC ANTI-HIV FORMULATIONS WITH IMPROVED DISSOLUTION CHARACTERISTICS

The human immunodeficiency virus (HIV) impacts up to 37 million people globally. Although not fatal on its own, HIV can develop into acquired immunodeficiency disease (AIDS), in which a person’s immune system becomes compromised. To date, there is no cure for HIV, although many treatment options are available. Despite their effectiveness, these treatments are commonly plagued by their inherent complexity. Factors such as doing regimen, pill burden, and undesirable side effects all contribute to variability in patient compliance, particularly in pediatric populations. Currently, there is no anti-HIV drug product readily available for pediatrics, despite close to 1.8 million children living with HIV. This is partially due to a diverse patient population (ranging from birth till adolescence age) with specific needs for various dosage forms and dosing unit size. In addition, taste preferences and toxicity of excipients and may differ in children compared to adults. In the present study, we aimed to develop a pediatric-friendly formulation for anti-HIV therapeutics. Two protease inhibitors, lopinavir (LPV) and ritonavir (RTV) (commercially available as Kaletra), were chosen as model drugs. Kaletra is a fixeddose combination (FDC) of LPV and RTV (4/1, w/w) in either a tablet or an oral solution form. However, neither of these dosage forms is suitable for children. The tablet is large, and therefore can be difficult to swallow for young children, especially for children under four years who generally cannot swallow tablets. In addition, the excipients used in the tablet formulation have been shown to induce adverse events in a pediatric population. On the other hand, the oral solution contains upwards of 40% ethanol and is not suitable for children. Both of these drugs exhibit very bitter taste profiles, which children are very sensitive to. In addition, both LPV and RTV, are inherently poorly water-soluble and suffer from low bioavailability. In order to develop a pediatric-friendly formulation for FDC of LPV and RTV, it is critical to improve dissolution and palatability of the therapeutics using safe excipient(s). Cyclodextrins (CD) are cyclic oligosaccharides that can form water-soluble complexes with hydrophobic drugs, and potentially enhance solubility and mask taste of the therapeutics. In this study, two CD derivatives, 2-hydroxoypropyl-β-CD (HP-βCD) and 2-HP--CD were investigated. Phase solubility, isothermal titration calorimetry (ITC), nuclear magnetic resonance (NMR) and molecular modelling studies were conducted to determine interactions between them and the two anti-HIV drugs, LPV and RTV. The results showed that complexes can be formed between drug and CD and the optimal complexion ratio of drug/CD is 1:1. The results from each study showed that RTV is capable of forming more stable complexes than LPV, with both types of CD. Stability constant values calculated via phase solubility studies indicated that β-CD formed more stable complexes with the drugs than -CD. However, a different trend was obtained from the NMR and molecular modelling studies, which showed that -CD formed more stable complexes. This suggested that non-inclusion complex formation was favored, which NMR and modelling are less sensitive to detecting, over traditional inclusion complex formation. These studies also showed that the specific interactions that occurred between LPV and CD, and RTV and CD, such as hydrogen bonding and hydrophobic interactions, were different, as each drug has a fundamentally unique molecular structure. Following this interaction analysis, formulation optimization of drug:CD complexes was conducted. The prepared drug:CD complexes were spray dried to obtain a final dry powder formulation. Solid state characterization of the spray-dried complexes was performed to determine physicochemical characteristics such as thermal profile, crystallinity, and morphology. Results showed that the spray-dried complexes did not exhibit a melting temperature, and were comprised of drug in an amorphous state, based on differential scanning calorimetry (DSC), X-ray diffraction (XRD) and polarized light microscopy (PLM) data. In addition, scanning electron microscopy (SEM) images showed that the spray-dried complexes exhibited a corrugated, raisin-like morphology. In vitro dissolution studies showed that RTV in an amorphous state exhibits a faster release profile than crystalline RTV. Spray-dried HPβ-CD/RTV complexes showed the most favorable dissolution profile, as 100% RTV was released in 45 minutes. Unexpectedly, converting LPV from crystalline to amorphous via spray-drying resulted in lower dissolution rate and extent. In addition, spray-dried CD/LPV complexes did not exhibit favorable dissolution characteristics, compared to the physical mixture of LPV, polymer, and CD. Overall, interactions between both drugs and both CDs were characterized, and CD/drug complexes were successfully prepared. Further studies will be conducted to assess taste masking effect and in vivo bioavailability of the prepared drug/CD complexes. In addition, other strategies such as freeze drying and kneading will be investigated in the future to further optimize a suitable formulation with improved dissolution characteristics for LPV.


