Cryo-crystallization under a partial anti-solvent environment as a facile technology for dry powder inhalation development

Abstract

Industrial research in the field of dry powder inhalation (DPI) technology is still confined to the micronization of active pharmaceutical ingredients (APIs). Earlier studies have identified triboelectric charge as the main reason for the reduced efficiency of most of the marketed formulations. The physical instability of the micronized product aggravates the situation. Direct conversion of an API into an inhalable dry crystal, without applying an attritional force, would be a solution for many issues such as loss of crystallinity or metal contamination of the final product. In this work, Isoniazid (INH) (an anti-TB drug) was converted to inhalable particles using cryo-crystallization under a partial anti-solvent environment. The crystals were dried using a lyophilization technique and furthermore, the powder was characterized. PXRD studies showed that a new polymorph was evolved by the proposed technique. Since lyophilization is a closed method, microbial contamination can also be avoided through this method. Accelerated stability studies at 40 °C with 75% relative humidity indicated that the newly formed crystals had similar stability to that of the raw material (the API).

---

Introduction

The legacy of the jet-milling strategy of active pharmaceutical ingredients (APIs) for making dry powder inhalation formulations has lasted for decades due to the ease of scale-up and technology transfer. Most commercially available formulations employ the above strategy but many of them have a dose deposition efficiency of less than 20%.¹ Jet-milling and other shear force indulging processes are notorious for triboelectric charge accumulation on the product.² This triboelectric charge causes excessive agglomeration of the milled particles, which then resist separation during the aerolization process. This imparts a defective dose deposition and batch to batch variability. Milling often leads to amorphization of the product either completely or partially.^3,4 This amorphous material often recrystallizes during storage. In addition to that, recrystallization may cause solid bridges between the particles, resulting in particle size growth (Ostwald ripening). In a disordered state (amorphous state) the stability is also reported to be less than that of the crystalline counterpart.

Spray drying (SD) evolved as an alternative strategy for jet-milling in inhalable particle generation. Various biodegradable or biocompatible polymers are reported to be effective in delivering a drug to the alveoli.^5–7 Considering the limitations in the clearing mechanisms of the lungs, no polymer has been approved by regulatory bodies for an inhalational application.⁸ DPI for antimicrobial indication needs a high dose delivery per administration.⁹ Apart from that, the product obtained after SD will be mostly in an amorphous state. So, many of the disadvantages of jet milling, such as amorphization or lower chemical and physical stabilities, still persist using this method too.

Crystal engineering is a new arena which can contribute much to DPI technology.¹⁰ Anti-solvent addition to a drug solution is widely utilized for fabricating crystals with desired properties.^11,12 Since the precipitation is a fast process, it often results in crystals with a wide range of particle size. Controlled crystallization under the presence of a polymer helps to reduce the variation in particle size.¹³ But polymers present in the final product, even in minute quantities, may be detrimental for the DPI product. Dhumal et al. crystallized salbutamol sulfate using anti-solvent addition under ultrasonication.¹⁴ Crystals were later separated from the solution through spray drying. The NANOEDGE™ technology employs an additional annealing step to convert the amorphous part into a crystalline state after the usual anti-solvent crystallization step.¹⁵ The complete removal of the amorphous material helps to keep away the possibility of crystal growth in storage. Cryo-conditions are reported to be helpful in the milling process for attaining smaller particle sizes.^16,17

From earlier studies it’s clear that an inhalational anti-TB therapy would be promising for the treatment of tuberculosis.^18,19 The spray dried capreomycin DPI formulation for anti-TB therapy, developed by Hickey et al., was successfully entered into clinical trials.²⁰ Though Isoniazid is a potent anti-TB drug, recent studies show that the main reason for the emergence of drug resistance is the defective delivery of the drug to the target. Through clinical trials, Katiyar et al. proved that a high dose of Isoniazid can cure even multi-drug resistant tuberculosis (MDR-TB).²¹ The conventional treatment regime, resorting to a prolonged high oral dose, often leads to patient non-compliance. Isoniazid is reported to have side effects such as hepato- and neuro-toxicities. In the present study, we have made a serious attempt to develop inhalable particles of Isoniazid. A DPI delivery system containing Isoniazid in purely crystalline inhalable particles could drastically increase the concentration of the drug in the vicinity of the pathogen, thus controlling and preventing the spread of the disease.

