Correlation between microstructure and bioequivalence in Anti-HIV ...

European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

Contents lists available at ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics journal homepage: www.elsevier.com/locate/ejpb

Research paper

Correlation between microstructure and bioequivalence in Anti-HIV Drug Efavirenz Cinira Fandaruff a, Marcos Antônio Segatto Silva a, Danilo Cesar Galindo Bedor b, Davi Pereira de Santana b, Helvécio Vinícius Antunes Rocha c, Luca Rebuffi d,e, Cristy Leonor Azanza Ricardo d, Paolo Scardi d, Silvia Lucia Cuffini a,f,⇑ a

Laboratório de Controle de Qualidade, Universidade Federal de Santa Catarina, Florianópolis, Brazil Departamento de Ciências Farmacêuticas, Universidade Federal de Pernambuco, Recife, Brazil c Laboratório de Sistemas Farmacêuticos Avançados, Instituto de Tecnologia em Fármacos/Farmanguinhos (FIOCRUZ), Rio de Janeiro, Brazil d Department of Civil, Environmental and Mechanical Engineering, University of Trento, Trento, Italy e Elettra-Sincrotrone Trieste, Trieste, Italy f Pós-Graduação em Engenharia e Ciências dos Materiais, Universidade Federal de São Paulo, São José dos Campos, Brazil b

a r t i c l e

i n f o

Article history: Received 23 September 2014 Accepted in revised form 24 January 2015 Available online 3 February 2015 Keywords: Anti-HIV Efavirenz Bioequivalence Microstructure Synchrotron Radiation Powder diffraction

a b s t r a c t Polymorphism and particle size distribution can impact the dissolution behaviour and, as a consequence, bioavailability and bioequivalence of poorly soluble drugs, such as Efavirenz (EFV). Nevertheless, these characteristics do not explain some failures occurring in in vitro assays and in in vivo studies. EFV belongs to Class II and the High Activity Antiretroviral Therapy (HAART) is considered the best choice in the treatment of adults and children. EFV is a drug that needs bioequivalence studies for generic compounds. In this work, six raw materials were analyzed and two of them were utilized with human volunteers (in vivo assays or bioequivalence). All the routine pharmaceutical controls of raw materials were approved; however, the reasons for the failure of the bioequivalence assay could not be explained with current knowledge. The aim of this work was to study microstructure, a solid-state property of current interest in the pharmaceutical area, in order to find an explanation for the dissolution and bioequivalence behaviour. The microstructure of EFV raw materials was studied by Whole Powder Pattern Modelling (WPPM) of X-ray powder diffraction data. Results for different EFV batches showed the biorelevance of the crystalline domain size, and a clear correlation with in vitro (dissolution tests) and in vivo assays (bioequivalence). Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction All solid-state characteristics of drugs can potentially impact their dissolution behaviour and, as a consequence, their bioavailability and bioequivalence. This is especially true in the case of poorly soluble drugs, for which crystalline structure or

Abbreviations: EFV, Efavirenz; HAART, High Activity Antiretroviral Therapy; WPPM, Whole Powder Pattern Modelling; UNAIDS, United Nations Program on HIV/ AIDS; HIV-1, human immunodeficiency virus type 1; BCS, Biopharmaceutical Classification System; API, active pharmaceutical ingredients; SLS, sodium lauryl sulphate; DE, Dissolution Efficiency; SD, standard deviation; SR, Synchrotron Radiation; XRPD, X-ray Powder Diffraction; SEM, Scanning Electron Microscopy. ⇑ Corresponding author. Pós-Graduação em Engenharia e Ciências dos Materiais, Universidade Federal de São Paulo, São José dos Campos, Brazil. Tel.: +55 1233099500. E-mail address: scuffi[email protected] (S.L. Cuffini). http://dx.doi.org/10.1016/j.ejpb.2015.01.020 0939-6411/Ó 2015 Elsevier B.V. All rights reserved.

polymorphism, particle size distribution, apparent density, flowability and compressibility are all carefully controlled by solid-state pharmaceutical routine protocols. However, other solid-state properties should be studied to understand some failures occurring in in vitro assays (dissolution tests) and in vivo studies (bioequivalence), which cannot be explained by the currently used protocols. According to the Joint United Nations Program on HIV/AIDS (UNAIDS), in 2012 the number of people living with AIDS worldwide was estimated at 35.3 million, and antiretroviral therapy averted 6.6 million AIDS-related deaths [1]. Efavirenz (EFV) was approved for the treatment of human immunodeficiency virus type 1 infection (HIV-1) in 1998 [2–4]. In the High Activity Antiretroviral Therapy (HAART), it is considered the best choice in the treatment of adults and children [5]. EFV is a Class II drug (low solubility, high permeability) according to the Biopharmaceutical Classification System (BCS) [6], showing

