AnalyticalMethods

PAPER

Department of Pharmaceutical Sciences, U

Italy. E-mail: giuseppina.ioele@unical.it;

493268

Cite this: Anal. Methods, 2013, 5, 1715

Received 28th November 2012Accepted 11th February 2013

DOI: 10.1039/c3ay26467j

www.rsc.org/methods

This journal is ª The Royal Society of

Lonidamine and related impurities: HPLC analysis,stability profile and degradation pathways

Giuseppina Ioele,* Michele De Luca and Gaetano Ragno

A sensitive HPLC-DAD method for the simultaneous determination of the anticancer lonidamine and five

potential impurities in pure powder and tablet dosage forms was developed and validated. The impurities

comprised by-products from thermal and photodegradation and potential secondary products from the

drug synthesis. The structure of the main drug impurity was elucidated with the aid of GC-MS

chromatography. A good resolution of all the analytes was achieved on a Gemini C18 column using

isocratic elution and DAD detection. The optimal mobile phase consisted of a mixture of methanol–

phosphate buffer (pH 3) (80 : 20, v/v). At a flow rate of 1.5 ml min�1, the total time of analysis was less

than 15 min. The method was determined to be linear in the range of 10–200 mg ml�1 for the

quantitation of lonidamine and 0.5–50 mg ml�1 for the related substances with a detection limit of

0.15 mg ml�1 for the impurities. Percentage recoveries ranged from 97.4 to 102.3 and 95.7 to 103.5 for

the drug and impurities, respectively, and the relative precision values were less than 3.3%. Finally, the

method was demonstrated to be robust, resistant to small variations of chromatographic variables such

as pH, mobile phase organic/aqueous composition and column temperature.

Introduction

Lonidamine (LND) or 1-(2,4-dichlorobenzyl)-1H-indazole-3-carboxylic acid is an anticancer molecule currently used to treata number of tumors, including head, neck, and breastcancers.1–3 It is believed that this drug would not induce mye-losuppression and other typical side effects of the classic cyto-toxic drugs because it would interfere with cell energyprocesses, by inhibiting aerobic glycolysis in cancer cells.4 Inrecent years, however, studies of this molecule have beenfocused on its antispermatogenic properties in several types ofanimals.5 But the molecule causes signs of toxicity such asmyalgia, asthenia, testicular pain, and gastrointestinaldiscomfort.6,7 Thus the goal of recent studies is to make analogsof the compound that block the production of sperm, but haveno toxicity problems. The recent studies about the pharmaco-logical activities of LND have been reported in a recent review.8

The synthesis of the drug is depicted in Fig. 1 and is based onthe alkylation of indazole-3-carboxylic acid (ICA) with 2,4dichloro-benzyl chloride (DBC).9 Some secondary products canbe obtained from this process. The most important is theisomer 2-(2,4-dichlorobenzyl)-2H-indazole-3-carboxylic acid(IS). The parent substances ICA and DBC could also be potentialimpurities because of their incomplete reaction.

niversity of Calabria, 87036 Rende, CS,

Fax: +39 0984 493201; Tel: +39 0984

Chemistry 2013

In recent times, there has been an increased tendencytowards the development of stability-indicating assays,10–12

using the approach of stress testing as enshrined in the Inter-national Conference on Harmonization (ICH) guidelineQ1AR(2) ICH, Q1A(R2).13 This guideline addresses the evalua-tion of stability data that should be submitted in registrationapplications for new drugs and associated drug products andprovides recommendations on establishing re-test periods andshelf lives.

LND has been determined in human plasma and urine14 andin serum,15 but to our knowledge, no report about a completepurity assay of LND has been published. Two papers dealingwith the simultaneous determination of the drug and IS havebeen reported16,17 but no stability-indicating method has beenreported. For this purpose, in the present work, the analyticalprole of LND was studied in depth including its behavior whenexposed to high temperatures or light. Since indazole-3-carboxylic acid undergoes easy decarboxylation by heating,18

indazole (IND) and the decarboxylated form from LND (LDC)were searched in the degraded samples of LND.

The photodegradation studies were conducted in accordancewith the recommendations of the ICH rules by performing thestressing tests bymeansof a cabinet equippedwith a xenon lamp,able to adopt different irradiation modes reported in the ICHguide.13 Identifying drug degradation products in LND formula-tions is of critical interest to ensure product quality with asignicant impact on safety and efficacy of the drug. Degradationproducts formed from LND subjected to forced-degradationwere investigated using a new simple and quick HPLC-DAD

Anal. Methods, 2013, 5, 1715–1720 | 1715

Fig. 1 Synthesis scheme of LND.