Experimental section

Materials

Isoniazid was obtained as a generous gift from Lupin research park, Pune, India. High performance liquid chromatography grade acetonitrile (Lichrosolv, Merck) was used as the anti-solvent and also for HPLC analysis. Milli-Q water of 18.2 MΩ cm at 25 °C was used for the studies. A size 2 capsule made of hydroxyl propyl methyl cellulose (HPMC) was purchased from ACG capsules (Mumbai, India). Inhalac 230 was provided by Meggle Pharma (Hamburg, Germany) as a gift sample. For the buffer preparation, disodium orthophosphate (Na₂HPO₄·2H₂O) from Merck, India was used.

Isoniazid crystal synthesis

Two batches of crystals were prepared at two different concentrations, keeping all of the other parameters identical. In the first batch, 100 mg of the drug was dissolved in 3 ml of water, which was poured into 50 ml of acetonitrile (HPLC grade) while stirring. We selected these solvents and optimized the proportions in such a way that the addition of the Isoniazid solution to the acetonitrile wouldn’t induce crystallization under room temperature (approx. 25 °C). This batch was to be called high concentration Isoniazid (HC INH). One more batch was prepared by reducing the Isoniazid quantity to half (50 mg) while all of the other parameters were the same. Henceforth, this batch was named LC INH (low concentration Isoniazid).

The clear drug solutions in the above co-solvent mixtures were super cooled by plunging the vessel into liquid nitrogen. At this stage, crystallization was visible and the system was further cooled until the co-solvent system was solidified. Direct pouring of liquid nitrogen into the drug co-solvent solution was avoided. Later, these samples were lyophilized to powder using an Operon lyophilizer. The chiller trap was maintained below −100 °C.

Scanning electron microscopy (SEM)

SEM images of the samples were used for particle size and surface morphology determination. SEM imaging was carried out on an EVO 18 special edition model of a Carl Zeiss (Munich, Germany) microscope. Samples were carefully loaded onto double sided adhesive carbon tape pasted on aluminum SEM stubs. Gold/palladium sputtering for 16S at 10 mA current (using EMITECH, Emitech Ltd, Ashford, UK) was carried out for making the surface conductive.

Differential scanning calorimetry (DSC)

Accurately weighed samples of around 5 mg each were taken in aluminum DSC pans which were sealed using a crimp sealer. An empty sealed pan was taken as the reference pan. DSC was performed on a DSC 6000 (PerkinElmer, Inc., Waltham, MA) system coupled with an intracooler (Intracooler SP, PerkinElmer, Inc., Waltham, MA). The analysis was carried out under nitrogen purging (UHP grade) at a rate of 20 ml min⁻¹. In the first step, the samples were heated from 30 °C to 200 °C at a rate of 10 °C min⁻¹ and then cooled to −20 °C at 5 °C min⁻¹. The heat flow was recorded during the heating and cooling processes using software supplied by PerkinElmer (Pyris).

Thermal gravimetric analysis (TGA)

Samples of an accurately weighed quantity were heated from 20 °C to 800 °C at a 10 °C per minute heating rate. A SDT 2960 TGA instrument (simultaneous TGA-DTA, TA Instruments, New Castle, DE) was used for the analysis of the samples.

Powder X-ray diffraction (PXRD) studies

The samples were analyzed in the angular range of 7–50 (2θ) using an X-ray diffractometer (X’pert, Philips, Eindhoven, The Netherlands) with nickel-filtered Cu Kα radiation (k 50.154 nm), employing an X’celerator and a monochromator at the diffracted beam side at a voltage of 40 kV and electric current of 30 mA.

Accelerated stability study

For the moisture uptake study, an earlier reported method was followed.²² In brief, accurately weighed samples (approximately 100 mg each) were kept in a watch glass under controlled relative humidity in a glass desiccator maintained at 40 °C in an incubator. A relative humidity of 75% was achieved inside the desiccator with the use of saturated sodium chloride solution. Periodically, samples were taken out and weighed in an analytical balance (Sartorius) to calculate the moisture uptake. At the end of the study, the moisture uptake was crosschecked using the Karl Fischer titration method (841 Titrando; Metrohm AG, Switzerland). The crystallinity of the stability study samples was confirmed using PXRD as mentioned above. An already reported method for indicating the stability using HPLC analysis of INH was adopted. Briefly, an isocratic method with an ultraviolet detector at 254 nm and a Waters column (SunFire™ C18, 5 μm, 4.6 × 250 mm) was used for the separation.