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

poor gastrointestinal absorption due to inadequate drug solubility in gastrointestinal fluids [2]. Besides that, the oral bioavailability of this drug is between 40% and 45% [5]. The generic version of EFV is a viable solution to offer quality medications to a much larger number of AIDS/HIV patients, provided that drug approval and registration are made according to existing national and international regulations covering the innovator’s version. In this context, bioequivalence assessment is the most important quality control tool in the process of a generic product development and registration, in order to ensure its therapeutic efficacy [7,8]. Even when the known critical conditions of purity and solid state characteristics were routinely controlled for the raw materials, significant differences were detected when administrated to healthy volunteers in bioequivalence studies. That was the case for two raw material batches of EFV studied in this work, used in bioequivalence studies where one was approved while the other one failed. The bioequivalence studies were conducted on volunteers whose age ranged between 18 and 45 years. The analytical tests and details of these studies were reported by Bedor et al. [9] and Honorio et al. [10], who followed the Brazilian protocol for bioequivalence studies [11], in accordance also with international recommendations for this kind of evaluation. Apart from the in vivo studies, so far there are no clear explanations for the dissolution (in vitro) tests with the specific conditions used in this work. These results showed significantly different behaviour from batch to batch, after the micronization process, for reasons that were not understood. Therefore, it is important to understand this discrepancy not only in vitro, but also in vivo in order to guarantee the bioequivalence of generic drugs and the reproducible quality in batch to batch production. It is crucial to acquire a deeper physicochemical knowledge to control the properties and solid-state characteristics of the pharmaceutical raw materials. Useful indications might be obtained by a detailed study of the micronization process, in particular of the mechanical effects on the microstructure of active pharmaceutical ingredients (APIs). This is a well-known subject in Materials Science, where the correlation between microstructure and performance of materials belonging to all major classes such as ceramics, metals and alloys has long been recognized [12,13]. So far, none of these concepts has been appropriately addressed in the pharmaceutical sciences. The aim of this study was to demonstrate the effect of microstructure of EFV (AIDS drug) raw material, expressed in terms of correlation of crystalline domain size with both in vitro assays (dissolution profiles) and in vivo studies of human oral bioequivalence. 2. Materials and methods 2.1. Materials Six EFV batches were provided by two Brazilian institutions, the governmental pharmaceutical laboratories Farmanguinhos-FIOCRUZ and LAFEPE. All batches were micronized. During the bioequivalence studies, raw material bathes 1 and 5 were used, respectively, in the approved and not approved biobatches. The reference drug product (StocrinÒ, 600 mg tablets) was commercially available at the time of the study. Details of clinical data from the approved biobatch can be found in Honorio et al. [10], and for the not approved biobatch in Bedor et al. [9]. Prior to volunteer recruitment, all protocols were approved by the research ethics committee, as detailed in each specific study. The bioanalytical conditions used in the bioequivalence studies can be found in the same publications, also following ANVISA recommendations at the time of the studies [14], which closely adhere to the international guidelines.