LIST OF TABLES
The current standard in the treatment of HIV involves the use of antiretroviral therapy (ART). ART is the combination of antiretroviral agents that target different processes in the HIV replication cycle. 6 This type of combination therapy has been 3 shown to be effective in dramatically suppressing viral replication and reducing HIV plasma concentrations to below detectable limits. 7 13 The structures of LPV and RTV are shown in Figure   1.2. LPV is a selective inhibitor of type 1 HIV (HIV-1) protease that works by preventing viral maturation that ultimately results in the spread of infection. 14 Administered alone, LPV has insufficient bioavailability due to extensive metabolism by the cytochrome P450-3A4 (CYP3A4) enzymes, which is the primary class of enzymes present in the liver responsible for metabolizing protease inhibitors. 15 RTV is a protease inhibitor, albeit a much less potent one than LPV, and is capable of inhibiting CYP3A4 enzymes, and therefore aids in boosting the plasma concentration of and increasing the bioavailability of LPV. 16 A recent study investigated the interaction between RTV and CYP3A4 enzymes and concluded that CYP3A4 inactivation results from RTV binding to Lys257 of the CYP3A4 apoprotein. 17 Therefore, formulations containing RTV may require less frequent dosing due to the protease inhibitor boosting effect, and may result in a reduction in side effects. 18

Pediatric Anti-HIV Formulations
Regardless of the disease, developing an oral pediatric formulation is often more complex than developing one for adults, as there are more confounding variables that impact overall performance. It has been shown that children exhibit fundamental pharmacokinetic differences compared to adults, including gastrointestinal permeability, rate of gastric emptying, and surface area available for drug absorption. 19 In addition, a child's size, age, and taste preference can play a key role in determining medication adherence, and ultimately overall effectiveness. 20

Cyclodextrin Complexes
Current challenges involved in the development of anti-HIV pediatric formulations include the need to improve drug solubility and dissolution, improve product stability, and provide taste masking effects. One potential approach aimed at addressing these issues is the development of drug-cyclodextrin complexes. Shown in  Drug-cyclodextrin complex formation arises when a drug is introduced to an aqueous cyclodextrin solution, resulting in the removal of enthalpy-rich water molecules from the hydrophobic cyclodextrin core. As water molecules are removed, non-polar drug molecules are able to maneuver into the center of cyclodextrin cavities, and form molecular interactions including hydrophobic, van der Waals, and electrostatic interactions. 31 No covalent bonds are formed or broken during this process and bound drug molecules inside the cyclodextrin molecules are in dynamic equilibrium with free drug molecules. 35 The simplest and most common type of complexation that can occur results in a binary system, in which only two different types of molecules are used (i.e. drug and cyclodextrin). As shown in Figure 1 Analytical techniques such as nuclear magnetic resonance, molecular modeling, and isothermal calorimetry can be used to help explain the interactions between guest and host compounds. Cyclodextrins have been used in various applications including use as food additives, cosmetics, and pharmaceutical drug carriers. 35 The toxicity of cyclodextrins has been studied in recent years, and studies have shown no apparent toxicity from cyclodextrin when developed as an oral formulation, mainly due to lack of absorption in the GI tract. 37 Native cyclodextrins (without modification) have limited aqueous solubility, and therefore in recent years they have been chemically modified to improve their aqueous solubility. 2-hydroxypropyl-beta-cyclodextrin (HP-β-CD) and 2-hydroxypropyl-gamma-cyclodextrin (HP-γ-CD) are two examples of cyclodextrin derivatives that have enhanced solubility profiles compared to parent cyclodextrins. 38 These modified cyclodextrins may be ideal candidates for pediatric formulations due to their impressive safety profile, as recent studies have shown no apparent toxicity in juvenile or adult rats. 39 Some work has previously been done to investigate techniques to improve the solubility and dissolution profiles of LPV and RTV. A recent study showed that cyclodextrin-LPV complexes can improve the in vitro dissolution profile of LPV, and that more research is needed to utilize their full potential. 40 However, in the described study, dissolution studies were performed under sink conditions using 0.06 M polyoxyethylene 10 lauryl ether, which does not accurately reflect the conditions of the GI tract. Additionally, the study investigated the use of cyclodextrins on the enhancement of LPV alone and did not investigate solubility enhancement strategies for RTV.
Cyclodextrins not only aid in the enhancement of stability and solubility in formulations, but they are also commonly employed for their taste masking abilities. This is significant attribute because a major concern in developing pediatric formulations is the palatability of the formulation. Studies have shown that children have a low tolerance for poor-tasting medications, which may negatively impact patient compliance, and overall medication effectiveness. 20 One study concluded that more than 90% of pediatricians reported that a drug's taste and palatability presented the largest obstacle to overcome for children completing treatment. 41 Additionally, many therapeutics, including LPV and RTV, have been reported as having very bitter, unpleasant taste profiles. 42 Cyclodextrins can act as taste masking agents by forming complexes with bitter drugs, effectively shielding this property of the drug. A recent study investigated the taste masking effect of HP-β-CD on ranitidine hydrochloride, a common therapeutic used to treat a variety of GI diseases such as duodenal ulcer, reflux oesophagitis, and Zollinger-Ellison Syndrome. It was concluded through the use 11 of electronic taste sensing systems that increasing the molarity of cyclodextrin in a formulation resulted in the reduction of bitter taste. 43 In some instances, the solubility of binary drug-cyclodextrin complexes is rather low, leading scientists to look for additional strategies to further enhance drug solubility. It has been shown that the addition of water-soluble polymers, forming a ternary formulation (i.e. drug-cyclodextrin-polymer), can enhance the traditional drugcyclodextrin complexation process, resulting in a system with improved water solubility and stability. 44 Other studies have corroborated this effect, showing that the addition of water-soluble polymers to a drug-cyclodextrin system resulted in higher drug solubility than when using cyclodextrin or polymer alone. 45 Water-soluble polymers can also aid in the overall stability of the final formulation, acting as drug crystallization inhibitors. 45 Despite these advantages, ternary systems are much more difficult to analyze in terms of identifying specific molecular interactions occurring between the drug, polymer, and cyclodextrin.

