Fast Screening Methods for the Analysis of Topical Drug Products

: Considering the recent regulatory requirements, the overall importance of in vitro release testing (IVRT) methods regarding topical product development is undeniable, especially when addressing particulate systems. For each IVRT study, several hundreds of samples are generated. Therefore, developing rapid reversed-phase high-performance liquid chromatography (RP-HPLC) methods, able to provide a real-time drug analysis of IVRT samples, is a priority. In this study, eight topical complex drug products exhibiting distinct physicochemical proﬁles were considered. RP-HPLC methods were developed and fully validated. Chromatographic separations were achieved on a XBridge TM C18 (5 µ m particle size, 150 mm × 2.1 mm), or alternatively on a LiChrospher ® 100 RP-18 (5 µ m particle size, 125 mm × 4.6 mm) at 30 ◦ C, under isocratic conditions using UV detection at speciﬁc wavelengths. According to the physicochemical characteristics of each drug, di ﬀ erent mobile phases were selected. Irrespective of the drug (hydrocortisone, etofenamate, bifonazole, clotrimazole, acyclovir, tioconazole, clobetasol, and diclofenac) and formulation, retention time values did not exceed 6.5 min. All methods were linear, speciﬁc, precise, and accurate at the intraday and interday levels, robust, and stable. These were successfully applied to establish product-speciﬁc IVRT proﬁles, thus providing a key database useful for topical pharmaceutical manufacturers.


Introduction
The market of skin drug delivery is projected to expand over the next five years [1,2]. Such a trend may certainly be boosted by the increasingly growing relevance of nanotechnology-based products. Reasons fueling this expansion include, among other factors, elderly population increase, demand for a superior topical therapeutic effect, and major patent expiration. Semisolid dosage forms intended for topical application enable prolonged in situ release, and for this reason are the preferred therapeutic vehicles when addressing topical conditions [3].
Semisolid dosage forms often use biphasic systems, such as oil-in-water emulsions, where the drug can be dispersed in both phases [4,5]. A complete microstructure characterization in such circumstances should include pH, rheological profile, polymorphism, active pharmaceutical ingredients, and globule size determination, together with a detailed understanding of both release kinetics and permeation behavior. In this context, even though semisolid preparations are one of the oldest classes of dosage forms, known for over 100 years, methods/general procedures able to fully characterize their microstructure/performance have been subjected to recent scientific and regulatory advances [4,6]. The standardization of these procedures will be even more complex with the introduction of nanosystems [7,8].
The European Medicine Agency (EMA) guideline on quality and equivalence of topical products as well as the US Food and Drug Administration's (FDA) acyclovir draft guidance are clear regulatory examples, which reinforce the growing need to fully characterize topical products [9][10][11].
Among the diversity of methodologies available for semisolid microstructure characterization, in vitro release testing (IVRT) assumes particular relevance. IVRT shares the same scientific principles as the dissolution methods, since the active substance ingredient must be released before it becomes bioavailable in the skin. In this context, IVRT methods are able to characterize the pharmaceutical performance of a given topical product; they are highly sensitive to changes in microstructure, thus representing a valuable source of information throughout a topical product lifecycle from the early development phase, as a formulation screening tool, to the late-stage development and post-marketing phases, as a quality control specification [4,[12][13][14].
Release testing is usually conducted in a vertical diffusion system that consists of a donor compartment where the formulation is applied, an artificial non-rate-limiting membrane that holds the product, and a receptor compartment fulfilled with the release media [15,16]. To standardize method conditions, release experiments should be temperature controlled and continuously stirred. This avoids the occurrence of stationary layers that limit drug diffusion and prevent homogenization of the drug within the receptor solution [11,[17][18][19]. Moreover, prior to the release experiment, it is vital to perform solubility studies to assure the maintenance of sink conditions. Throughout IVRT studies, samples from the donor compartment are taken and processed by analytical methods such as reversed-phase high-performance liquid chromatography (RP-HPLC) in order to trace the release profile.
According to recent regulatory requirements, 12 replicates should be considered and sampling should be performed at least for 6 times within the linear portion of the drug release profile [9]. Therefore, for each IVRT study, from several dozen to hundreds of samples are generated, which demand rapid analysis to avoid stability issues and provide timely evaluation as well as real-time assessment of specification conformity. This assumes critical importance for routine decision making, especially when out-of-trend (OOT) or out-of-specification (OOS) results occur.
The present study aims to create a fully validated library of RP-HPLC methods, directed to the real-time drug analysis of eight blockbuster topical drug products. Furthermore, a framework for the development of IVRT studies, based on active pharmaceutical ingredient (API) physicochemical characteristics, is actively pursued. To the best of our knowledge, there is no literature reporting this methodological assembling to assist with fast screening in topical product development.