53

2.2. Dissolution profiles EFV raw materials (200 mg of pure drug) were added to 300 ml of sodium lauryl sulphate 0.25% (SLS, dissolution medium) at 37 ± 0.5 °C. The test was made with 75 rpm stirring in standard dissolution equipment (Nova Ética, Brazil). The apparatus 2 (USP) was used and samples of 5 ml were withdrawn at 5, 10, 15, 20, 30, 45, 60, 90 and 120 min. The medium was refilled with an equal amount of fresh solution to maintain a constant total volume. The specimens were analyzed by high-performance liquid chromatography. The mobile phase consisted of acetonitrile: ammonium acetate buffer pH 7.5 (50:50 v/v) and wavelength detection at 252 nm. The other chromatographic conditions were: a PerkinElmerÒ C18 (150 mm  4.6 mm, 5 lm) column, flow rate of 1 ml/min and injection volume of 20 ll. The Dissolution Efficiency (DE) was calculated as the area under the dissolution curve (AUC0–120) up to a certain time, t, expressed as a percentage of the area of the rectangle described (AUCTR) by 100% dissolution over the same time interval [15]. The DE was calculated to compare the relative performance of the six batches. Results were expressed as mean values ± standard deviation (SD). Statistical comparisons were made by Student’s t-test using the GRAPH PAD PRISM INSTAT Program (San Diego, CA, USA), considering P < 0.05 to be significant. 2.3. X-ray Powder Diffraction (XRPD) XRPD pattern was recorded on a XPERT PANalytical diffractometer, equipped with X’Celerator detector, using Ni filtered ka radiation from a Cu tube operating with 40 kV and 45 mA, 2 Theta range from 5–30°, scan step size of 0.033° and scan step time of 45 s. The Soller, divergent and antiscattering slits used were 0.04 rad and 0.25E respectively. 2.4. Synchrotron Radiation (SR) X-ray Powder Diffraction (XRPD) Structural and microstructural analysis was based on SR XRPD measurements made at the MCX beamline of the Italian synchrotron Elettra-Sincrotrone Trieste, using the Debye–Scherrer (capillary) geometry [16]. Specimens of the six EFV batches were loaded in KaptonÒ capillaries and measurements were carried out in duplicate. Structural information was obtained by modelling the data with the software TOPASÓ [17,18]. All samples were identified as EFV polymorph 1, with space group P21212 and cell parameters a = 16.781 Å, b = 27.258 (Å), c = 9.698 (Å) (data collected at 250 K [19]). This structural information was used as initial model in the refinement of data collected at 298 K, giving a = 16.88(1) Å, b = 27.335(5) (Å), c = 9.765(2) (Å), where the number in parentheses is standard deviations of the distribution of values in the six batches. Microstructural information was provided by the analysis of the diffraction line profiles, using the software PM2K [20,21], implementing the Whole Powder Pattern Modelling (WPPM) approach [22–24]. WPPM is based on a physical model of the microstructure to generate theoretical expressions for the line profiles, and relies on the assumption that the observed diffraction line profile is a convolution of profile components produced by all contributing effects, such as the instrumental profile, coherent scattering domains size/shape, lattice distortions, etc. Microstructural parameters can then be obtained from a non-linear least squares fitting of the experimental powder pattern, using Fourier Transforms of the individual profile components to handle the usually complex convolution problem. Details can be found in the literature [22–24]: here the method was applied representing the EFV crystallites as equiaxed crystalline domains. The latter were modelled as spheres with a lognormal distribution of diameters D (hDi = d[1, 0] = arithmetic mean, hDis = d[3, 2] = surface-weighted mean, hDiv = d[4, 3] = volume-weighted mean). The modelling then

54

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

provided lognormal mean (l) and variance (r) from which arithmetic mean and all other possible mean sizes can be easily obtained. We also considered the possible presence of a strain broadening contribution, based on a rather general strain model constrained by the symmetry of the elastic tensor for the specific Laue group of EFV [21,24,25]. However, no significant improvement to the modelling was observed by adding this component, which can therefore be considered as negligible in the present case. 2.5. Differential Scanning Calorimetry (DSC) and thermogravimetric analysis (TGA) DSC curve of EFV was obtained using a DSC-60 cell (Shimadzu) in aluminium pan containing about 2 mg of sample, under a dynamic nitrogen atmosphere (50 ml/min) at a heating rate of 10 °C/min from 25 to 350 °C. The DSC cell was calibrated using a standard reference, Indium. The TG measurement was obtained with a thermobalance model TGA-50 (Shimadzu) in temperature range above described, using platinum crucible with sample of 4.5 mg, under dynamic nitrogen atmosphere (50 ml/min) and heating rate of 10 °C/min. 2.6. Scanning Electron Microscopy (SEM) The morphology of the raw materials was evaluated by SEM. Specimens were mounted on metal stubs using double-sided adhesive tape, vacuum-coated with gold (350 Å) in a Polaron E-5000, and analyzed in a Philips (model XL 30) instrument, with a voltage of 15 kV and 3000 magnification. 2.7. Particle size distribution The particle size distribution of EFV micronized batches was determined by Laser diffraction, using the wet mode (Malvern Mastersizer 2000, Hydro 2000 SM, Worcestershire, UK). Specimens were prepared by dispersing approximately 10 mg of drug in a beaker containing 10 ml 0.02% (w/v) polysorbate 80 solution. A volume of 100 ml water was used as the dispersion medium and the mixture was stirred at 2.000 rpm. This technique provides d[3, 2], d[4, 3] and the Span values: the former is the surface-weighted mean size, directly related to the surface area; d[4, 3] is the volume-weighted mean size, or mass moment mean diameter, whereas Span is a measure of the size distribution width (the narrower the distribution the smaller the Span). The Span definition used in this paper was given by Glover et al. [26] as the difference in particle diameters at 10% and 90% of the cumulative volume, d(0.1) and d(0.9) respectively, divided by the volume median diameter d(0.5): Span = [d(0.9)  d(0.1)]/d(0.5). 2.8. Fourier Transform Infrared (FT-IR) Spectroscopy FT-IR spectra of EFV were recorded using FT-IR (Prestige-21 Shimadzu), in KBr (potassium bromide) discs over the range of 4000–1000 cm1. 2.9. Impurities profile The impurities profile of all the batches was tested according to the USP 38 [27] EFV official monograph. All the impurities were tested based on the relative retention time and the relative response factor established on the monograph. The chromatographic conditions followed the pharmacopeic recommendations. There was used a HPLC Shimadzu and the results were evaluated with the software of the same equipment (Shimadzu Class – VP version 6.13 SP2).