Phase Solubility Studies
Phase solubility studies of LPV and RTV were carried out in aqueous solutions containing either HP-β-CD or HP-γ-CD from 1. 25 where S0 is the solubility of drug in water, and the slope comes from the linear fit of the plotted data.

Isothermal Titration Calorimetry (ITC)
ITC was used to analyze the thermodynamic parameters of interactions between drug and CD at 298. 15

DG = DH -TDS
where T is the temperature of the system.

Nuclear Magnetic Resonance (NMR) Analysis
All NMR titrations, 1 H-NMR and 2D NMR, were performed in a Bruker 400 MHz NMR spectrometer at room temperature. Methanol-d4 and D2O were used as solvents to determine any solvent effects. TMS was used as an internal standard. NMR titrations were performed in two different directions to determine whether higher order drug/CD complexations such as 1:2 or 2:1 occurred. First, the host molecule (i.e. HPβ-CD and HP-γ-CD) concentration was held constant (2 mM) while the guest (i.e. LPV or RTV) concentration was varied from 0 to 25 mM. Then the host concentration was varied from 0 to 60 mM while the guest concentration was held constant (6 mM).
The chemical shift data obtained from the NMR titration study was used to plot NMR titration curves as described in Thordason's work. 47 The software used for the non-15 linear curve fit model was BINDFIT, which is offered as a freeware from supraolecualr.org.
Based on the change in chemical shift values as drug concentration increased, stability constants based on 1:1 guest-host stoichiometry were calculated using the following equations: Where Δ is the change in chemical shift value to concentration of complexes formed

Molecular Modelling
Molecular modelling was performed using Spartan16 and Molecular Operating Environment (MOE) software. The models were first constructed in Spartan and were then subjected to energy minimizations using molecular mechanics and semi-empirical level calculations. Once the energy was minimized, further docking studies were performed using MOE, where the host molecule (i.e. HP-β-CD and HP-γ-CD) was treated as the receptor and the drugs were treated as ligands. AMBER force field was used to perform the docking studies. The receptor and ligands were set for flexible alignment to ensure free movement. Once the docking was completed, docking scores were tabulated, and ligand interactions were obtained to visualize the binding inside the cavity of cyclodextrins.

Binary Complex Formation
Binary drug-CD complexes were initially formed at a 1:1 molar ratio of drug/CD.  where drug was dissolved in methanol and CD was dissolved in water, followed by the combination of these two solutions to allow for drug:CD complex formation. added to the spray dryer to be dried and collected as micron-size particulates in the particle collection chamber.

Ternary Complex Formation
The effect of various hydrophilic polymers (e.g. PVP K30, PVP VA 64, HPMC E15, Soluplus ® , and Kollidon ® ) on the solubility improvement of LPV and RTV in the presence of HP-β-CD, and HP-γ-CD was investigated. The polymer showing the greatest enhancement was used to form ternary drug-CD-polymer complexes. Ternary complexes consisting of drug-CD-polymer were produced in a similar manner as the binary complexes as described in Section 2.6.1 above. The final complex formulation contains 1% (w/v) polymer.

