Mefenamic acid: new polymorph or crystal defect?

Mefenamic acid (MA) [N–(2, 3–xylyl) anthranilic acid] is a high-dose, non-steroidal anti-inflammatory agent available in immediate-release capsule and tablet formulation. It is poorly soluble in aqueous media, sticking to any surface, creating problems in granulation and tabletting processes.¹ Mefenamic acid shows bioavailability ranging from 28–86% in capsule form.

The difference in bioavailability was found to be a result of the variation in dissolution rate involving multiple factors, such as the particle size and packing of the drug, and the de-aggregation rate of MA.² Our studies revealed that MA from different manufacturers has different crystal properties. The complete solid-state characterization of MA has also been performed.^3–4 We found that crystallization of MA from ethyl acetate (EA) at a fast cooling rate generated a new crystal form (or a crystal defect). To explore the feasibility of preparing this new polymorph, this study was conducted by altering the polarity of EA with different organic solvents under various crystallization conditions.

Materials

• A commercial sample of MA

(

www.bluecrosslabs.com

)

• Ethanol (E), EA and tetrahydrofuran (THF) were of absolute, analytical and HPLC grade, respectively.

• Potassium bromide (KBr) used for recording Fourier Transform infrared (FTIR) spectra was FTIR grade.

Methods

Microscopic examination

Light microscopy (LM). Recrystallized samples of MA were viewed/photographed under light and polarized light microscopes at various magnifications (×10, ×25 and ×45).

Hot stage microscopy (HSM). MA samples recrystallized from various solvents at varying degrees of supersaturation, cooling rate and agitation speed were screened for the required polymorph under a hot stage microscope (Leica DMLP; www.leica.com) at a constant heating rate from 32–250 °C to observe any transition.

Scanning electron microscopy (SEM). The surface morphology of MA-recrystallized samples was studied using an SEM (D–6000; www.jeol.com). Samples were fixed on a cylindrical aluminium stub with conductive double-sided adhesive tape and sputter-coated with gold.

Thermal characterization

Differential scanning calorimetry (DSC). Thermal analysis of MA samples was recorded in a differential scanning calorimeter (DSC 821; us.mt.com/home) using Mettler STARe software with the Solaris operating system (version 5.1).

Samples of 1–5 mg were weighed directly into aluminium crucibles sealed with pin-holed lids and scanned at 25–300 °C with a heating rate of 10 °C/min under nitrogen purging (80 mL/min). The temperature axis, heat flow and cell constant were previously calibrated with a pure sample of indium. An empty aluminium crucible sealed with a pin-holed lid was taken as the reference cell.

Thermogravimetric analysis (TGA). This was performed on the weighed quantities of samples by heating from 25–300 °C at the constant rate using a thermogravimetric analyser (us.mt.com/home). Any weight loss occurring before the melting point of the drug can be a result of solvent residues occluded in drug crystals during recrystallization.

Spectroscopic characterization

FTIR. A pellet was prepared by mixing 2–5% of the sample with KBr using a mortar and pestle, and compressed at a pressure of 15000 psi. Spectra were recorded in the transmission mode in the region of 4000–4500 cm^-1 by an FTIR spectrophotometer (www.perkinelmer.com).

Powder X-ray diffractometry (pXRD). To minimize preferred orientation, all samples were prepared by passing through a sieve (number 60) for pXRD analysis. The pXRD pattern of recrystallized samples was recorded at room temperature using an X-ray diffractometer (www.philips.com) with Cu kα1 radiation operated at 350 kv. Diffractograms were recorded over an angular range from 5–50° 2θ in a continuous scan mode using step size of 0.04° 2θ and a step time of 0.5 s.

Dissolution and solubility studies

Intrinsic dissolution rate (IDR) studies. IDR studies of recrystallized samples were performed using rotating disc apparatus with optimized parameters such as sample weight, compression force and stirring speed. Sample powder weighing 100 mg was compressed using a hydraulic press at a force of 3000 psi for 1 min to yield pellets of surface area 50.27 mm² that would not disintegrate during the test.