In vitro lung deposition simulation studies

The official compendia (both the USP and EP) recommended method was followed for aerodynamic particle size determination of the prepared samples. An eight-stage cascade impactor (Thermo Fisher Scientific, Waltham, MA, USA) was used for the experiment. The fine particle fraction (FPF) and mass median aerodynamic diameter (MMAD) were calculated from the obtained data. For simulating the human throat, a throat piece was used along with a pre-separator. The pre-separator is recommended by the USP to prevent the large sugar carriers from entering into the cascade impactor. Accurately weighed samples (drug-carrier blend) were filled into size 2 capsules made of HPMC. The inspirational flow rate was maintained at 60 l min⁻¹ inside the cascade impactor. The particles were administered into the ACI with the use of Rotahaler® (Cipla, Mumbai, India). After each experiment the particles deposited on the filter disc were carefully washed and transferred to a 10 ml measuring cylinder. After dilution, the samples were analyzed using an already reported HPLC isocratic method, using an ultraviolet detector at 254 nm.²³ A disodium orthophosphate (Na₂HPO₄) buffer at pH 7.5 and acetonitrile in a 40 : 60 proportion were used as the mobile phase. A Waters column (SunFire™ C18, 5 μm, 4.6 × 250 mm) was employed as the stationary phase for the separation. The cumulative percentage of the dose collected from stages 2–5 represents the FPF of the formulation (aerodynamic diameter 1.1–5.8 μm).

Result and discussion

The INH API was dissolved in its most easily dissolving solvent, i.e. water. The anti-solvent (acetonitrile) was crucially selected because at room temperate the system should not induce crystallization and both of the solvents should be miscible. We tried with acetone and methanol in our initial trials. Acetone induces a sudden visible crystallization in room temperature, which is not under the control of the formulator. Similar to acetonitrile, methanol didn’t induce crystallization, but due to the very low freezing point of methanol (−95 °C) lyophilization was not possible. The higher freezing point of acetonitrile (−45 °C) proved it to be a better candidate for the easy removal of solvent in the lyophilization process.

Crystallization was visible through the transformation from a clear solution to a thick white colloidal suspension. In HC INH, the crystallization started just after it was plunged into liquid nitrogen and the sample was further cooled until complete solidification of the whole suspension occurred. In LC INH, crystallization and solidification of the system occurred almost simultaneously.

SEM analysis

The morphological characterization was done using scanning electron microscopy (Fig. 1). The particles of the raw material (INH API) did not have any particular morphology. In addition to that, they showed varying particle size too. Due to their low density and size, imaging of HC INH and LC INH was found to be extremely difficult at higher magnifications because of the vibration of the crystals when hit by the electron beam. In HC INH, much thinned uniform needle-like particles were observed. Each particle was discrete without solid bridges and had a smooth surface. Similar needle shaped crystals were also observed in the LC INH batch but have connections between the crystals. Many crystals of a rosette shaped morphology were also present. In contrast to HC INH, for LC INH the crystal surface was observed to be rough and had many asperities. The thickness of the HC INH crystals was less than that of the LC INH crystals and was in the range of 1–2 μm.

Fig. 1 SEM images of the INH API, LC INH and HC INH. A, C and E correspond to the INH API (500×), LC INH (500×) and HC INH (1000×). B, D and F correspond to higher magnification images of INH API (15 000×), LC INH (3000×) and HC INH (3000×), respectively.

Supersaturation leads to either homogenous or heterogeneous nucleation. The presence of a partial anti-solvent, i.e. acetonitrile, and the cryogenic conditions, induces the formation of crystal nuclei in the HC INH batch much faster when compared to LC INH due to the higher concentration. We speculate that the crystallization of the HC INH occurred during the cooling phase of the liquid when it was plunged into liquid nitrogen. Due to the high concentration, the abrupt generation of so many crystallization nuclei resulted in the formation of uniform needle shaped crystals (homogenous epitaxy). The cryo-cooling crystallization in LC INH may happen in multiple phases. The lower concentration causes the formation of a lower number of nuclei and the further addition of INH molecules during the crystal growth phase results in thicker crystals, as in the case of LC INH. In addition, secondary nucleation and further crystal growth on the previously formed crystals were evident in the SEM pictures of LC INH (Fig. 1C and D). This could be the reason for the presence of the rosette type crystals and solid bridges between the crystals in the LC INH batch. It could be inferred that crystallization progressively occurred until the solidification of the co-solvent system in the LC-INH batch.