Solubility Study
Solubility studies of final complex formulations were performed in millipore water. Saturated solutions were prepared by adding each formulation into vials until visible precipitate was observed (indicating saturation). Samples were prepared in triplicate and shaken 37°C for 48 hours. The amount of LPV and RTV in solution was determined using HPLC, as described in Section 2.2.

Differential Scanning Calorimetry (DSC)
Thermal behavior of the raw compounds and spray-dried ternary drug/CD complexes were analyzed using a TA Q10 DSC system (TA Instruments, New Castle, DE) connected to an RSC-90 cooling accessory. Approximately 5 mg of each sample was hermetically sealed in an aluminum pan and analyzed at 10°C/min from 0 to 200°C. An empty, sealed aluminum pan was used as reference.

X-Ray Diffraction (XRD)
The crystallinity of the raw compounds and spray-dried ternary drug/CD complexes was determined using a Rigaku Multiflex X-ray diffractometer (The Woodlands, TX) with a Cu Kα radiation source of 40 kV and 44 mA. Samples were placed on a 3 mm horizontal quartz glass holder prior to analysis. Scans were taken from 5-60° in 2 with a step width of 0.2 and a scan rate of 2°/min.

Polarized Light Microscopy (PLM)
Samples of raw drugs and spray-dried ternary drug/CD complexes were obtained and analyzed for birefringence activity using an AmScope polarized microscope (Irvine, CA). Samples were mounted onto microscope slides, dispersed in mineral oil, and imaged at 10x magnification at room temperature.

Scanning Electron Microscopy (SEM)
The morphology of the raw drugs and spray-dried ternary drug/CD complexes were analyzed via a Zeiss Sigma VP Field Emission-SEM (FE-SEM, Germany). Dry powder samples were placed on aluminum SEM stubs (Ted Pella, Inc., Reading, CA) via adhesive carbon tabs. Samples were sputter coated with a film of gold/palladium alloy using a BIO-RAD sputter coating system at a 20 µA for 75 seconds under argon.

20
The moisture content of the spray-dried ternary drug/CD complexes was measured via Karl Fischer (KF) titration using a 737 KF coulometer (Metrohm, Riverview, FL). Approximately 3-5 mg of sample was dissolved in anhydrous methanol before being injected into the reaction cell filled with Hydranal ® reagent.
Water content was then quantified using anhydrous methanol as the reference sample.

In Vitro Dissolution Studies
In vitro dissolution studies were carried out using a USP apparatus II method Final formulations were created following drug:CD:Soluplus ® complexation and spray drying to produce micro-sized dry particulates, which were characterized for their morphology, thermal stability and crystallinity, water content, and in vitro dissolution.

Phase Solubility Studies
One of the major challenges of the described study was the need to improve the water solubility of two poorly-soluble small molecule drugs, LPV and RTV. Table 1 shows the molecular weight and solubility of LPV and RTV in water and in 0.1 N HCl, where both drugs exhibited aqueous solubilities below 10 µM. The solubility of RTV in 0.1 N HCl was significantly higher than LPV, mainly as a result of the protonation ability of thiazole groups present in RTV (Figure 1.2). In addition, LPV solubility decreased nearly 10-fold in acidic conditions (compared to water), whereas RTV increased 585-fold. A solubility enhancement technique that is commonly used for poorly-soluble drugs is the cyclodextrin complexation approach, in which drug molecules can interact with cyclodextrin molecules to form either inclusion or non-inclusion complexes (Figure 1.4). In an initial attempt to increase LPV and RTV solubility, phase solubility studies of LPV and RTV were carried out in aqueous solutions containing increasing concentrations of HP-β-CD or HP-γ-CD from 1.25 mM to 10 mM. This initial study was vital in that it helped elucidate the effect of CD concentration on drug solubility. Based on the Higuchi-Connors classification of phase solubility profiles (Figure A.1), the A-type phase solubility profiles were obtained for each system, indicating that solubility of the substrates increased (i.e. drug) with increasing ligand (i.e. 23 cyclodextrin) concentration. 48 More specifically, the relationship between each drug and CD were found to exhibit AL-type phase solubility profiles (Figure A.1  were also calculated based on the experimental solubility data (S0) for both drugs (Table 1).
K1:1 can be used to determine the strength of interactions between a drug and CD.
Based on the values in Table 2, the strength of interactions between β-CD and LPV and γ-CD and LPV were similar, as evidenced by their comparable K1:1 values.
Despite this similarity, the K1:1 for β-CD and LPV was higher than γ-CD and LPV, indicating that β-CD may form stronger complexes with LPV than γ-CD. This has been previously shown in literature, with the explanation that β-CD has a slightly smaller cavity than γ-CD, allowing for a better fit and intermolecular interactions between β-CD and LPV. 37 The interactions between β-CD and RTV were much higher than those between γ-CD and LPV, as indicated by the nearly 3-fold increase in K1:1, which is likely due to the smaller β-CD cavity as discussed above.
With respect to how S0 and Sint impact K1:1, the binding constants based on S0 for