Study conditions include a 1000 mL jar, containing 900 mL of borate buffer maintained at 37 °C and 100 rpm. A sample volume of 5 mL was withdrawn at specific time intervals and analysed for MA.

Equilibrium solubility studies. Equilibrium solubility of MA acid was determined in two solvent mixtures EA:E (10:90% v/v) and EA:THF (50:50% v/v) at 32 °C and 60 °C. The drug was added to 5 mL solvent mixtures (kept in vials) until no more drug was dissolved at the specified temperatures, and closed with aluminium foil and rubber caps to prevent vaporization. All solutions were kept under constant stirring at 100 rpm at constant temperature in a water bath (www.hetoholten.com). At set time intervals, a fixed amount of solution was withdrawn with a microlitre syringe (www.hamilton.ch) and analysed by UV-vis spectrophotometry at 280 nm (DU640i; www.beckmancoulter.com) after appropriate dilutions.

MA recrystallization

Recrystallization of MA was performed from various solvents with differing degrees of supersaturation, cooling rates and with or without stirring. For recrystallization, the crystallizer was designed by using a 500 mL beaker jacketed with a 1000 mL beaker, whose inlet and outlet were connected to a circulator water bath (www.julabo.com) capable of controlling the temperature to 0.1 °C/min. Both the inner beaker containing sealed ampoules and outer beakers were closed separately.

Recrystallization was performed in sealed ampoules containing magnetic beads and 15 mL drug solutions of varying concentrations in different solvents. The entire crystallizer containing ampoules was placed on a magnetic stirring plate and the temperature increased to 70 °C, at which point all the drug in ampoules was solubilized. Cooling was then performed at slow and fast rates with the aid of a circulator bath, with and without stirring. The crystals were filtered and dried in a dessicator for 12 h.

To study the effect of harvesting time on polymorph generation, crystals from drug solutions of EA were filtered at two different times, and a visible mass of crystals was seen in solution after fast cooling. The crystals were kept in a dessicator for 12 h and analysed.

Results and discussion

Effect of solvents on recrystallization.

Solvents showing less solubility and high solute–solvent interactions are preferred for metastable polymorph generation.⁵ Furthermore, an appropriate balance between solubility, and hydrogen bond donating and accepting propensity of solvents is necessary.

MA solubility increases with decreasing polarity of solvent and has very high solubility in THF; therefore to achieve a balance between solubility and solute–solvent interactions, mixtures of solvents were prepared (THF:EA [50:50% v/v]; E:EA [90:10% v/v]).

The generation of a metastable polymorph was demonstrated in a previous study of EA at different recrystallization conditions (i.e., the degrees of supersaturation were 50.25 mg/mL, 53.00 mg/mL and 55.75 mg/mL, and the cooling rate was 20 °C/h with stirring). 100% EA was also selected for recrystallization.⁴

MA equilibrium solubility was determined at 32 °C and 60 °C in two solvent mixtures using a shaker water bath. The equilibrium solubility of MA increased with decreasing solvent polarity, as expected, at both temperatures. In addition, equilibrium solubility at 60 °C was determined to decide the degree of supersaturation for recrystallization experiments, while 32 °C was used to calculate the theoretical yield of recrystallization.⁶

Characterization of recrystallized samples.

Recrystallization of MA was performed in solvent–solvent mixtures (E:EA, EA and THF:EA) of three different polarities at three degrees of supersaturation and two cooling rates, with and without stirring. HSM and DSC screened all the crystals for new crystal properties and polymorphic forms. However, a significant difference in recrystallized product with respect to solid state was observed only in crystals obtained from E:EA mixture at various recrystallization conditions.

These crystals were further characterized by additional techniques to confirm the existence of the new polymorph.

Light and polarized microscopy. All crystals were observed under light and polarized microscopes to identify any change in morphology as well as optical properties. Since all the samples showed birefringence under a polarized microscope, the possibility of the amorphous form being produced at a fast cooling rate was ruled out.

There was a marked difference in the size of crystals obtained under different conditions where crystal size increased with decreasing cooling rate, unstirred conditions and increased degree of supersaturation. Some surface grooves were also seen for THF:EA and EA-recrystallized samples.

