An exploratory study of a simpli ﬁ ed approach to evaluate drug solubility in milk related vehicles

Objectives: Milk related materials are frequently used as a vehicle for drug product administration. Therefore, drug sol-ubility information in milk related vehicles is desirable for prediction of how they may in ﬂ uence in vivo drug release and bioavailability. However, there are very limited data published on this topic. This study explored a practical method to address the key challenges associated with solubility assessment in milk, including the sample equilibration time and cleanup procedures. Methods: Amitriptyline, acetaminophen, dexamethasone, nifedipine, piroxicam, and prednisolone were selected as model drugs to represent a wide range of physicochemical properties. Their solubilities were determined at room temperature in pH 6.8 phosphate bu ﬀ er, skim milk, whole milk, reconstituted milk powder, and unprocessed raw milk. Results: The overall results con ﬁ rmed that milk greatly improves the solubility of poorly water-soluble drugs. However, the extent of improvement and mechanism of solubilization appeared unique for each drug, highlighting the importance of evaluating milk solubility experimentally. Conclusions: The method used in this exploratory study can be applied in future investigations of a broader range of drugs and milk-related vehicles.


Introduction
Drug release from oral solid dosage forms is a critical step impacting the overall bioavailability.Milk-based drinks are often used when patients take whole capsules and tablets.They are also frequently used as liquid vehicles to disperse capsule contents or crushed tablet powder to facilitate dose administration for certain patient populations [1][2][3].A variety of cow milk commodities are available to consumers, and the most common ones in grocery stores vary in the percent of milk fat from 0 % (skim milk) to approximately 3.3 % (whole milk) [2,4].Components of milk are routinely used as key ingredients to manufacture a wide range of palatable nutrition products which may be used directly or after reconstitution for drug administration.Several recent studies also evaluated lyophilized milk as an excipient matrix for orally dispersible tablets which are particularly valuable for pediatric patients [5,6].
Milk contains a variety of chemical components which are expected to influence drug solubility [1][2][3][4].Despite the frequent use of milk-based products in drug delivery, there is a paucity of primary literature data on drug solubility in these vehicles or on how that solubility effect can vary as a function of the drug's physicochemical characteristics.Searches in PubChem and DrugBank databases revealed that most drug monographs contain only data of drug concentrations in milk from nursing mothers or dairy cows within the context of pharmacokinetic distribution and excretion to address safety concerns [7,8].Based on the PubMed [9] search results, only one research group has published comprehensive reports on drug solubility in milk, using a flow injection serial dynamic dialysis technique [10][11][12].More recently, another research group studied solubility of two model drugs in milk along with a panel of other soft foods and drinks [13,14].
There are several key challenges in performing drug solubility studies in milk due to its complex compositions and properties.Traditionally, solubility samples are agitated with excess solid drug particles at room temperature for >4 h (and often overnight) to reach equilibrium.However, such prolonged test procedures are not feasible for milk-related vehicles due to the potential for rapid microbial growth.Moreover, after equilibration, the excess drug particles are typically removed by centrifugation or filtration through 0.45 µm filters.This step can be problematic for milk, because it is an oil-in-water emulsion system and the aqueous phase also contains colloidal protein components.Centrifugation leads to phase separation, and 0.45 µm filters can be blocked by oil globules and colloidal protein particles.Finally, drug concentrations in saturated solution samples are typically analyzed directly by HPLC-UV methods.Milk samples, on the other hand, require proper extraction steps prior to analysis to account for all the solubilized drug in the system, including free and protein-bound drug in the aqueous phase, drug in the oil phase, and drug at the aqueous-oil interface.Additional sample treatment steps may also be necessary to reduce analytical interferences from other components in milk, such as protein.
The objective of this study was to explore a practical method to evaluate drug solubility in milk-based vehicles.Several representative vehicles and model drugs were selected and evaluated.The results provide useful preliminary data for future applications to a broader range of vehicles and drugs.The method can also be incorporated in various drug release studies to aid in drug product development activities and for the assessment of palatable, nutrientbased matrices for drug administration to special populations.

Materials
APIs: Six model drugs with diverse physicochemical properties were selected for this exploratory solubility study as listed in Table 1.The source information is provided in Table 2.
Vehicles: All milk vehicles were stored under refrigeration (2-8 °C) until use.The skim and whole milk products were purchased from the local supermarket (Wegmans Food Markets, Rochester, NY) and used within the expiration dates.The fresh raw milk was provided directly by a local farm (Craft Dairy Farm, Williamson, NY) and used within 24 h.The cows were of Holstein breed on corn and forage diet.The dry whole milk powder, Nestle Nido ® , was purchased from Amazon or Walmart.The reconstitution was performed according to product instructions by mixing 1 oz.(31 g) powder with 8 fl.oz.(240 mL) warm (30-40 °C) water.The reconstituted milk was used within 24 h.
Solvents for solubility sample treatment and HPLC analysis: A Milli-Q Direct eight system from Millipore Sigma (Burlington, MA) was used to produce purified water.Acetonitrile (MeCN) and trifluoroacetic acid (TFA), both HPLC grades, were purchased from Thermo Fisher Scientific (Waltham, MA).