3. Results 3.1. Dissolution behaviour The powder dissolution profiles of the six EFV batches in SLS are shown in Fig. 1A, with the corresponding Dissolution Efficiencies (DEs) plotted in Fig. 1B. The dissolution assay conditions highlight the differences between biobatches, i.e., batches 1 and 5. Batch 1, a positive result in the in vivo bioequivalence test, showed the best dissolution behaviour with the highest DE (82%); in contrast, batch 5, which did not pass the bioequivalence test, was around 66%.

3.2. Solid state properties The six batches of EFV were analysed by several techniques (XRPD, DSC-TG and FT-IR) in order to do a complete solid state characterization (Fig. 2A). However, these results did not evidence any difference between the samples. Furthermore, the impurity profiles of all batches presented acceptable results and were approved by the USP pharmacopeia [27] (data not shown). The origin of the DE behaviour here observed is also commonly explained by other factors, such as morphology, particle size distribution and/or impurities [28–30]. The morphology was not the main cause of DE results since the SEM micrographs in Fig. 2B show that the morphology of EFV grains is rod-like, with similar shape (length to width) ratio in all batches. Some authors have illustrated the influence of impurities on crystal growth and also on the final crystalline properties of the substances [31–33]. In the present paper, the impurity profiles cannot be traced as a relevant influence in the dissolution behaviour. Moreover, increase in the impurities concentration can be detected as a consequence of the micronization process [34]. In our case, the differences in the micronization process for each sample did not change the impurity profile and so that it is possible to exclude this parameter as the reason of the dissolution behaviour observed. Particle size distribution parameters are reported in Table 1. Batch 1 presents the lowest value of average particle size. d[4, 3] values for batches 2–6 are higher, although still acceptable in a pharmaceutical micronization routine process. In general, it is expected that smaller d[4, 3] values correspond to higher values of dissolution profile and DE [26], a correlation actually verified for batches 1 and 6, which lay at the two extremes of both size and DE ranges. However, contrary to this expectation, batch 2 shows a higher DE (70%) than batches 3, 4, and 5, all below 66% (see Table 1). It is important to highlight that, even if the mean particle size of batch 1 (bioequivalent) is smaller than batch 5 (not bioequivalent), the latter has a d[4, 3] value well within the acceptable range. Therefore, neither morphology nor particle size characteristics can account for the in vitro differences and in vivo assays. Furthermore, some unexpected relations between particle size distribution and dissolution behaviour were observed. From a pharmaceutical technology point of view, the particle size parameters most used traditionally to correlate with dissolution and bioavailability results are volume-based ones [35], and in particular d[4, 3] seems to be the most adequate [26]. Based on theoretical calculations, the so-called volumeweighted mean diameter (d[4, 3]) should have the best correlation with the average rate constant of dissolution [28]. Even the discrimination in d(0.9), d(0.5) and d(0.1) does not class batches 1 and 5 in two opposite extremes. So, according to the results and to the routine quality control specifications, all six batches would be approved if just particle size

55

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

Fig. 1. (A) Powder dissolution profile in SLS 0.25% versus time. (B) Dissolution Efficiency of six EFV batches of raw materials. Batch 1 passed the bioequivalence test, whereas batch 5 did not. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2. (A) XRPD, DSC-TG and FT-IR of EFV polymorph 1. (B) SEM micrographs of EFV raw material batches 1–6.

Table 1 Particle and crystalline domain size analysis of the six batches of EFV and corresponding values of Dissolution Efficiency. See text for details. EFV raw material batch

Dissolution Efficiency (%)

hDi (nm)

SD (nm)

Average particle size d[4, 3] (lm)

Average particle size d[3, 2] (lm)

d(0.9) (lm)

d(0.5) (lm)

d(0.1) (lm)

Span

1 2 3 4 5 6

82 ± 7 70.0 ± 0.7 66.1 ± 0.5 62.1 ± 0.8 66.3 ± 0.3 61.1 ± 0.1

30 ± 3 64 ± 7 95 ± 11 133 ± 22 208 ± 42 282 ± 56

22 41 51 55 74 80

2.6 5.2 4.4 4.4 4.0 8.8

2.1 2.1 2.3 2.0 2.2 3.4

5.3 13.2 10.1 9.5 8.2 19.1

1.9 2.8 3.0 2.5 3.1 6.2

0.7 1.0 1.1 1.0 1.0 1.4

2.5 4.3 3.0 3.4 2.3 2.8

distributions were determined as the main parameter for raw material analysis before production. Obviously, if DE with the specific medium conditions presented in this work was introduced as a new quality control parameter, batch 1 could be clearly differentiated from the other batches. However, in this case, it would seem contradictory according to the current, accepted paradigm, in which dissolution is directly correlated with particle size distribution. In our case, the mean particle size does not correlate, either with the (in vitro) dissolution profiles or with the (in vivo) bio-