Isothermal Titration Calorimetry (ITC)
ITC was used to confirm the binding stoichiometry between each drug and cyclodextrin. ITC is an analytical technique used to determine the stability and thermodynamics of cyclodextrin inclusion compounds in solution. 49 It can be used to determine parameters such as number of binding sites (n, binding stoichiometry), Typically, for guest-host (drug-CD) interactions, the complexation process is exothermic, as enthalpy-rich water molecules evacuate the CD cavity, making room for the more hydrophobic guest molecules. The amount of heat released due to complexation is proportional to the binding enthalpy of the system and to the number of complexes formed. 49 Non-linear curve fitting is used to generate the binding isotherm for each system, as depicted in  Based on the values of n listed in Table 3, it is evident that the binding stoichiometry between each drug and each CD is 1:1, since all values are near 1, confirming the results of the phase solubility analysis. The ΔH and ΔS values can be used to determine the specific types of interactions that occur during drug:CD complexation. In particular, the binding enthalpy reflects the guest-host interactions in terms of van der Waals interactions and hydrogen bonding, whereas the entropy is a reflection of hydrophobic interactions between the guest and host. 50 Based on the results, each system exhibited a negative binding enthalpy and a negative change in entropy, indicating that both van der Waals interactions and hydrophobic interactions were involved in the complexation processes.
The Gibb's free energy value for each system was found to be negative, indicating that complexations were exothermic, spontaneous processes. ΔG values for RTV for both CD were higher than ΔG values calculated for LPV systems. This indicates that RTV can form more stable complexes than LPV, which is supported by phase solubility studies that showed increased binding constant values for RTV than LPV. Overall, these data show that the binding stoichiometry for each system was 1:1 (drug:CD) based on the calculated n values, and that multiple types of interactions play a role in the complexation processes. The binding stoichiometry between drug and CD is a vital piece of information needed during formulation development.
Lastly, the calculated ΔG values for each system were found to be negative, indicating spontaneous complex formation. 51

Nuclear Magnetic Resonance (NMR)
NMR was used to investigate the changes in chemical shifts of particular protons present in CD molecules when exposed to varying drug concentrations, which can  Figures A.2-A.9). In this study, changes in chemical shift values (Δδ) were not found to be consistent with the formation of 30 higher order complexes, as indicated in Table 4, as no stability constants were available for 1:2 and 2:1 complexations.  to the right, which indicates that these protons are being shielded, an event that commonly occurs during complexation between guest-host molecules. 52 These results indicate that successful complexation occurred between LPV or RTV and the two CD. The stability constants (Ka) were calculated from the NMR data were slightly different from the experimental data. The Ka values listed in Table 4 indicate that γ-CD has the potential to form more stable complexes than β-CD for both LPV and RTV. These results are contrary to those discovered during the phase solubility studies, which indicated that β-CD forms were more stable complexes than γ-CD with the two drugs. This may be explained by the fundamental differences in the way the stability constant is calculated for each experiment. In the phase solubility study experiment, the stability constant is calculated based on regression analysis of experimental solubility data, and this study involves saturating each aqueous solution with pure, undissolved drug. In contrast, in NMR experiments the trials were performed in deuterated methanol, allowing both drugs to remain in solution for the entirety of the titration. These experimental differences may affect the trends associated with guest-host interactions, and may therefore impact the calculated stability constant values. Overall, the NMR studies were able to confirm 1:1 stoichiometry between the guest and host during complexation.