Physicochemical characterization of milk vehicles
pH: A Mettler-Toledo (Columbus, OH) Seven Easy model pH meter was used with a gel-filled pencil-thin pH electrode from Thermo Fisher Scientific (Waltham, MA).All milk vehicles were equilibrated to room temperature prior to the pH measurement.
Osmolality: A µOsmette Model 5004 osmometer from Precision Systems (Natick, MA) was used for osmolality measurement, which operates based on the freezing point depression method.For each measurement, 50 µL of the vehicle was placed in the sample tube and lowered into the freezing chamber.After the solenoid-induced pulse freezing, the liberated heat of fusion was related by the microprocessor to the freezing point of the sample, and the osmolality was automatically calculated and displayed.
Viscosity: A DHR-2 rheometer with a Peltier temperature-controlled cup, and a double-gap concentric cylinder measurement system from TA Instrument (New Castle, DE) was used for viscosity measurements at 25 °C.For each measurement, 12 g of the vehicle was placed in the cup, and the shear rate was scanned from 0.001 to 1 s −1 on a log-scale with 5 points/decade.The method was set to determine the steady-state viscosity result with a maximum equilibration time of 480 s at each point.
Globule size distribution: A LA-960 laser scattering particle size analyzer from Horiba (Piscataway, NJ) was used to analyze the globule size distribution of the milk vehicles at room temperature.A refractive index of 1.45 was used for the oil phase and 1.333 for the aqueous phase.
The vehicle was added dropwise to deaerated water to achieve 90-95 % transmission for the red laser.A circulation speed of 2 was used with no sonication, and a data acquisition of 10,000 per read.

Solubility assessment and sample treatment for HPLC analysis
Solubility assessment was performed in triplicates for each drug and milk vehicle combination.A 10 mM sodium phosphate buffer pH 6.8 was also included as a control vehicle.A general flow chart is provided in Figure 1.For each solubility sample, an excess amount of drug powder was weighed in a 20-mL glass vial.Ten mL of the vehicle, preequilibrated to room temperature (RT; 20-25 °C), was added to the drug powder.The samples were stirred at RT using a small stir bar at 100 rpm for 60 min.It was assumed that the pH of the vehicles and the chemical stability of drugs remained unchanged during this period.Upon completion, approximately 2 mL of each solubility sample was passed through the Phenex-GF 1.2 µm filter to remove the excess solid.An aliquot of 0.5 mL filtrate was accurately transferred to a microcentrifuge tube followed by the addition of 1 mL acetonitrile (MeCN) for drug extraction and protein precipitation.The mixture was then centrifuged at 10,000 rpm for 10 min, and about 1 mL of the supernatant was withdrawn passed through a Phenex-RC 0.45 µm filter.The filtrate was collected in an HPLC autosampler vial and analyzed by the HPLC method described below.Placebo milk vehicles without drug were treated the same way to serve as controls for HPLC analysis.Considerable back pressure was observed during the 1.2 µm filtration step of the reconstituted whole milk powder and raw milk, with or without drug.Additional investigation revealed that back pressure was inversely correlated to milk temperature.The solubility experiment was repeated at 38 °C with one model drug, nifedipine, for confirmation.

HPLC analysis
A gradient HPLC method, described in Table 3, was used to analyze the concentration of all drug solubility samples after treatment.The UV detection wavelength and sample injection volume were selected to streamline the HPLC analysis.