Analytical Methods Paper

method which allowed a very good resolution of all the analytesstudied. The method was validated with respect to specicity,linearity, precision, accuracy and robustness. The structure of themain degradation productwas investigated and conrmedby theGC/MS technique. The method was applied to LND pure powderand drug pharmaceutical formulations.

Materials and methodsa Chemicals and reagents

LND, ICA, DBC and IND were purchased from Sigma-Aldrich(Milan, Italy). IS was isolated as a secondary product from thealkylation of ICA with DBC to synthesize LND following theclassical reaction of Silvestrini.9 For this purpose, ICA wasreacted with 3 equivalents of aqueous NaOH and 1 equivalent ofDBC, stirring the solution at 50 �C for 7 h. The mixture solutionwas then fractionated on a silica column and the fractions wereanalyzed by HPLC. IS was isolated and its identity wasconrmed by comparing the UV spectrum from DAD with thatreported in the literature.14 The fraction was dried under highvacuum at 60 �C for at least 48 h and IS was obtained as a whitepowder. The purity of the product was determined to be greaterthan 94% by HPLC analysis of standard solutions.

LDC was obtained from thermal degradation of a LNDstandard solution 100 mg ml�1 in a sealed vial, maintained at70 �C for 60 days. The mixture solution was fractionated on asilica column and analyzed by HPLC. LDC was isolated and itsidentity was conrmed by GC/MS.

Phosphoric acid 85% p.a. was obtained from Sigma-Aldrich(Milan, Italy), methanol and water for HPLC from J.T. Baker(Holland). PTFE 0.45 mm membrane lters were purchasedfrom Supelco (Milan, Italy). All other reagents were of thehighest purity commercially available.

The specialty Doridamina� tablets (30 mg) were commer-cially obtained.

b Analytical instrumentation

HPLC analysis was carried out by using a HP 1100 pump ttedwith a diode array detector G1315B (Agilent Technologies) and aRheodyne 7725 manual injector. The LC column was a C18Gemini (Phenomenex), 250 � 4.6 mm � 5 mm. The mobilephase consisted of methanol–phosphate buffer (pH 3) (80 : 20,v/v) at room temperature. The mobile phase was lteredthrough a 0.45 mm nylon membrane under vacuum. The mobilephase was pumped isocratically at a ow rate of 1.5 ml min�1

1716 | Anal. Methods, 2013, 5, 1715–1720

during analysis. The injection volume was 5 ml. The optimumwavelength for the detection of all analytes with adequatesensitivity and specicity was found to be 275 nm.

Gas-chromatographic (GC-MS) analysis was performed by agas chromatograph Agilent 6890N with a Mass SelectiveDetector Agilent 5973. The GC conditions were as follows:column, HP capillary (30 m � 0.25 mm i.d.); 250 nm lmthickness; injection port temperature, 250 �C; carrier gas,helium; ow-rate 0.3–0.6 ml min�1; column temperature, pro-grammed from 60 to 280 �C at 13 �C min�1; initial time 3 min,nal time 54 min.

Forced photodegradation was conducted in a light cabinetSuntest CPS+ (Heraeus, Milan, Italy), equipped with a xenonlamp. The apparatus was tted with an electronic device forirradiation and temperature measuring and controlling insidethe box. The system was able to select spectral regions byinterposition of lters. In the present study, samples wereirradiated under a direct lamp, with interposition of a singlequartz lter, furnishing irradiation over 270 nm, and thenbetween 320 and 800 nm, by means of a glass lter, according tothe ID65 standard of ICH rules.

c Standard samples

Stock solutions of all the single compounds were prepared bydissolving appropriate amounts in methanol. The referencesolutions were prepared by diluting with methanol the respec-tive stock solutions to the nal concentrations of LND(10.0–200.0 mg ml�1), ICA, DBC, IS, LDC, IND (0.5–50.0 mg ml�1)in volumetric asks. A series of mixture solutions were preparedwith all the analytes within the same concentration ranges ofthe single-component standards. The ratio between the impu-rities and LND was maintained in a percentage range from 1 to83%. In all degraded samples the percentage of any impuritynever exceeded the value of 83% with respect to starting LND.

All solutions were stored in the refrigerator (4 �C) beforeanalysis.

d Sample preparation

The content of ve tablets of the specialty Doridamine� waspulverized and the equivalent to one tablet was suspended inmethanol in a 25.0 ml volumetric ask. The suspension wassonicated for 10 min and then ltered through a 0.45 mm PTFEmembrane. Samples for HPLC analysis were obtained bydiluting 1 ml of the ltrate to 10.0 ml with methanol.