DSC analysis

To assess the crystallinity, the lyophilized samples (of both HC INH and LC INH) and the raw material (INH API) were taken for differential scanning calorimetry (DSC). The heating curves (Fig. 2) show that all of the samples were in a crystalline condition and had the same melting point (180 °C). Just after the melting was completed, the samples were allowed to cool according to the DSC program (5 °C min⁻¹). During the cooling phase, data were also acquired to get an insight about the crystallization behavior of the samples. A single sharp exothermic peak was observed for all three samples. This indicated the chemical purity at the time of crystallization and that there was no degradation of the drug.

Fig. 2 DSC thermograms of the LC INH and HC INH in comparison to the INH API.

In DSC thermograms, a metastable form shows a lower melting point, but here we observed the same melting point in all three cases.²⁴ This indicated the possibility of there existing two stable polymorphs for INH, as reported in the case of the acyclovir derivative.²⁵ This phenomenon is very desirable because crystals formed through the proposed method will be as stable as the API crystals.

We know that for a given material in a spherical morphology, the temperature of melting and the temperature of crystallization should be the same.²⁶ In actual conditions, during DSC studies, a difference is noticed between the melting curve temperature and the crystallization temperature. This phenomenon is called a hysteresis. This may be the result of shape effects or the difference in the extent of cooling in the super cooling phase of the sample. In our case the samples showed different temperatures of crystallization. HC INH, which was observed to have the smallest crystals in the SEM analyses, had the highest temperature of crystallization (94 °C) here. The crystallization temperature of LC INH was observed to be 89 °C, whereas an even lower temperature was necessary to crystallize the large crystals of the API (81 °C). From these DSC data, we expect that the difference in crystallization may be the result of the particle size difference or the change in the aspect ratio, which can detrimentally affect the crystallization temperature. To check further the possibility of an existing polymorph, powder X-ray diffraction studies were carried out.

PXRD analysis

The PXRD spectra showed that both HC INH and LC INH had peaks at similar 2θ values which were different from the INH API (Fig. 3). The observed d-spacing values also ascertained that both HC INH and LC INH had similar crystal structures (Table 1). It is already reported that changes in the crystallization solvent or solvent mixture can produce a new polymorph. Since the PXRD data showed that both HC INH and LC INH had the same crystal nature, the difference in the temperature of crystallization observed in the DSC thermogram could be the result of differences in the particle size or porosity.

Fig. 3 PXRD data of INH API, HC INH and LC INH.

Table 1 List of d-spacing values from the PXRD study

INH API	INH high conc.	INH low conc.
7.38455	8.9844	9.0325
6.19951	8.0886	8.1085
5.72684	7.3861	7.3762
5.68063	6.6612	6.6894
5.33146	6.1717	6.2087
5.29699	6.0733	6.0835
4.51685	5.6724	5.6595
3.72008	5.2825	5.2901
3.68935	4.7588	4.7673
3.53977	4.5012	4.5112
3.5091	3.8491	3.8652
3.41426	3.6819	3.6923
3.2743	3.5374	3.5475
3.11657	3.5028	3.5069
3.09462	3.4029	3.412
3.03381	3.266	3.2769
2.93191	3.0889	3.0918
2.78805	3.0301
2.61791	2.9245
2.45816	2.7811	2.786
2.41586	2.4573	2.4572
2.40226	2.405	2.4081
2.33735	2.3243	2.326
2.32549	2.246	2.2416
2.25703	1.9714	1.9791
2.24408
2.06967
2.00112
1.99023
1.97583
1.94663
1.60987

---

TGA analysis

Derivative curves were plotted from the data obtained from the TGA analysis (Fig. 4). The major degradation peak of the INH API was at 240 °C, whereas HC INH and LC INH showed peaks at 227 and 220 °C, respectively. The thermal stability of a given material depends on the density of that material. This affirmed that the crystals present in HC INH were more tightly packed than those in LC INH. The smooth surface of HC INH (as observed in the SEM) may be the result of this peculiar crystal packing. This could be the reason for the enhanced stability of HC INH compared to LC INH. The rough surface and the crevices of LC INH may also be accelerating the thermal degradation process.