equivalence results. Takano et al. propose d[3, 2] should correlate with dissolution [36]. However, they assume particles are spherical in their model, which is a strong simplification. Moreover, coefficients of determination for d[3, 2] and d(0, 5) do not differ significantly in their analysis. As shown in Fig. 3(A and B) our data show no significant correlation between DE and d[4, 3] or d[3, 2]. This evidence calls for a new explanation, which might well be achieved by studying other solid state and/or mechanical properties.

56

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

Fig. 3. DE versus average particle size, and DE versus average crystalline domain size, for six batches of EFV raw materials. Batch 1 passed the bioequivalence test, whereas batch 5 did not. (A) DE versus average particle size d[4, 3]. (B) DE versus average particle size d[3, 2]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. (A) Schematic definition of particle, cluster and crystalline domain (crystallite) size. (B) Example of the XRPD data analysis by the WPPM approach (EFV batch 5): data (circle), model (line) and their difference, or residual (line below). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

For all those reasons, it was not possible to establish clear evidences to understand and explain the DE results. Therefore, another parameter, not analysed by these solid state techniques and routine pharmaceutical analyses must be evaluated. It is worth recalling here the distinction between terms used so far to describe sizes. Particle size is used to refer to the size of any small, localized object, labelled as P in the scheme of Fig. 4. Based on SEM micrographs, P might be considered regions where the localized shapes show visible contours.

3.3. Microstructure Particles may show an internal structure, highlighted in Fig. 4, made of smaller crystalline domains.1 Diffraction is sensitive to the size of the crystalline domains (or crystallites), which are regions of coherent scattering (labelled as D in the scheme of Fig. 4). These may be portions of apparently larger crystals, whenever the microstructure is such that it interrupts coherent scattering across a larger object [24]. Dissolution and bioavailability of pharmaceutical raw materials and products have been traditionally correlated with several physicochemical characteristics, such as crystallinity/amorphism, polymorphism, particle size, surface charge, density, porosity, to cite a few [37,38]. So far crystalline domain size has not been an issue for pharmaceutical analysts and formulators, and so microstructure properties to the level of crystallite size have not been commonly 1 A crystalline domain is, in quite general terms, a region of a polycrystalline material which is coherently scattering.

correlated with biopharmaceutical characteristics or process considerations. Torrado and collaborators [39] have mentioned the importance of crystallite size in understanding dissolution results, but their work does not consider any aspect of particle size. It is then impossible to distinguish between crystallite and particle size, although the authors maintained that crystallite size impacted their results. Other authors propose correlations of physical and mechanical properties with crystallite size, mainly focusing on tableting/compression process using the Scherrer formula [40,41]. Riippi et al. [41] proved the influence of mechanical stress on crystallite size and so it is possible to suppose that micronization will proceed in the same way, considering that this process does also impose stress on the powder. Therefore, micronization is a good alternative for bioavailability enhancement but even in this kind of situation the solid state characterization is focused only on evaluating possible amorphization or disorder induced in the drug [42], surface modifications [43], occurrence of phase transformations and in general, stability during processing [44]. Recently, Fornico et al. [45] also mentioned the crystallite size parameter after micronization of the drug. Prompted by the increase of research and market launch of nanocrystalline drug products, a discussion has started about the correlation of nanoparticle size with dissolution and solubility. Some authors claim that, according to the Ostwald–Freundlich equation, solubility saturation can be enhanced if particle size falls below 100 nm, because of the effect of surface curvature [46,47]. Even in recent works, the correlation is based only on particle and/or nanocrystal sizes [48]. Perhaps in this size range of crystals, it is possible that crystalline domains influence drug dissolution and bioavailability.

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

Fig. 5. Crystalline domain size distributions for the six EFV batches of this study. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. DE versus average crystalline domain size hDi (nm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