This journal is ª The Royal Society of Chemistry 2013

Paper Analytical Methods

Results and discussion

Several analytical columns were tested to achieve an optimalresolution of LND and the major related substances in itssynthesis, shown above in Fig. 1. The Phenomenex Gemini C18reversed-phase bonded phase provided the best peak shape andsample resolution. This phase is designed for extended pHstability (pH range 1–12). Different mobile phases consisting ofmethanol and phosphate buffer were tested and the pH of thebuffer was varied from 2 to 6 with 85% phosphoric acid. Thebest results were obtained using a mobile phase consisting ofmethanol and phosphate buffer in the ratio 80 : 20 (v/v), pH 3.0in such a way as to achieve a stable baseline and better peakshapes. The mobile phase was isocratically eluted with a owrate of 1.5 ml min�1. The analysis time was maintained lessthan 11 min in which all the components were well resolved.Under the described conditions, it is possible to observe a goodseparation between all the peaks, with LND eluting at 6.05 min.The reagent products ICA and DBC eluted at 1.58 and 9.83 min,respectively, whereas the isomer IS eluted at 7.03 min. Thesedata conrm that the method developed can be suitable as arapid and reliable quality control method of LND as a bulkmaterial.

a Method optimization and validation

Chromatographic separation was optimized with respect to thestationary and mobile phase compositions, ow rate, samplevolume and detection wavelengths. The method was oppor-tunely validated to demonstrate suitability for routine applica-tion. The resolution of individual components and repeatabilityof injections were evaluated by retention time and peak areavalues.

Calibration curves of LND and the other main products werecalculated from data of 10 reference solutions of eachcompound, with concentrations ranging between 10 and 200 mgml�1 for LND and 0.5 and 50 mg ml�1 for the impurities,distributed on 5 levels and analyzed in triplicate. The relativepeak areas (dependent variables) were correlated to the

Table 1 Accuracy (%recovery) and precision results (standard deviation) from applmg ml�1

Sample

Nominal

LND ICA DBC IS IND LDC

1 10.02 0.52 1.83 4.50 14.85 50.222 25.05 1.56 5.49 13.50 44.55 0.623 50.1 4.68 16.47 40.50 0.55 1.864 100.2 14.04 49.41 0.50 1.65 5.585 200.4 42.12 0.61 1.50 4.95 16.746 10.02 4.68 49.41 1.50 14.85 0.627 25.05 14.04 0.61 4.50 44.55 1.868 50.1 42.12 1.83 13.50 0.55 5.589 100.2 0.52 5.49 40.50 1.65 16.7410 200.4 1.56 16.47 0.50 4.95 50.22MeanSD

This journal is ª The Royal Society of Chemistry 2013

respective drug concentrations (independent variables). Thecorrelation coefficients were all higher than 0.98.

The homoscedasticity of replicate measurements over theconcentration ranges explored was tested to verify that thevariances of measurements are independent of the concentra-tion levels. The homoscedasticity was calculated by Cochran'stest19 in which the parameter Cochran C (relative to n standardswith p replicates) should be less than a critical value for eachdrug. The critical value for C5,3 at 95% condence interval is0.684. The six calibrations passed the homoscedasticity testsince the C values were less than the critical value. The regres-sion equations were validated by assaying an external set of 10synthetic standard mixtures, with component concentration inthe same range of the calibration set. The method showedpercent recovery between 97.4–102.3 and 95.7–103.5 for thedrug and for the impurities, respectively, and values of %RSDnever above 3.

The results of accuracy (%recovery) and precision (standarddeviation) from application of HPLC analysis on this predictionset are listed in Table 1. The study of interday and intradayprecision at different concentration levels was performed byanalysing several standards of known concentration. Themethod provided good precision for interday and intradaymeasures independent of the concentration, with a relativestandard deviation (%RSD) between 1.9–4.2% and 2.4–5.8,respectively. Thus, the repeatability and reproducibility of themethod were satisfactory.

The robustness of the HPLC method was determined byobserving the inuence of limited changes in the mobile-phasecomposition on the peak areas and retention times. Where thechanges were maintained under a value of �6%, relative peakareas related to the peak areas for optimal mobile phasecomposition resulted in a range from 95% to 105%. When thechange exceeded the value of �6%, some peaks were notcompletely separated.