Fig. 4 TGA derivative curves of the INH API, HC INH and LC INH.

Accelerated stability studies

The samples were analysed for their moisture uptake tendency under accelerated conditions (40 °C/75% relative humidity). Periodically, the samples were taken out and weighed to determine the moisture uptake during the period of study. From the moisture uptake kinetics, it was clear that neither the raw material nor the polymorphs (LC INH and HC INH) had hygroscopic tendencies. Earlier observations from a moisture uptake study indicate that each quantum increment in weight is an indication of a conversion into a hydrated crystal.²⁷ In our study, the absence of this phenomenon denied the possibility of conversion into the hydrated crystal form. To rule out this possibility, PXRD analysis was also carried out on the stability study samples at the end of the second month (Fig. 5). The PXRD spectrum showed matching 2θ values of the starting materials which confirmed the stability of their crystallinity even under accelerated conditions. This affirms our earlier speculation that the crystal structure evolved through this new method is a stable polymorph of INH.

Fig. 5 PXRD spectrum of the stability study samples (HC INH and LC INH) after 2 months of study.

Impurity profiling of HC INH and LC INH was carried out using an already reported HPLC method for indicating stability at the end of the second month of study. INH in all its forms (INH API, HC INH and LC INH) was found to be stable. Apart from the impurities present from the raw material, no new peaks were observed in the chromatogram (Fig. 6). In addition to that, the area of the API relative to that of the impurity was conserved in all of the samples. These findings indicate that the API in the formulation didn’t undergo degradation during storage.

Fig. 6 HPLC chromatograms of the stability study samples along with a percentage purity table.

From the stability studies we can assume that neither physical nor chemical degradation occurred in the samples under study. At this moment, we can’t conclude that this has been achieved by the peculiarity of our crystallization method. This may be contributed to by the simpler and more stable structure of the API molecule itself.

In vitro lung deposition efficiency

To study the lung deposition efficiency of the lyophilized samples, an approved method described in the official compendia (United States Pharmacopeia and British Pharmacopeia) was adopted. An eight stage Anderson cascade impactor (ACI) was employed for this.

The fluid flow dynamics of a lung during inhalation has been well studied by various researchers. It is a fact that the cross sectional area of the airway from the trachea to the alveoli decreases progressively, causing enormous air resistance during inhalation. This imparts an efficient particulate filtering mechanism to the respiratory system. Reports say that only particles in the size range of 2–5 μm can reach the terminal part, i.e. the alveoli.^28,29 In the case of non spherical particles, particularly elongated ones, the aspect ratio plays a key role. If the smaller dimension (width or thickness) of the elongated particles is below 5 μm, they will reach the alveoli.^30,31 In our case, the SEM analysis of both batches (HC INH and LC INH) proved them to have a thickness within the inhalable size range. HC INH had a thickness of 2 μm and LC INH of 4 μm. From this we expected that both HC INH and LC INH could deliver the drug to the alveoli. So the particles were added directly, without admixing with an excipient, to the ACI under normal breath flow rate (28.3 l min⁻¹). Upon analysis at each stage of the ACI, we found that approximately 80% of the administered dose was deposited on the first stage itself. We were able to see the dose as aggregated chunks on the first stage of the ACI.

We opted for the conventional method of admixing with inhalation grade lactose to overcome the hurdle of aggregation. To avoid pressure during geometrical mixing with a spatula we used a screw capped plastic 50 ml centrifuge tube (Tarson tube). The samples (HC INH and LC INH) were mixed with lactose only using a gentle reciprocating motion for two minutes. This could be reproduced with a double cone blender in future scale up studies. Afterwards, the samples were analysed using the eight stage ACI with the pre-separator attachment. According to USP guidelines the pre-separator was employed for separating large lactose crystals and to avoid the overloading of the stages of the ACI during operation. As expected, both batches were able to deliver the medicament to the alveoli. The fraction of the dose administered under the inhalable size of 5 μm is termed as the fine particle fraction. In the eight stage ACI, particles above 5 μm will be filtered off before reaching stage 3. So, the percentage fraction of the amount of particles deposited below stage 3 (stages 4, 5, 6 and 7 and the filter) is taken as the fine particle fraction (FPF). During the operation of the ACI the particle deposition is directly proportional to the size and inversely proportional to the density. This density and particle size coupled parameter is expressed as the MMAD. The smaller the MMAD, the better the system will be for DPI application. The ACI characterization showed that the MMAD and FPF of HC INH were better than those of LC INH as a DPI candidate (Fig. 7). In the case of the INH API, most of the drug was found to be deposited on stage 0 and stage 1. Only a small percentage of the INH API progressed further to the lower stages. In our earlier SEM analysis, we also observed the presence of fine particles adsorbed on the surface of the large crystals of the API. Due to these, a large MMAD for the INH API was observed.