Molecular Modelling
Molecular modelling was performed to determine the most stable conformation  Table 5. Based on these data, γ-CD exhibited lower free energy values than β-CD upon interacting with either LPV or RTV, indicating that γ-CD can form more stable complexes with these two drugs compared to β-CD. These results do not correlate well with phase solubility results, in which β-CD was found to have higher binding constants for both drugs compared to γ-CD. One reason for this could be that in modeling studies only the lowest free energy conformations are examined, which are based on ideal conditions in the computational method. In addition, molecular modelling studies do not take into account non-inclusion based phenomena, in which drug and cyclodextrin complexes interact with one another to further aid in solubilization. Negative docking score values indicate that complexation can occur spontaneously and is an exothermic process, which supports the data from ITC studies.   Following the docking score evaluation, the most stable conformation of each system was selected in order to visualize the potential interactions between each drug and CD, as shown in Figure 3.4. From these models, it is possible to investigate the specific interactions involved in the complexation process, and is particularly helpful in gaining insight into the interactions that occur in the CD cavity. In addition, in each two-dimensional figure, the blue shaded regions represent the portions of the drug molecule that are exposed (not located inside of the CD cavity). For β-CD+LPV it appears that one of the aromatic regions of LPV is able to insert itself into the CD cavity. This conformation likely occurs because the interior of the CD cavity is hydrophobic relative to the exterior, and the drug aromatic rings are inherently hydrophobic due to the C-C and C-H bonds that comprise their structure. It is also likely that hydrogen bonding, in addition to van der Waals interactions play a key role in complex stabilization, as denoted by the blue and yellow dashed lines, respectively. It appears that hydrogen bonds can form between the hydroxyl groups located on the wider rim of the CD molecule and the double-bonded oxygen carbonyl groups found on LPV. For β-CD+RTV the structure of RTV is inherently different than the structure of LPV, which results in different interactions between each drug and CD molecule. RTV contains two thiazole groups, which can act as hydrogen 36 acceptors that can readily participate in hydrogen bonding. This phenomenon can be confirmed from the modeling simulation, where the sulphur atom in the thiazole group interacts with exterior rim hydroxyl groups on the CD molecule. In addition, the terminal isopropyl group present in RTV is capable of inserting itself into the CD cavity, a process most likely driven by hydrophobic interactions, as high energy water molecules release from the CD cavity, making room for hydrophobic moieties.
Additional hydrogen bonding occurs between CD hydroxyl groups and other protons present in RTV, which further aids in complex stabilization.
γ-CD+LPV interactions were similar to β-CD+LPV interactions. From the model, extensive hydrogen bonding occurs, especially between the carbonyl located on the pyrimidine on LPV and protons located on the outer CD rim. In addition, hydrogen bonding occurs between the amine group of the pyrimidine on LPV and the hydroxyl group on the CD rim. For γ-CD+RTV one of the thiazole groups in RTV appears to insert itself into the CD cavity and extend straight through to the back of the CD molecule. It is like that hydrogen bonding occurs between the inserted RTV moiety and hydroxyl groups present on the narrow rim of the CD. This phenomenon may be due to the increased size and internal volume of the γ-CD cavity, as a result of it being comprised of an additional glucopyranose unit, in comparison to β-CD. These phenomena may also explain why γ-CD+RTV exhibits the lowest docking score, and thus results in the most stable conformation. In particular, more of the drug molecule is able to insert itself into the CD cavity, and interactions occur with the narrow rim of the CD molecule. In addition, the other thiazole group present on RTV appears to hydrogen bond to exterior protons on the CD rim. Overall, these molecular modeling studies were able to provide a unique look into how drug and CD molecules form complexes. Many interactions can form between drugs and CD, however, extensive hydrogen bonding between the drug and exterior rim hydroxyl groups appears to provide a significant stabilizing effect. In addition, hydrophobic interactions appear to play a key role in forming stable complexes.

Binary and Ternary Complex Formation
Phase solubility studies were initially conducted to investigate the influence of increasing CD concentration on drug solubility. Based on these values, further studies were performed to enhance the solubility of the drugs in solution with CD through the addition of a hydrophilic polymer. Various water-soluble polymers such as PVP, Soluplus ® , Kollidon ® , HPC, PVA, and HPMC were investigated. in the presence of various polymers (0.5 weight%) in aqueous media.
A follow-up solubility study was conducted to determine the optimal concentration of Soluplus ® to use in final formulations containing drug, CD, and the polymer. Soluplus ® is a poly(ethylene glycol)-polyvinyl acetatepolyvinylcaprolactame-based grafted copolymer that is amphiphilic, containing both hydrophilic and hydrophobic moieties, which provides an ideal platform to form interactions with hydrophobic drugs while maintaining water solubility. Soluplus ® has 39 been shown to not only improve solubility of poorly water-soluble drugs, but also aids in the stability of formulations, acting as a crystal growth inhibitor for drugs in amorphous solid dispersions. 53 As seen in Figure 3.6, 1 wt% of Soluplus ® resulted in the maximum enhancement in solubility for both drugs. The addition of Soluplus ® had the greatest impact on the solubility of LPV in comparison to RTV. Based on these results, ternary formulations were prepared that include drug, CD, and Soluplus ® .