Materials
Micronized hydrocortisone, etofenamate, bifonazole, clotrimazole, acyclovir, tioconazole, clobetasol propionate, and sodium diclofenac were kindly supplied by Laboratórios Basi. Water was purified with a Millipore MILLI-Q reagent water system and filtered through a 0.22 µm nylon filter before use. All other chemicals were of analytical grade or equivalent.

Preparation of Stock Solutions, Calibration Standards, and Quality Controls
All stock solutions were prepared by weighing approximately 10 mg of drug in 10 mL of an appropriate solvent, yielding ca. 1 mg/mL concentration.
Methanolic stock solutions were prepared for diclofenac, clobetasol, tioconazole, etofenamate, and clotrimazole. Alternatively, for bifonazole and hydrocortisone, acetonitrile stock solutions were prepared instead. For acyclovir, due to its high hydrophilicity, the stock solution was directly prepared in water. Two working standards were considered for each drug. These were prepared by further dilution of each stock solution with the corresponding release media used in IVRT studies. Table 2 summarizes the release media used for each drug. The rationale used for IVRT release media selection is carefully explained in Section 2.4.8. Six to ten standard solutions were considered for each calibration curve. Moreover, as quality controls, five replicates of three different concentration standards were used.
Since the focus of the present work was to quantify IVRT samples, preliminary in vitro tests were performed to determine the appropriate range for the calibration standards. Table 3 summarizes the concentrations used for each molecule. In all cases, IVRT and standard samples were directly injected. All stock and working solutions were freshly prepared each day.

Method Validation
Validation studies were performed according to the International Council for Harmonization (ICH) guidelines for each drug substance [20,21]. These included system suitability, limits of detection, and quantification, linearity, accuracy, precision, robustness, and stability assessments.

System Suitability
System suitability was evaluated by injecting the same standard solutions six times. The following requirements were considered: relative standard deviation (RSD) of the detector response and retention time for all standard injections was not more than 2%, capacity factor (k') was higher than 2, tailing factor (T) of drug peak was not more than 2.0, and theoretical plate number (N) was higher than 2000. These parameters were obtained by injecting five times an intermediate standard for each drug [22].

Limits of Detection and Quantification
A specific calibration curve obtained from six standard solutions was traced in order to determine the limits of detection (LOD) and quantification (LOQ).
The following expressions were used: LOD = 3.3 σ /S and LOQ = 10 σ /S, where σ is the standard deviation of the response and S the slope of the calibration curve [21].

Linearity
To test the linearity of the detector response, a set of six to ten calibration standards were prepared for each drug, in the corresponding release media. A specific range for each active substance was adopted, taking into account IVRT results. Please see Table 3.
At least four calibration curves were considered for each active substance. Linearity was determined through the calculation of a regression line, attained from the peak area as a function of the standard concentration, by the method of the least squares. To comply with regulatory requirements, curves that did not present a R 2 of at least 0.99 were not considered [23].

Accuracy and Precision
Precision measures the closeness of agreement, i.e., the degree of scatter between a series of measurements obtained from multiple sampling of the same homogeneous sample. This validation parameter was assessed by repeatability (intraday) and intermediate precision (interday) for three days, of the established quality control (QC) standards. Five samples of each concentration level were prepared. The RSD determined at each concentration level should not exceed 15%, except for the lower concentrations, where a maximum of 20% was allowed [20,24].
To demonstrate accuracy, the closeness of agreement between the true value and the found value, Equation (1), was used. Afterwards, the mean bias percentage for the five replicates of the three QCs was calculated.
Equation (1): The same acceptance criteria regarded for precision were also considered to establish the methods accuracy [20,21].