Fig. 7 Eight stage Anderson cascade impactor data for HC INH and LC INH.

It is known that a lactose carrier increases the aerolization efficiency of DPI by decreasing the electrostatic attraction between drug particles. To explore further the role of lactose in our system, the blends were analyzed using SEM (Fig. 8). Amazingly, along with the de-aggregation of drug particles, the lactose mixing step was found to induce particle size reduction in both HC INH and LC INH. The needle shaped particles were broken and adsorbed on the surface of the lactose crystal. Even the minimal impact forces acting on the HC INH and LC INH crystals during the mixing process are able to break the needle shaped crystals into small pieces. It is clear from the SEM images that more uniformly sized broken crystals are seen in the HC INH batch than in the LC INH. We could infer that crystals with a lower thickness are more prone to breakage during the mixing step. Though there are chunks of aggregated needle shaped crystals on the asperities of the lactose crystals, no agglomeration was seen between the broken pieces of the drug. This may be because the larger work function of the lactose crystal is sufficient to overcome the cohesive forces between the broken pieces of HC INH and LC INH.³²

Fig. 8 HC INH and LC INH after mixing with inhalation grade lactose.

The LC INH and HC INH batches of crystals were found to be suitable for DPI. The stability of these crystals under accelerated conditions is also promising regarding the shelf life period. In the widely accepted jet milling strategy, the API crystals are first brought into the inhalable size range by applying an enormous attritional force. This causes triboelectric charging along with a partial or complete conversion of the API crystal into the amorphous state. Even after the lactose mixing step, the work function of lactose won’t be able to separate the tightly cohered API micronized particles due to the excessive static (triboelectric) charges. So, in most of the commercially available formulations, the micronized drug on the surface of the lactose crystals exists as multiplets of drug–drug or drug–lactose fine particles.^33–35 Formulations using the HC INH and LC INH crystals would be superior to the products of micronization or spray drying in terms of the conservation of crystallinity.

Conclusion

To date no polymorph of Isoniazid has been reported. A new polymorph has been synthesized using an amalgamation of extreme low temperatures and the polarity of the solvent to encourage crystallization. Since crystallization under cryo-conditions is an ultra fast process, uniform crystals can be achieved through this method. We found this to be a highly facile method for the conversion of the API into inhalable particles. The chances of triboelectric charging and metal contamination are eliminated by avoiding milling. Unlike in various milling processes, there is no generation of heat during the process so chances of the conversion of the material into a less stable amorphous state are totally eliminated. We believe this platform technology can be used for the formulation development of inhalable drugs, especially, thermolabile biological molecules.

Acknowledgements

MVV is thankful to the Indian council for Medical research for providing senior research fellowship. We are thankful to Dept. of Biotechnology, India for providing facilities and funding. For SEM and TGA analysis we are grateful to CSIR-NIIST and SCTIMST of Thiruvananthapuram.