57

non-random discrepancy is observed between data and model, as a possible result of more complex dispersion of the domain size, or non-spherical domain shapes. Using this methodology, besides mean size values, the distribution can be determined. Fig. 5 shows the crystalline domain size distributions for all the measured specimens of this study. Arithmetic mean size (hDi, also known as d[1, 0]) and SDs are reported in Table 1. Beyond fine differences and possible discrepancies between data and models, the crystalline domain size was found to span quite different values, from a few tens to hundreds of nanometres. In order to determine the effect of microstructure in the dissolution assays and bioequivalence studies, DE was compared with the average crystalline domain size. Fig. 6 (see also Table 1) clearly shows the correlation: larger crystalline domain sizes (above 100 nm) correspond to the lower DE values, as opposed to results in Fig. 3A and B, showing no clear correlation between DE and average particle sizes, d[4, 3] and d[3, 2], mainly for batches with comparable particle sizes. Fig. 7 summarizes these results in a 3D representation, showing the relationship between average particle sizes, crystalline domain sizes and DE. For batches with comparable particle size distribution (2–5), DE is clearly related to the crystalline domain size (Figs. 6 and 7). Most importantly, the significant difference in bioequivalence essays or in vivo studies presented by batches 1 and 5 can be explained by this evidence. Batch 1, with the approved bioequivalence essay, shows the smallest crystalline domain size (D = 30 ± 3 nm) whereas batch 5, with a not approved bioequivalence assay, has crystalline domains more than 6 times larger (D = 208 ± 42 nm). Finally, an important remark: the present results suggest that there could be a critical crystalline domain size and particle size distributions to ensure the bioequivalence. Further studies will be required to gather additional evidence and also to establish the biorelevance of the microstructure in other pharmaceutical compounds. 4. Conclusion The biorelevance of the microstructure in a poorly water-soluble drug such as EFV was investigated in the light of the correlation between crystalline domain size distributions and bioequivalence studies. Besides the scientific relevance, this result is an alert that not only the polymorphism, particle size distribution and surface properties but also the microstructure would modify the solidstate properties of pharmaceutical drugs, their dissolution behaviour and bioavailability. Therefore, a deeper knowledge of the crystalline domain size distribution should be obtained during pharmaceutical developments, in order to determine whether the microstructure is critical for bioavailability and bioequivalence success. Conflict of interests The authors have declared that there are no conflict of interests regarding this manuscript.

Fig. 7. Synoptic 3D view of the relationship among average crystalline domain size, average particle size and dissolution efficacy. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

From our point of view, the investigation of microstructure opens a new and broad field in the pharmaceutical area. Fig. 4 also shows a typical result of the PM2K software, based on the WPPM approach described above. The result is acceptable, although some

Acknowledgements We are grateful to MSc Livia Deris Prado and Dr. Joel Bernstein for the review of the technical aspects of the paper and also to the Laboratory of Solid State Studies (Farmanguinhos) for the particle size analysis. The XRPD and SEM measurements were performed at Laboratório de Difração de Raios-X (LDRX) and Laboratório Central de Microscopia Eletrônica (LCME) at UFSC. For the financial

58

C. Fandaruff et al. / European Journal of Pharmaceutics and Biopharmaceutics 91 (2015) 52–58

support for the development of the present study, the authors acknowledge CNPq, CAPES, FAPESC, PDTIS/FIOCRUZ. This work was partly supported by Project 2013-0247 ‘‘Mechanical activation to improve bioavailability and to reduce adverse effects of drugs in the therapy of chronic diseases’’, Fondazione Caritro, Trento. We kindly thank the MCX beamline staff at Elettra-Sincrotrone Trieste for their support. We are sincerely grateful to Dr. Kevin Charles Prince for his final revision of this paper.