Hot scanning microscopy. Crystals from EA and THF:EA showed a wide transition range of about 10 °C between 150–215 °C, without recrystallization. Other than this, samples recrystallized from EA with the shortest harvesting time did not show any transition before melting. There was a sharp melting of crystals obtained from E:EA under stirred conditions at 222 °C while the melting point of MA is 231 °C (Figure 1).

Figure 1 HSM of crystals from E:EA under stirred conditions. All the crystals showed a sharp melting between 220–223 °C.

Furthermore, the melted product recrystallized at 223 °C and could not be seen clearly since melting and recrystallization were almost simultaneous.

Melting without any transition appears to be caused by too high a heating rate. The sharper melting and recrystallization events of crystals from E:EA can be attributed to melting at 222 °C, followed by recrystallization at 223 °C leading to melting at 231 °C. A similar trend of solid–liquid transition was also observed for carbamazepine, which melted at 175.92 °C, recrystallized at 181.66 °C and melted at 192.1 °C. ⁷

The wide transition of crystals from EA and THF:EA is representative of solid–solid transition. This was observed at 150 °C for solid–solid transition of monoclinic retinoic acid to its triclinic form followed by its complete melting at 183 °C.⁸

However, a different trend in transition and melting of recrystallized product from the E:EA mixture and other solvents indicated the possibility of a different crystal form.

Scanning electron microscopy (SEM). In light microscopy, some grooves were seen on the crystal surface of samples from THF:EA and EA compared with those from the E:EA and the United States Pharmacopeia (USP) reference sample, hence SEM photomicrographs of representative recrystallized and USP reference samples were recorded. The reference sample was observed to be an aggregated microcrystalline mass with columnar plate-like properties.

The recrystallized samples were found to have larger particle size with the same properties as that of reference sample. At higher magnification, THF:EA and EA recrystallized products had some depressions on the crystal surface unlike E:EA-recrystallized samples (Figure 2).

Figure 2 Scanning electron photomicrographs of crystals from different solvents. Key: MR USP reference; (A) photomicrographs at low magnification; (B) photomicrographs at high magnification. Crystals from THF:EA and EA showed some depressions on the surface.

Disc spinning calorimetry (DSC). There was no significant difference in the transition or melting of crystals — except those obtained from EA with the shortest harvesting time, which melted without any transition. Moreover, the transition temperature range was wider for samples recrystallized from THF:EA and EA than from E:EA.

All the crystals obtained from E:EA under slow and fast cooling with stirred conditions showed a sharp melting endotherm at 222 °C, a recrystallization exotherm at 223 °C and another melting endotherm at 231 °C.

The thermal transitions for samples from EA and THF:EA appear to be the result of an increased proportion of crystal defects, whereas sharper thermal transitions for E:EA-recrystallized product may be a result of increased crystallinity.^9–12 However, the highest heat of transition may be caused by too high a proportion of crystal defects interacting with each other, causing relaxation of the crystal lattice leading to the stable state and requiring high energy for melting.

Thermogravimetric analysis (TGA). When crystals from THF:EA, EA and E:EA underwent TGA no weight loss was found prior to the melting point. There was weight loss between 235–260 °C, which indicates that MA melts with degradation. No weight loss before melting signifies the absence of any pseudopolymorph formation from used solvent–solvent mixtures.

Although the solvent systems of lower polarity interacted strongly with the drug, no solvent had enough capacity to occlude into the crystal lattice, thereby defeating possibilities of solvate formation.¹³

Fourier transform infrared (FTIR)

spectroscopy. Since polymorphs differ from each other with respect to hydrogen bonding, spectral differences in MA within a characteristic region as a result of the functional group stretching (e.g., C=O and N–H), can be attributed to polymorphism. According to Burger's IR rule, less stable polymorphs will have a first absorption band at a higher frequency.¹⁴ It is well known that forms I and II of MA show characteristic differences in detailed shape and intensities at the N–H stretching region between 3350–3300 cm^{-1 4}. Further N–H stretching frequency at 3313 cm^-1 and 3347 cm^-1 is assigned for form I and form II, respectively. In addition, there are differences between form I and II below 1600 cm^-1 , which required recording the spectra of all recrystallized samples from 4000–4500 cm^{-1 9}.