Thermal Analysis of Spray-Dried Microparticles
Thermal analysis of the spray dried formulations and raw compounds was performed using differential scanning calorimetry (DSC). As shown in Figure 3.7, a strong endothermic peak was evident at 113°C for raw LPV, which indicates the melting temperature of LPV. Similarly, a strong endothermic peak was present at 130°C for raw RTV, which indicates the melting temperature of RTV. These results are in agreement with previously reported LPV and RTV melting temperatures. 54 Raw β-CD exhibited a glass transition near 149°C, and γ-CD exhibited a glass transition near 156°C. The sharp, endothermic peaks that correspond to raw LPV and RTV disappear after spray-drying each drug, which is an indication that the drugs transition from crystalline to amorphous states during the spray-drying process.
Spray-drying is a process where dry powders are created from a feed solution containing the formulation components. The feed solution is fed through a small diameter nozzle, forming atomized droplets of solution. Depending on temperature and solvents used, most of the solvent is quickly vaporized, and the droplets quickly condense into a dry powder for collection. Since the solution is heated, drug molecules that are initially arranged in an orderly, crystalline structure become excited and rearrange themselves into a more unstable, higher energy amorphous arrangement.
This amorphous solution solidifies as it is quickly cooled. As a result, when pure drug is spray dried, its crystalline properties are lost, and thus its melting temperature disappears. The molecules no longer exhibit long order arrangement, and so the bonds 42 between drug molecules have different strengths. Therefore, there is no single amount of heat required to break the bonds; instead it requires a range of energies.

Crystallinity Analysis of Spray-Dried Microparticles
The crystallinity of pure compounds as well as spray-dried formulations was further examined using X-Ray diffraction (XRD). It can be seen in

Scanning Electron Microscopy (SEM)
SEM images were taken of raw and spray-dried drugs and final ternary formulations to investigate their size and morphology. As seen in Figure 3.11, pure LPV appears crystalline and rectangular in nature, exhibiting a size close to 20 µm.
Analogously, raw RTV exhibits a very ordered structure, with narrow, stick-like morphology, with crystals greater than 10 µm in length. The final formulations (i.e. β-CD/LPV/SOL, γ-CD/LPV/SOL, β-CD/RTV/SOL, and γ-CD/RTV/SOL) were imaged, 48 and consisted of microparticles with predominantly raisin-like morphology. The morphology of spray-dried particles is dependent on a number of process parameters such as inlet temperature and feed concentration. 55 A recent study showed that increasing the spray drying inlet temperature from 70°C to 140°C resulted in particles transitioning from spherical, smooth morphology to displaying raisin-like morphology. 56 Therefore, it is not unexpected that formulations spray-dried at an inlet temperature of 100°C and with relatively low feed concentration (20 mg/mL) would exhibit raisin-like morphology. A high inlet temperature and low feed concentration causes fast vaporization of the solvent from solution and subsequent, internal bubble nucleation, particle inflation and particle surface deformation during drying, thereby preventing droplets from forming spherical particles.

Karl Fischer Titration
Karl Fischer titration was used to determine the water content of the final ternary formulations. As seen in Table 6, water content ranged from 1.06 to 1.24 wt%. The moisture content of formulations is an important characteristic relating to product stability, where low moisture content can reduce particle agglomeration as well as prevent crystal growth formation from amorphous drug in a formulation. 57 These values indicate that dry microparticles were in fact produced, and that the spray-drying process is capable of removing most of the moisture from the samples. dissolution profile of the samples tested. Unexpectedly, the cumulative release of the spray-dried LPV was around 1.3% over 120 minutes, which is significantly lower than that of the pure LPV. This result indicates that transitioning LPV from crystalline to amorphous does not necessarily result in improved dissolution rate and extent. This was consistent with a recently reported study, 58 in which different metastable forms of LPV were studied.
Considering that LPV has four crystalline forms, it may be possible that amorphous LPV recrystallizes to form a less soluble crystalline form of LPV under super saturated conditions (non-sink conditions). Further studies need to be conducted to elucidate this phenomenon. Ternary formulations (i.e. β-CD/LPV/SOL and γ-CD/LPV/SOL) exhibited enhanced dissolution rates over the pure drug and spraydried drug (Figure 3.12A). This is most likely due to the effects of CD complexation aided by the addition of water-soluble polymer Soluplus ® rather than LPV being amorphous. It is worthy to note that large pieces of solid aggregates were observed during the dissolution testing of the ternary formulations, indicating that a significant amount of drug may not have access to the dissolution medium and hence hampered dissolution. In addition, the physical mixtures of CD, LPV, and SOL resulted in an enhanced dissolution profile with a decrease in dissolution at 90 minutes followed by a sharp increase in dissolution. This indicates that polymorphic transition may have occurred during the dissolution testing process. It may be possible that LPV  Overall, these results indicate that LPV and RTV exhibit fundamentally different physiochemical characteristics and hence different outcomes of dissolution enhancement by forming complexes with CD. β-CD/RTV/SOL complexes showed the most profound dissolution enhancement.