References [1] UNAIDS, Joint United Nations Programme on HIV/AIDS, 2013. (accessed 02.03.14). [2] S. Sathigari, G. Chadha, Y.H.P. Lee, N. Wright, D. Parsons, V. Rangari, O. Fasina, R.J. Babu, Physicochemical characterization of Efavirenz-cyclodextrin inclusion complexes, AAPS PharmSciTech 10 (2009) 81–87, http://dx.doi.org/10.1208/ s12249-008-9180-3. [3] J.A. de Aquino Ribeiro, L.M.M. de Campos, R.J. Alves, G.P. Lages, G.A. Pianetti, Efavirenz related compounds preparation by hydrolysis procedure: setting reference standards for chromatographic purity analysis, J. Pharm. Biomed. Anal. 43 (2007) 298–303. http://dx.doi.org/10.1016/j.jpba.2006.06.010. [4] FDA, in: CDER Fast Track Products Approved Since 1998 through June 1, 2010. (accessed 15.12.10). [5] D.A. Chiappetta, C. Alvarez-Lorenzo, A. Rey-Rico, P. Taboada, A. Concheiro, A. Sosnik, N-alkylation of poloxamines modulates micellar assembly and encapsulation and release of the antiretroviral efavirenz, Eur. J. Pharm. Biopharm. 76 (2010) 24–37. http://dx.doi.org/10.1016/j.ejpb.2010.05.007. [6] G.L. Amidon, H. Lennernäs, V.P. Shah, J.R. Crison, A theoretical basis for a biopharmaceutic drug classification: the correlation of in vitro drug product dissolution and in vivo bioavailability, Pharm. Res. 12 (3) (1995) 413–420. [7] European Medicine Agency, 2010. (accessed 09.03.14). [8] Brasil, Resolução n° 391, de 9 de agosto de 1999. (accessed 09.03.14). [9] D.C.G. Bedor, J.H.S. Filho, V.L.S. Ramos, T.M. Gonçalves, C.E.M. Sousa, D.P. Santana, A sensitive and robust LC–MS/MS method with monolithic column and electrospray ionization for the quantitation of efavirenz in human plasma: application to a bioequivalence study, Quim. Nova 34 (6) (2011) 950–955. [10] T.S. Honório, E.C. Pinto, H.V.A. Rocha, V.S.D. Esteves, T.C. Santos, H.C. Casto, C.R. Rodrigues, V.P. Sousa, L.M. Cabral, In vitro–in vivo correlation of efavirenz tablets using GastroPlusÒ, AAPS PharmSciTech 14 (3) (2013) 1244–1254. [11] Brasil, Resolução 1170 – Guia para provas de biodisponibilidade relativa/ bioequivalência de medicamentos, 2006. (accessed 29.06.14). [12] E.J. Mittemeijer, Fundamentals of Materials Science, Springer-Verlag, Berlin Heidelberg, 2010. http://dx.doi.org/10.1007/978-3-642-10500-5. ISBN 978-3642-10499-2. [13] J.C.M. Li, Microstructure and Properties of Materials, World Scientific Publishing, Singapore, vol. 1, 1996 (ISBN: 978-981-02-2403-5) and vol. 2, 2000 (ISBN: 978-981-02-4180-3). [14] Brasil, Resolução 899, Guia para validação de métodos analíticos e bioanalíticos, 2003. (accessed 29.06.14). [15] P. Costa, L. Sousa, Modeling and comparison of dissolution profiles, Eur. J. Pharm. Sci. 13 (2) (2001) 123–133. [16] L. Rebuffi, J.R. Plaisier, M. Abdellatief, A. Lausi, P. Scardi, MCX: a synchrotron radiation beamline for X-ray diffraction line profile analysis, Z. Anorg. Allg. Chem. 640 (15) (2014) 3100–3106. http://dx.doi.org/10.1002/zaac.201400163. [17] Bruker-AXS, ‘‘TOPAS Version 3, Academic’’. and . [18] R.W. Cheary, A. Coelho, A fundamental parameters approach to X-ray lineprofile fitting, J. Appl. Crystallogr. 25 (1992) 109–121. [19] S. Mahapatra, T.S. Thakur, S. Joseph, S. Varughese, G.R. Desiraju, New solid state forms of the Anti-HIV drug efavirenz. conformational flexibility and high Z0 issues, Cryst. Growth Des. 10 (2010) 3191–3202, http://dx.doi.org/10.1021/ cg100342k. [20] M. Leoni, T. Confente, P. Scardi, PM2K: a flexible program implementing Whole Powder Pattern Modelling, Z. Kristallogr. Suppl. 23 (2006) 249–254. [21] P. Scardi, M. Ortolani, M. Leoni, WPPM: microstructural analysis beyond the Rietveld method, Mater. Sci. Forum 651 (2010) 155–171.