Ruggedness
To prove the methods ruggedness, five duplicates of the quality control samples were alternatively analyzed using different HPLC equipment, a Shimadzu LC-10AD apparatus, equipped with a quaternary pump (LC-10AD), an autosampler unit (SIL-10ADVP), a CTO-10AVP oven, and a CBM-20 A detector.

Specificity and Selectivity
Specificity, viz., the ability to accurately measure the analyte in the presence of all potential sample components, was evaluated by comparing the chromatograms of three different control matrices: (i), blank receptor medium solution (negative control, C n ), (ii) quality control solutions prepared in receptor medium (positive controls, C p ), and (iii) release medium solution of an IVRT run conducted with the tested formulations (matrix positive control, C pm ). The main purpose of this assessment was to ensure that the integrity of each active substance retrieved from IVRT samples was not compromised by any formulation excipient.
The negative controls (C n ) were analyzed to confirm the absence of any detectable drug concentration. The mean retention time and concentration values (regarding the nominal concentrations) for the C p and C pm were used to set the acceptance criteria for the drug retention time, and the difference in both parameters should not exceed 15% [14].

Stability
Five replicates of quality control solutions were prepared to evaluate the stability of IVRT samples at room temperature storage (25 • C) for 24 h in the autosampler and in short-term storage at 4 • C for 72 h [24,25]. The same acceptance criteria established during precision and accuracy assessments were used.

Franz Cell Receptor Fluid Screening
The maintenance of sink conditions is crucial throughout the release experiments, therefore, screening different receptor media (co-solvents and pH effects), according to the active substance physicochemical profile, should be a priority concern [26].
In order to rationally establish solubility conditions to be tested, in silico studies were performed first. For that, the chemical structure of each API was introduced in Chemaxon ® software (ChemAxon, Budapest, Hungary) in order to calculate the respective chemical descriptors, such as size, geometry, lipophilicity, solubility, and surface topology [27].
For solubility studies, the adopted protocol included weighing 60-100 mg of active substance into 2 mL flasks. These were then filled with the tested release medium and continuously stirred for 24 h in an oven (37 ± 0.1 • C). Afterwards, the samples were centrifuged, and the supernatant was filtered and appropriately diluted with a mobile phase. The active substance solubility in the different media was determined by RP-HPLC. The results were statistically compared using a two-way analysis of variance (ANOVA), with a Tukey multiple comparison test. The differences between the means were considered to be significant for values of p < 0.05. The statistical tests were applied using GraphPad Prism, version 6.01 (GraphPad Software, San Diego, CA, USA).

IVRT Studies
Release studies of the different drug products were conducted using static vertical Franz diffusion cells (PermeGear, Inc., PA, USA) with a diffusion area of 0.636 cm 2 and a receptor compartment of 5 mL. All studies followed the indications provided in the Topical and Transdermal Drug Products-Product Performance Tests section of the United States Pharmacopoeia (USP) [23]. Briefly, 300 mg of the semi-solid formulations were applied in the donor compartment, separated from the receptor compartment by a polysulfone membrane (HT: 200 Tuffryn diameter 25 mm, pore size 0.45 µm, Pall Corporation, USA), previously soaked in distilled water for 30 min. Specific receptor media were used for each drug product, see Table 2. The receptor media was in all cases continuously stirred at 600 rpm and maintained at a temperature of 37 • C, while a circulating water bath maintained the membrane surface at 32 • . Before the release experiments, the system was allowed to equilibrate for at least 30 min. Throughout the release studies, the donor compartment and the receptor sampling arm were carefully covered with Parafilm ® to avoid unnecessary release media volatilization and to conduct all the release experiments under occlusive conditions. Samples of the receptor phase (300 µL) were withdrawn at 0.25, 0.5, 1, 2, 3, 4, 6, and 8 h.
After each collection, the same volume of medium was replaced with fresh receptor solution. The cumulative amount of drug released as a function of time was calculated in relation to the amount of formulation placed in the donor compartment using Equation (2): where C n corresponds to the drug concentration of the receptor medium at each sampling time, C i to the drug concentration of the i th sample, A to the effective diffusion area, and V 0 and V i to the volumes of the receptor compartment and the collected sample, respectively. The release rates were calculated from the slope of the regression line obtained by plotting the cumulative amount of drug diffused per cm 2 versus the square root of time [14,28,29].
Mass balance studies were also performed to characterize the extent of dose depletion during the IVRT. For that, the average cumulative amount released at the last sampling point (Q f ) was divided by the actual amount of active substance placed in the diffusion cell, and multiplied by 100. An n = 12 was considered for each product.