CHAPTER 4 CONCLUSIONS AND FUTURE WORK
The interactions between two types of cyclodextrin (i.e. HP-β-CD and HP-γ-CD) and two different anti-HIV model drugs (i.e. LPV and RTV) were thoroughly investigated. Phase solubility studies confirmed that as CD concentration increases, there is a corresponding increase (linear relationship) in drug solubility for both LPV and RTV. It appears that HP-β-CD has a more significant impact on enhancing solubility. This can be due to the smaller size of the CD cavity itself, increasing the likelihood of interactions forming between drug and CD. In addition, it is possible that the drugs are forming non-inclusion complexes, and LPV and RTV are interacting with the functional groups (e.g. hydroxypropyl) on the outside of the CD and not just with the hydrophobic interior. ITC studies were performed mainly to investigate the binding stoichiometry between drug and CD. This information is extremely important when deciding on the ratio of drug:CD to use in the actual formulations. It was found that for both types of CD and both drugs, a 1:1 interaction was favored based on their thermodynamic behavior. The negative Gibb's free energy values indicate that the complexation is a spontaneously exothermic process. NMR titration and molecular modeling studies were then performed to further confirm these finding, and to closely examine the particular the interactions that occur between the drugs and the CD. It was shown that multiple types of intermolecular forces such as hydrogen bonding and hydrophobic interactions play a role in the formation and drug:CD complexes.
To further enhance the dissolution characteristics of the two model compounds, a variety of polymers were studied. It was found that 1% Soluplus ® (SOL) provided the greatest increase in solubility for both drugs, and therefore this polymer was chosen for further studies. Ternary complexes were prepared using drug, CD, and Soluplus ® , and then spray-dried to yield dry powder microparticles. Various solid-state characterization techniques were used to analyze the dry complexes. SEM imaging revealed that microparticles exhibited mainly raisin-like morphology, with sizes ranging from 2-4 microns. DSC results showed that LPV and RTV had melting temperatures of 113°C and 130°C. Data also suggests that spray-dried formulations may contain amorphous drug, as indicated by the disappearance of melting temperature peaks. XRD results confirmed this finding, as both pure LPV and pure RTV exhibited shark, distinct peaks that disappeared in all complex formulations.
PLM images were obtained to provide visual evidence of this phenomenon. Both drugs showed a significant degree of birefringence, indicative of crystalline structures.
On the contrary, spray-dried drug as well as binary and ternary formulations exhibited a lack of birefringence, indicating that both drugs converted from crystalline to amorphous during the microparticle formulation process. In this study, we were able to successfully create spray-dried microparticles and characterize them based on their solid-state properties.
In vitro dissolution study showed that LPV and RTV had vastly different release profiles in both conditions, and that the effect of crystallinity had a different impact on each drug. For LPV, the physical mixture of LPV, CD, and SOL provided the greatest release profile. It was also found that spray-dried LPV did not offer any dissolution 56 enhancement, and in fact resulted in a lower release profile. On the other hand, the release rate of RTV was drastically enhanced in both formulations containing RTV, CD, and SOL, when compared to the controls. In addition, it was found that spraydried RTV exhibited a better dissolution profile than crystalline drug, therefore indicating that spray-drying has a positive effect on overall RTV dissolution kinetics.
Ternary RTV:CD complexes were able to successfully enhance dissolution rate and extent when compared to the controls.
Based on these findings, a different formulation technique will be required to enhance solubility and dissolution profile of LPV. It appears that converting LPV from crystalline to amorphous form via spray drying adversely affected dissolution characteristics of LPV, and therefore other methods such as kneading and lyophilization that do not involve high temperature will be studied. The taste assessment study of the optimized formulations will be performed to assess tastemasking capability of CD. In addition, stability studies will be conducted to analyze the effects of CD and SOL on the stability of both drugs. Following these studies, oral bioavailability study of the optimized formulations will be performed using a rat model.