[22] P. Scardi, M. Leoni, Diffraction line profiles from polydisperse crystalline systems, Acta Crystallogr. A A57 (2001) 604–613. [23] P. Scardi, M. Leoni, Whole powder pattern modelling, Acta Crystallogr. A A58 (2002) 190–200. [24] P. Scardi, Microstructural properties: lattice defects and domain size effects, in: R.E. Dinnebier, S.J.L. Billinge (Eds.), Powder Diffraction: Theory and Practice, The Royal Society of Chemistry, Cambridge, 2008, pp. 376–413. ISBN 978-0-85404-231-9 (Chapter 13). [25] M. Leoni, J. Martinez-Garcia, P. Scardi, Dislocation effects in powder diffraction, J. Appl. Cryst. 40 (2007) 719–724. [26] W. Glover, H.K. Chan, S. Eberl, E. Daviskas, J. Verschuer, Effect of particle size of dry powder mannitol on the lung deposition in healthy volunteers, Int. J. Pharm. 349 (2008) 314–322. [27] USP The United States Pharmacopeial, Efavirenz Official Monograph, 2014. [28] A.P. Tinke, K. Vanhoutte, R. Maesschalck, S. De Verheyen, H. De Winter, A new approach in the prediction of the dissolution behavior of suspended particles by means of their particle size distribution, J. Pharm. Biomed. Anal. 39 (2005) 900–907. [29] N. Blagden, M. Matas, P.T. Gavan, P. York, Crystal engineering of active pharmaceutical ingredients to improve solubility and dissolution rates, Adv. Drug Deliv. Rev. 59 (2007) 617–630. [30] S. Willmann, K. Thelen, C. Becker, J.B. Dressman, J. Lippert, Mechanism-based prediction of particle size-dependent dissolution and absorption: cilostazol pharmacokinetics in dogs, Eur. J. Pharm. Biopharm. 76 (2010) 83–94. [31] N. Kubota, Effect of impurities on the growth kinetics of crystals, Cryst. Res. Technol. 36 (2001) 749–769. [32] C.L. Burcham, D.J. Jarmer, Industrial Crystallization of Pharmaceuticals: Capability Requirements to Support an Outsourcing Paradigm. (accessed 12.12.14). [33] S. Bordawekar, Z. Kuvadia, P. Dandekar, S. Mukherjee, M. Doherty, Interesting morphological behavior of organic salt choline fenofibrate: effect of supersaturation and polymeric impurity, Cryst. Growth Des. 14 (8) (2014) 3800–3812. [34] M. Aoki, H. Nishimura, A. Mimura, S. Kita, T. Yasuzawa, K. Terada, Identification of the degradation products of the steroid sulfatase inhibitor KW-2581 in jet mill-micronized powder, J. Pharm. Sci. 102 (6) (2013) 1760–1772. [35] M. Mosharraf, C. Nyström, The effect of particle size and shape on the surface specific dissolution rate of micronized practically insoluble drugs, Int. J. Pharm. 122 (1995) 35–47. [36] R. Takano, N. Takata, K. Shiraki, S. Higo, Y. Hayashi, S. Yamashita, A theoretical– empirical analysis on the initial dissolution rate of drugs from polydispersed particles, Biol. Pharm. Bull. 32 (11) (2009) 1885–1891. [37] L. Shargel, A.B.C. Yu, Applied Biopharmaceutics & Pharmacokinetics, fourth ed., Appleton & Lange, 1999. [38] G. Brittain, Preparation and identification of polymorphs and solvatomorphs, in: Moji C. Adeyeye, H.G. Brittain (Eds.), Preformulation in Solid Dosage form Development, Drugs and the Pharmaceutical Sciences, vol. 178, Informa Healthcare USA Inc., New York, NY, 2008. [39] G. Torrado, S. Fraile, S. Torrado, Process-induced crystallite size and dissolution changes elucidated by a variety of analytical methods, Int. J. Pharm. 166 (1998) 55–63. [40] E. Fukuoka, M. Makita, S. Yamamura, Preferred orientation of crystallites in tablets. III. Variations of crystallinity and crystallite size of pharmaceuticals with compression, Chem. Pharm. Bull. 41 (3) (1993) 595–598. [41] M. Riippi, V.P. Tanninen, J. Yliruusi, Effect of compression force on the crystal properties of erythromycin acistrate tablets, Eur. J. Pharm. Biopharm. 50 (2000) 365–371. [42] T. Feng, R. Pinal, M.T. Carvajal, Process induced disorder in crystalline materials: differentiating defective crystals from the amorphous form of griseofulvin, J. Pharm. Sci. 97 (8) (2008) 3207–3221. [43] D. Olusanmi, D. Jayawickrama, D. Bu, G. McGeorge, H. Sailes, J. Kelleher, J.F. Gamble, U.V. Shad, M. Tobyn, A control strategy for bioavailability enhancement by size reduction: effect of micronization conditions on the bulks, surface and blending characteristics of an active pharmaceutical ingredient, Powder Technol. 258 (2014) 222–233. [44] G.G.Z. Zhang, D. Law, E.A. Schmitt, Y. Qiu, Phase transformation considerations during process development and manufacture of solid oral dosage forms, Adv. Drug Deliv. Rev. 56 (2004) 371–390. [45] R.G. Fornico, G.R. Williams, J. Brum, F.G. Vogt, The study of micronization induced disorder in active pharmaceutical ingredients, Am. Pharm. Rev. October 2013. (accessed 27.06.14). [46] C. Brough, R.O. Williams, Amorphous solid dispersions and nano-crystal technologies for poorly water-soluble drug delivery, Int. J. Pharm. 453 (2013) 157–166. [47] B. Van Eerdenbrugh, J. Vermant, J.A. Martens, L. Froyen, J. Van Humbeeck, G. Van den Mooter, P. Augustijns, Solubility increases associated with crystalline drug nanoparticles: methodologies and significance, Mol. Pharm. 7 (5) (2010) 1858–1870. [48] X. Zhai, J. Lademann, C.M. Keck, R.H. Müller, Nanocrystals of medium soluble actives – novel concept for improved dermal delivery and production strategy, Int. J. Pharm. 470 (2014) 141–150.