Over the entire region, spectra exactly matched the USP reference sample (Figure 3). It can be inferred that recrystallization from all the solvents resulted in form I.

Figure 3 FTIR spectra of samples recrystallised from THF:EA, EA and E:EA compared with reference USP standard MR. In functional group as well as fingerprint region, spectra of all the crystals were similar to USP reference form.

However, it is now difficult to rule out the ability of FTIR to differentiate the small differences at molecular level for MA.

Powder X-ray diffraction (pXRD). Two solid forms are said to be different from each other if the scattering angles of the ten strongest reflections differ by ± 0.20° and the relative intensities of these reflections vary by more than 20%.¹⁵ In the present study, the recrystallized samples did not differ from each other by the above levels. Only small changes in the intensities of some peaks were observed which may be caused by preferred orientation.

Samples recrystallized from E:EA, EA and THF:EA differed from each other and the USP reference with respect to peak intensity (i.e., increasing intensity with increasing solvent polarity, according to DSC).

Intrinsic dissolution rate (IDR). The IDR of recrystallized samples was determined under similar conditions, except the compression was 3000 psi.⁴ There is no statistically significant difference at the 95% confidence interval (P > 0.623) for the samples harvested from E:EA. The observed minor difference can be attributed to the altered crystallinity (i.e., the crystal defect), which agrees with the DSC and pXRD results. Although recrystallization increases the crystallinity of samples recrystallized from E:EA, it may not be too high to affect the IDR statistically. Hence, there is no effect on the rate of solubility and many do not have any effect on bioavailability when compared with the reference if all other conditions remain constant.

Effect of various factors on recrystallization

Effect of solvent. Recrystallization of MA from all solvents yielded only polymorph I. However, the transition temperature is wider for samples recrystallized from THF:EA and EA compared with those from E:EA. From this, it is clear that there was a change in crystallinity with solvent polarity.

The crystals obtained from E:EA showed maximum crystallinity as is evident from pXRD and DSC patterns.

An increase in transition temperature range for the conversion of form I to II (broadened DSC endotherms), as well as diminished pXRD peaks with a decrease in the polarity of solvent, seems to be evident for crystal defects.¹²

Recrystallization can be visualized as a process of molecular aggregation, which can be subdivided into desolvation of MA, self-assembly of desolvated MA and aggregate solvent interaction.¹³ The fact that decreasing solvent polarity improves MA crystallinity explains that the crystalline state is formed in conditions favouring desolvation of MA as self-assembling takes place.

Therefore, it can be inferred that a nucleus originating from the solvent phase is composed of unsolvated MA molecules. Among the solvents under study, this happened with the E:EA system. It is clear that THF:EA and EA are relatively less polar than E:EA and MA has more non-polar groups.

Moreover, strong interaction between the two-solvent systems and MA occupy the non-polar portions. Hence, aggregates formed in the presence of these systems are formed by solvated MA molecules, resulting in a poorly ordered crystal structure, suggesting more defects in the crystal lattice.

Conversely, E:EA has a higher polarity index therefore less interaction with MA, which favoured more ordered structure, hence a higher heat of fusion and sharper melting point.

Effect of stirring. In all the solvents, it was always form I that was obtained under stirring and non-stirring conditions. Agitation has a profound effect on polymorph generation. Generally, the rate of polymorphic transformation is higher for stirring than that for non-stirring. This is observed in the case of taltirelein as well as sulphamerazine.⁵

The initially nucleated metastable polymorph dissolved rapidly into the surrounding phase and again nucleated as the stable form I of MA as a result of stirring. For stirred conditions Mier's supersaturation theory cannot be applied since nucleation occurred before the concentration reached the metastable zone.

It might have occurred just across the solubility line rather than the supersolubiltiy line, therefore, it seems that the number of nuclei formed were insufficient to further grow as metastable crystals and dissolved into the solution phase and gave form I only. Under unstirred conditions, the nucleation rate was less; therefore, some of the nuclei initially formed converted immediately into the stable form before a sufficient number of nuclei appeared to form metastable crystals.