System Suitability
Chromatographic separation of all drugs in their respective release media was successfully achieved using the previously described analytical conditions, see Figure 1.
Please note that the products used in this study are complex formulations with multiple ingredients. Preservatives such as benzyl alcohol or methyl parahydroxybenzoate, present in some of the studied products, absorb in the UV region. For this reason, additional peaks are present in IVRT samples of bifonazole, clobetasol, and tioconazole (pertaining to benzyl alcohol) and in hydrocortisone (corresponding to methyl parahydroxybenzoate). In all cases, drug peaks were resolved from the additional components, thus highlighting the specificity of the proposed methods.
To further illustrate the chromatographic similarity between the standards and IVRT samples, the retention times are presented in Table 4.  Please note that the products used in this study are complex formulations with multiple ingredients. Preservatives such as benzyl alcohol or methyl parahydroxybenzoate, present in some of the studied products, absorb in the UV region. For this reason, additional peaks are present in IVRT samples of bifonazole, clobetasol, and tioconazole (pertaining to benzyl alcohol) and in System suitability tests aim to assess if the chromatograph and respective modules are able to generate acceptable accuracy and precision results. To provide quantitative data, several parameters should be taken into consideration. These include: number of theoretical plates, separation factor, resolution, tailing factor, and precision [30].
System suitability results and acceptance criteria are summarized in Table 5. The precision of both peak area and retention time was compliant for all drugs, thus documenting the system's ability to detect and analyze the compounds in their respective release media.
Column efficiency was also evidenced for all drugs, since the number of theoretical plates (T. Plates), tailing factor (T. Factor), and resolution were compliant with ICH criteria. Even though the capacity factor (k ) for acyclovir and clobetasol was slightly beneath 2.0, according to the British Pharmacopoeia, this value can still be accepted (k < 1.5) [31].

Limits of Detection and Quantification
Since different dosage forms were considered for the present study, different release mechanisms and release ranges are expected. Therefore, preliminary IVRT results were used to determine specific and plausible LOD concentrations, please see Table 3. LOD and LOQ values are summarized in Table 6.

Linearity
IVRT sample concentration is interpolated by using the regression results obtained from calibration curves. These are often calculated through the least squares method that estimates the correlation coefficient (R 2 ) and the regression equation [24,32]. Table 7 depicts the results from linearity studies. Table 7. Results obtained from the regression analysis using the least squares method for all studied drugs. Results are expressed as mean ± standard deviation (SD), n = 4. Please note that a weighted linear regression was performed on calibration data using 1/x 2 as the weighting factor. All R 2 were superior to 0.99 and, therefore, evidence is provided regarding data quality and linearity in the proposed range.

Accuracy and Precision
Calibration curve quality should also be documented through the evaluation of precision and accuracy. Both parameters should be assessed at the intraday and interday levels.
According to our results, displayed in Table 8, RSD values did not exceed 15%, even for the lowest concentration QC. The same trend was also verified for the accuracy results.  These results substantiate that the developed methods are accurate, reliable, and reproducible, since they all met the acceptance recommendations.

Ruggedness
The ability of the method to provide reliable results despite minor variations in the analytical conditions can be defined as ruggedness or robustness. Nevertheless, both terms present slight differences [33]. ICH defines analytical method robustness as, "the measure of its capacity to remain unaffected by small, but deliberate variations in method parameters and provides an indication of its reliability during normal usage" [34]. Furthermore, ICH provides examples of such variations. These include mobile phase composition/pH variations, different columns, and different temperature analysis or flow rates [33,34]. USP defines ruggedness as a measure of reproducibility of the test results under the variation conditions normally expected from laboratory to laboratory and from analyst to analyst [35]. According to these compendial definitions, the term ruggedness is more accurate when considering two different equipment types. Table 9 summarizes robustness results.
Even though all results are in good acceptance with regulatory standards, some borderline results were verified for hydrocortisone, etofenamate, and clotrimazole.