Milk vehicles and properties
Milk is a biological fluid with complex chemical compositions and physicochemical properties [2,4].Raw and whole milk include all categories of nutrients, and the protein and lipid components are expected to have a substantial impact on the solubility of many drugs, especially the ones with poor aqueous solubilities.In addition, pH is a critical factor for solubility of drugs with ionizable functional groups.
Four milk vehicles were included in this exploratory study with the following rationales.Skim milk and whole milk are widely available at most food stores which are suitable materials for routine solubility experiments.The main difference in their lipid contents would help elucidate its effect on drug solubility.Whole milk powder provides a convenient alternative to fresh whole milk, as it can be stored at RT for extended period of time.Using the powder product from an established brand also minimizes potential variability issues of whole milk due to bovine breed, environment and diet [21,22].Raw milk is consumed in certain regionals, and it is important for the development and evaluation of certain veterinary drug products intended for intramammary delivery.It is not pasteurized or homogenized, and the effect of these processes on drug solubility has not been well characterized in the past.Several relevant physicochemical properties were evaluated for the four milk vehicles, and the results are summarized in Table 4.The pH values of all milk vehicles ranged from 6.6 to 6.8, which is consistent to published data [22].The osmolality values of all vehicles, except reconstituted whole milk powder, were within the isotonic range of approximately 280-300 mOsm/kg.The low osmolality of the reconstituted milk powder was possibly attributed to the loss of some ingredients during the manufacturing process.All vehicles exhibited low viscosity at room temperature.The globule size D 50 data reflected marked differences across the milk vehicles.As expected, whole milk contained much smaller oil droplets than raw milk due to the homogenization process.Interestingly, the reconstituted whole milk powder exhibited smaller D 50 than whole milk.This was probably due to the presence of soy lecithin additive in the milk powder formulation which facilitated the emulsification process during reconstitution.Nevertheless, there were still some large globules detected in the reconstituted milk which were not reflected by the D 50 result.In other words, it exhibited a wider dispersion of globule size as compared to that of whole milk.Skim milk contained very little milk fat, and the small D 50 value most likely reflected the size of the colloidal protein components.

Model drug solubility in milk vehicles
Six model drugs were used in this study to represent a wide range of physicochemical properties as shown in Table 1.Their milk solubility experiments were performed using the exploratory method, and the data are summarized in Table 5.For all six drugs, the solubility values in the pH 6.8 buffer control were in general agreement with the published data in water as shown in Table 1.This confirms that the 60-min equilibration time and the 1.2 µm filtration step (to remove excess drug) were suitable for regular drug powder materials.
As expected, the solubilities of the four poorly watersoluble drugs were markedly improved in all milk vehicles (Table 5).However, the exact trend for each drug seemed unique and not fully rationalized by their plasma protein binding or o/w partition characteristics (Table 1).Dexamethasone and prednisolone have closely related chemical structures with no ionizable groups.Their solubilities increased considerably in skim milk vs. aqueous buffer but no additional increases in solubility occurred when tested in whole or raw milk.On the other hand, the solubility of piroxicam exhibited no difference between the control buffer and skim milk but did exhibit a substantial increase when tested in whole and raw milk.This was somewhat surprising given the high % plasma protein binding but relatively small Log P of piroxicam.For nifedipine, its solubility benefited from components in both skim and whole milk.Surprisingly, the solubility of nifedipine in raw milk was substantially lower than that in skim or whole milk.Overall the results highlighted the complexity of drug solubility behavior in milk vehicles.Besides the plasm protein binding and o/w partition properties, other factors may need to be considered, e.g., the protein compositions in milk vs. plasma and drug distribution at the interfaces.More experiments and data in this area are needed to enhance the overall understanding and enable accurate predictions in future.
One consistent and encouraging finding is that the whole milk powder, after reconstitution, offered similar solubility results as whole milk.Therefore, drug solubility in either vehicle can be assumed to be interchangeable and reconstituted whole milk can be considered a possible substitute in routine solubility studies.However, it was noted that the reconstituted whole milk and raw milk samples, with or without any drug, exhibited high back pressure when being filtered through the 1.2 µm filters.This was likely due to the presence of large oil globules as discussed above.When these two vehicles were warmed to >35 °C, the back pressure dropped considerably, probably due to the reduced viscosity of the oil.The solubility experiment was repeated at 38 °C for nifedipine in these two vehicles without any filtration issues.The solubility increased as expected at higher temperature, and the trend was comparable for raw milk and reconstituted whole milk.

Conclusions
Drug solubility data in milk provide valuable information to support many aspects of drug product development and administration.An exploratory method was developed to overcome key challenges associated with solubility studies in milk vehicles.The utility of the method was demonstrated in this study by six model drugs in four milk vehicles.Future studies are encouraged to apply and improve this method for the investigations of a broader range of drugs and milkrelated vehicles.
c . (base) . (acid) Negatively charged −. (pH .) . (pH .) ∼ % Prednisolone Very slightly soluble (.- mg/mL) Non-ionizable Neutral ∼.   - % a The solubility descriptions and ranges are based on USP definitions [].b Amitriptyline was only available as the HCl salt commercially at the time of the study.c Solubility of piroxicam is pH dependent.At pH .,  °C, it was reported as . mg/mL [].

Table  :
Material source information of model drugs.
Figure 1: Flow chart of the solubility assessment and sample treatment steps.Li et al.: A simplified approach to evaluate drug solubility in milk related vehicles

Table  :
HPLC method for drug analysis.

Table  :
Physicochemical properties of the milk vehicles.