References

1. P. M. Young, R. Price, M. J. Tobyn, M. Buttrum and F. Dey, Drug Dev. Ind. Pharm., 2003, 29, 959–966.
2. A. Bailey, Powder Technol., 1984, 37, 71–85.
3. R. J. Malcolmson and J. K. Embleton, Pharm. Sci. Technol. Today, 1998, 1, 394–398.
4. H. Steckel, N. Rasenack, P. Villax and B. W. Müller, Int. J. Pharm., 2003, 258, 65–75.
5. P. O’Hara and A. J. Hickey, Pharm. Res., 2000, 17, 955–961.
6. P. Muttil, J. Kaur, K. Kumar, A. B. Yadav, R. Sharma and A. Misra, Eur. J. Pharm. Sci., 2007, 32, 140–150.
7. M. V. Vadakkan, K. Annapoorna, K. Sivakumar, S. Mundayoor and G. V. Kumar, Int. J. Nanomed., 2013, 8, 2871.
8. A. Misra, A. J. Hickey, C. Rossi, G. Borchard, H. Terada, K. Makino, P. B. Fourie and P. Colombo, Tuberculosis, 2011, 91, 71–81.
9. S. Claus, C. Weiler, J. Schiewe and W. Friess, How can we bring high drug doses to the lung?, Eur. J. Pharm. Biopharm., 2014, 86(1), 1–6.
10. W. Kaialy, H. Larhrib, M. Ticehurst and A. Nokhodchi, Cryst. Growth Des., 2012, 12, 3006–3017.
11. D. Murnane, C. Marriott and G. P. Martin, Cryst. Growth Des., 2008, 8, 2753–2764.
12. J. Schöll, C. Lindenberg, L. Vicum, M. Mazzotti and J. Brozio, Cryst. Growth Des., 2007, 7, 1653–1661.
13. S. Xie, S. K. Poornachary, P. S. Chow and R. B. Tan, Cryst. Growth Des., 2010, 10, 3363–3371.
14. R. S. Dhumal, S. V. Biradar, A. R. Paradkar and P. York, Int. J. Pharm., 2009, 368, 129–137.
15. S. Khan, M. D. Matas, J. Zhang and J. Anwar, Cryst. Growth Des., 2013, 13, 2766–2777.
16. S. Sugimoto, T. Niwa, Y. Nakanishi and K. Danjo, Int. J. Pharm., 2012, 426, 162–169.
17. T. Niwa, Y. Nakanishi and K. Danjo, Eur. J. Pharm. Sci., 2010, 41, 78–85.
18. A. Gupta, G. Pant, K. Mitra, J. Madan, M. K. Chourasia and A. Misra, Mol. Pharm., 2014, 11, 1201–1207.
19. R. K. Verma, J. K. Mukker, R. S. P. Singh, K. Kumar, P. R. P. Verma and A. Misra, Mol. Pharm., 2012, 9, 1011–1016.
20. A. S. Dharmadhikari, M. Kabadi, B. Gerety, A. J. Hickey, P. B. Fourie and E. Nardell, Antimicrob. Agents Chemother., 2013, 57, 2613–2619.
21. S. Katiyar, S. Bihari, S. Prakash, M. Mamtani and H. Kulkarni, International Journal of Tuberculosis and Lung Disease, 2008, 12, 139–145.
22. A. K. Salameh and L. S. Taylor, J. Pharm. Sci., 2006, 95, 446–461.
23. B. Glass, S. Agatonovic-Kustrin, Y.-J. Chen and M. Wisch, J. Chromatogr. Sci., 2007, 45, 38–44.
24. E. H. Lee, Asian J. Pharm. Sci., 2014, 9, 163–175.
25. A. Kristl, S. Srčič, F. Vrečer and D. Vojnovic, Int. J. Pharm., 1996, 139, 231–235.
26. K. Ishikiriyama, M. Todoki and K. Motomura, J. Colloid Interface Sci., 1995, 171, 92–102.
27. G. Buckton and P. Darcy, Int. J. Pharm., 1995, 123, 265–271.
28. L. M. Nagao, S. Lyapustina, M. K. Munos and M. D. Capizzi, Cryst. Growth Des., 2005, 5, 2261–2267.
29. I. Schultz, F. Vollmers, T. Lühmann, J.-C. Rybak, R. Wittmann, K. Stank, H. Steckel, B. Kardziev, M. Schmidt and P. Högger, Pulmonary insulin-like growth factor I delivery from trehalose and silk-fibroin microparticles, ACS Biomater. Sci. Eng., 2015, 1(2), 119–129.
30. K. Ikegami, Y. Kawashima, H. Takeuchi, H. Yamamoto, N. Isshiki, D.-I. Momose and K. Ouchi, Pharm. Res., 2002, 19, 1439–1445.
31. H. Larhrib, G. P. Martin, C. Marriott and D. Prime, Int. J. Pharm., 2003, 257, 283–296.
32. C. Gallo and W. Lama, J. Electrost., 1976, 2, 145–150.
33. P. Lucas, K. Anderson and J. N. Staniforth, Pharm. Res., 1998, 15, 562–569.
34. S. Sulihtyowati Soebagyo and P. J. Stewart, Int. J. Pharm., 1985, 25, 225–236.
35. S. S. Desai, A. A. Aher and P. T. Kadaskar, Drug Dev. Ind. Pharm., 2013, 39, 1589–1598.

---