Stability
Stability testing should mimic common sampling procedures, so as to provide reliable data analysis. The stability conditions assessed included autosampler and short-term stability. Results are presented in Table 10.  According to the results, it can be inferred that the analytes are stable under the studied conditions, since precision and accuracy requirements were met.

In Silico Studies: Using Chemical Predictors to Support IVRT Development
In order to determine the most appropriate conditions for solubility studies, the physicochemical properties of each molecule were firstly predicted in silico and used to infer their ionization, lipophilicity, permeability, and solubility status, please see Table 11. The distribution coefficients (Log D) were determined within a physiological range of pH values compatible with skin application. The distribution coefficient is the appropriate descriptor for ionizable compounds, since it relates to the solubility of all chemical forms depending on the selected pH.
Hydrocortisone, etofenamate, acyclovir, and clobetasol, as neutral actives, do not present changes in LogD at different pH values. By contrast, all antifungal drugs and diclofenac show opposite behaviors. For diclofenac, the Log D at 4.6 was higher than the Log D at 6.5, since the non-ionized form is predominant at an acidic pH [36]. As expected, the reverse situation is observed for all antifungal drugs, where an increase of pH is followed by an increase of the Log D. Bifonazole, tioconazole, and clotrimazole are mainly present in their ionized form at an acidic pH (1.7), however, this pH is not compatible with a dermal application.
According to the Lipinski rule of five, it is more likely for a compound to exhibit an enhanced permeability if the number of hydrogen bond donors/acceptors is no more than 5 or 10, respectively, the molecular weight does not exceed 500 Da, and the LogP is not more than 5. Hydrocortisone, etofenamate, acyclovir, clobetasol, and diclofenac comply with these requirements, while the other active substances (antifungals) are expected to have poorer permeability [27].
Regarding lipophilicity, different profiles can be identified. Hydrocortisone and acyclovir exhibit moderate lipophilicity (0 < LogP < 3), while the other compounds are predominantly more lipophilic. The solubility of the compounds was carefully inspected in order to select the most appropriate IVRT conditions. Acyclovir and hydrocortisone, as previously mentioned, display the highest intrinsic solubility amongst the studied drugs (acyclovir is very soluble and hydrocortisone is freely soluble) [37]. Moreover, the solubility of both actives is not affected by the pH. These reasons support the selection of PBS (pH = 7.4) as the release media during IVRT experiments. Nevertheless, taking into account the information displayed in the Topical and Transdermal Drug Products Product Performance Tests section in the USP, in which an IVRT method for a hydrocortisone cream is presented, a water/ethanol (70:30, v/v) release media is suggested [23]. Therefore, for hydrocortisone, this particular release media was considered.
Diclofenac and etofenamate can be classified as sparingly soluble, hence the effects of two co-solvents, ethanol and propylene glycol, were tested. Etofenamate Log D displays no alteration within the considered pH range, therefore, a pH =7.4 was solely considered. The solubility was then assessed in the following conditions: PBS/co-solvent 80:20 and 70:30, v/v. On the other hand, diclofenac solubility is impacted by the pH, consequently, screening studies were performed at a pH values of 3.6, 5.5, and 7.4 in the condition PBS/co-solvent 80:20, v/v [11].
Clobetasol is considered slightly soluble and its Log D is not affected by pH. Taking this into account, the following conditions were tested: PBS/ethanol 70:30 and 50:50, v/v. Antifungal molecules present the lowest solubility, as they are considered as very slightly soluble molecules. As previously mentioned, these molecules do not exhibit change in the Log D at a physiological pH compatible with skin application. However, for tioconazole, at a pH = 4.6, the molecule evidences some degree of solubility. Therefore, the effect of the same co-solvents was screened in the following conditions: PBS/co-solvent 70:30, 60:40, and 50:50, v/v, for both bifonazole and clotrimazole. Alternatively, for tioconazole, PBS/co-solvent ratios of 80:20, 70:30, and 50:50, v/v, were tested at a pH = 7.4. Since ethanol provided the highest solubility results, the PBS/ethanol 80:20 and 50:50, v/v, conditions were repeated, but at a pH of 4.5. Solubility results are presented in Figure 2.
Ethanol was herein selected due to its prevalence in the literature, while the selection of propylene glycol was mainly related to the presence of this component in the qualitative formulation of some products. In fact, among the polyvalent alcohols, propylene glycol is the most frequently used co-solvent in topical products [11,38]. Nevertheless, for all drugs, ethanol demonstrated a higher solubility.
For etofenamate, PBS/ethanol pH = 7.4 at a 70:30, v/v, ratio was selected as the release medium. For bifonazole, clotrimazole, and clobetasol, a 50:50, v/v, ratio of PBS/ethanol pH = 7.4 enabled the optimal solubility conditions. For tioconazole, the same media displayed good results, however, a pH = 4.5 was selected. Finally, for diclofenac, a PBS/ethanol 80:20, v/v, at a pH = 7.4 was selected. Please note that diclofenac conditions were already optimized in a previous study directed at IVRT method optimization [11]. Ethanol was herein selected due to its prevalence in the literature, while the selection of propylene glycol was mainly related to the presence of this component in the qualitative formulation of some products. In fact, among the polyvalent alcohols, propylene glycol is the most frequently used co-solvent in topical products [11,38]. Nevertheless, for all drugs, ethanol demonstrated a higher solubility.
For etofenamate, PBS/ethanol pH = 7.4 at a 70:30, v/v, ratio was selected as the release medium. For bifonazole, clotrimazole, and clobetasol, a 50:50, v/v, ratio of PBS/ethanol pH = 7.4 enabled the optimal solubility conditions. For tioconazole, the same media displayed good results, however, a pH = 4.5 was selected. Finally, for diclofenac, a PBS/ethanol 80:20, v/v, at a pH = 7.4 was selected. Please note that diclofenac conditions were already optimized in a previous study directed at IVRT method optimization [11].