Effect of supersaturation. At all three levels of supersaturation, each solvent only produced stable form I except the crystals obtained from E:EA solvent. There was a maximum crystallinity for an intermediate degree of supersaturation, as was evident from DSC and pXRD results. At the constant temperature, supersaturation ratio and interfacial energy are key factors for the relative nucleation rate of the polymorph.

According to Oswald's step rule, at low supersaturation the difference of the supersaturation ratio between the polymorphs is influential on the crystallization and the stable form may preferentially precipitate. Conversely, at high supersaturation, the difference of the interfacial energy between the polymorphs becomes relatively dominated and the metastable form may tend to precipitate. However, such behaviour was not observed.

Effect of cooling rate. Cooling is an important variable for polymorph generation by altering the degrees of supersaturation. At faster cooling rates there is more supersaturation which leads to a less stable polymorph. In the present case, both cooling rates seem to be insufficient, giving only the polymorph of higher stability.

Effect of harvesting time. It is evident from the data available that the crystals produced during harvesting were of form I. Generally, the shorter the contact time between the initially formed metastable nuclei and solution phase, the greater the retardation of polymorphic transformation to a more stable form, and there are more possibilities of obtaining the metastable polymorph.^16,17

In the present study, although there was a very short contact time between the crystals and solution, the stable polymorph is generated, possibly during filtration or drying — the processes that may not be too rapid to prevent transformation to stable form.

Conclusions

Characterization indicates possibilities of generating crystals with improved crystallinity from low polarity solvents. IDR studies revealed no significant difference in the dissolution rate for products derived from low and high polarity solvents. These crystals may have improved material characteristics, which will help to solve the sticking and tabletting problems of MA.

Ramesh Panchagnula is a professor at the School of Biomedical Sciences, University of Ulster (UK).

Tata Bapurao, Yasvant Ashokraj and P Pillai are all post graduate students at NIPER (India).

References

1. A. Adam, S. Leopold and C.S. Peter, Drug Dev. Ind. Pharm. 26, 477–487 (2000).

2. D. Shinkuma et al., Int. J. Pharm. 21, 187–200 (1984).

3. S. Agrawal, O. Pillai and R. Panchagnula, Solid State Characterization of Mefenamic Acid, 55th Indian Pharmaceutical Congress, New Delhi: AP, 141, 175–176 (2001).

4. R. Panchagnul et al., J. Pharm. Sci. 93, 1019–129 ( 2004).

5. H.G. Chong et al., J. Pharm .Sci . 90, 1878–1890 (2001).

6. W.L. Badger and J.T. Banchero, "Crystallization" in Introduction to Chemical Engineering (Tata Mcgraw-Hill, New York, NY, USA, 2000)

7. C. Rustichelli et al., J. Pharm. Biomed. Anal.23, 41–54 (2000).

8. V. Berbenni et al.,Int. J. Pharm. 221, 123–141 (2001).

9. S. Romero, B. Escalera and P. Bustamante, Int. J. Pharm. 178, 193–202 (1999).

10. S.D. Clas, C.R. Dalton and B.C. Hancock, Pharm. Sci. Tech. Today. 8, 311–319 (1999).

11. S.P. Duddu and D.J.W. Grant, Thermochemica Acta. 248, 131–145 (1995).

12. D.J.W. Grant, "Theory and origin of polyymorphism," in H.G. Brittain, Ed., Polymorphism in Pharmaceutical Solids (Marcel Dekker, New York, NY, USA, 1999) pp. 1–35.

13. M.L.P. Leitoo et al., Thermo. Acta. 30, 1–8 (2003).

14. Y. Lian, M.R. Susan and A.S. Gregory, Pharm. Sci. Tech. Today.1, 118–127 (1998).

15. H.G. Brittain, J. Pharm. Sci. 91, 1573–1580 (2002).

16. W. Beckmann, Org. Proc. Res. Dev. 4, 372–383 (2000)

17. G. Nicholas and C.S. Frampton, P. J. Pharm. Sci., 87, 684–693 (1998).