IVRT
The release profile provides useful information on the pharmaceutical performance characteristics of topical products, including nanotechnology-based formulations, since the API must be released before it becomes bioavailable in the skin. Moreover, from both scientific and regulatory perspectives, IVRT methods are responsive to product microstructure, making them a useful tool to assess product similarity [4,[39][40][41].
The analytical methods developed in this work were used to monitor drug concentrations in IVRT samples. Figure 3 shows the IVRT profiles.

IVRT
The release profile provides useful information on the pharmaceutical performance characteristics of topical products, including nanotechnology-based formulations, since the API must be released before it becomes bioavailable in the skin. Moreover, from both scientific and regulatory perspectives, IVRT methods are responsive to product microstructure, making them a useful tool to assess product similarity [4,[39][40][41].
The analytical methods developed in this work were used to monitor drug concentrations in IVRT samples. Figure 3 shows the IVRT profiles.
To meet formal requirements, topical product release mechanisms should follow the Higuchi model, in other words, drug release should exhibit linearity with respect to the square root of time [9,42]. As portrayed in Figure 3, for all products, this criterion was fulfilled [19]. To meet formal requirements, topical product release mechanisms should follow the Higuchi model, in other words, drug release should exhibit linearity with respect to the square root of time [9,42]. As portrayed in Figure 3, for all products, this criterion was fulfilled [19].

Conclusions
Considering the current regulatory background, product performance evaluation techniques, such as IVRT, are becoming increasingly common in all stages of topical products' lifecycles. In this context, one of the principal contributions of the present paper relies on the establishment of a portfolio of HPLC methods specifically tailored for commercially available topical products. The analytical methods presented here are specific, linear, accurate, reliable, and reproducible. Due to their simplicity and a high sample throughput potential, this database can constitute a useful tool for (i) fast screening of formulation performance, and (ii) timely analysis of batch compliance, supporting a routine implementation.
Lastly, an in silico approach was successfully developed to assist the IVRT design.

Conflicts of Interest:
The authors declare no conflict of interest.