The next study was a stability study, carried out by exposingLND in both solid and solution forms under stressing condi-tions. Degradation of the drug was examined under heat andlight exposure.

ication of HPLC analysis on the prediction set. The nominal values are expressed as

Found

LND ICA DBC IS IND LDC

98.20 94.91 102.28 94.91 94.76 95.4887.70 103.24 101.74 96.27 99.07 100.2199.40 95.67 100.42 98.31 98.24 98.41

102.10 94.07 98.07 100.22 103.33 99.27100.98 95.29 99.50 101.48 98.37 102.7499.02 94.75 96.29 103.2 99.21 101.78

102.30 102.53 94.80 102.21 101.04 103.1197.40 101.21 101.90 100.74 97.70 98.24

101.50 100.98 95.03 97.28 100.21 97.72100.08 99.27 96.24 99.44 94.88 95.2998.87 98.17 98.63 99.41 98.64 99.231.74 2.23 2.01 2.98 2.56 2.43

Anal. Methods, 2013, 5, 1715–1720 | 1717

Analytical Methods Paper

b Thermal degradation

Two series of six LND methanol solutions 100 mg ml�1 in sealedvials were maintained in a thermostatic oven at 40 and 60 �C,respectively. HPLC analysis was run just aer sample prepara-tion (t ¼ 0) and at the following times: 2, 5, 10, 20, 30 days.Thermal degradation tests were also performed on the solidproduct, kept in sealed vials and maintained in a thermostaticoven, at 60 and 100 �C. HPLC analysis was run at t¼ 0 and at thefollowing times: 10, 20, 60, 90 days. Analogous samples of LND,in solution and in solid state, were maintained at 21 �C andanalysed at the same times.

The sample maintained at 40 �C showed no signicantchanges along the entire test, demonstrating very high stability.In contrast, the sample kept at 70 �C showed the formation of anew peak at 13.85 min and the simultaneous decrease of theLND peak. This degradation product was identied by GC/MSanalysis as the decarboxylated form of LND (LDC), with amolecular signal with m/z 277 and the characteristic signalsm/z118, corresponding to the indazole derivative, and m/z 162corresponding to the indazole carboxylic derivative. A series ofsecondary products were detected in small amounts withretention times under 6 min but their identication was notcarried out. Aer 30 days the residue percentage of LND resul-ted to be 37%. A HPLC chromatogram of a thermally stressedLND sample under 70 �C is shown in Fig. 2.

For the control, standard solutions of LND in methanol atthe same concentrations were kept at room temperature (25 �C)and in a refrigerator (4 �C). No signicant changes in thecontent of LND were observed during the storage at bothconditions even aer 90 days.

The thermal stressing tests were also performed on the solidproduct, kept in closed vials at 60 and 100 �C and analyzed every10 days up to a nal time of 90 days. Only traces of LDC occurredin these experiments, conrming the high stability of solid LNDunder stressing temperature.

Fig. 2 HPLC chromatogram of a 100 mg ml�1 LND solution kept at 70 �C for30 days.

1718 | Anal. Methods, 2013, 5, 1715–1720

c Photodegradation

In the present study, samples were submitted to forced photo-degradation in the light cabinet Suntest CPS+ under twodifferent conditions. In a rst test series, six samples of LND100 mg ml�1, were subjected to direct irradiation of the xenonlamp with interposition of only a quartz lter, corresponding toa wavelength range between 270 and 800 nm. In a second testseries, six LND solutions with the same concentration wereirradiated under a wavelength range between 320 and 800 nm,by means of glass lters, according to the ID65 standard of ICHrules. In both the experiments, the irradiation power was xedat 350 W m�2, corresponding to an energy value of 21 kJ (minm2)�1. The inner temperature was maintained constant at 25 �Cin all these experiments. HPLC analysis was run just aersample preparation (t ¼ 0) and at the following times: 5, 15, 30,60, 90, 120, 150, 210, 270, 330, 390, 450, 510, 600 min. Photo-degradation tests were also performed on the solid product,kept in sealed vials, under the above described conditions.HPLC analysis was run at t ¼ 0 and at the following times: 5, 10,30, 60, 90 h. Samples of ICA 50.4 mg ml�1 and IND 50.5 mg ml�1

were also submitted to forced irradiation for studying thephotodegradation mechanism of LND.

When the rst experiment with irradiation over 270 nm wasperformed, a rst photoproduct appeared aer 30 min with aretention time of 1.58 min. This product was identied to beICA, derived from the debenzylation of LND. Its concentrationincreased constantly during degradation up to a value of 12.4 mgml�1 aer 600 min, calculated by means of the regressionequation. When the logarithmic values of the concentrationpercentage (LND%) were plotted against the degradation time(min) up to 120 min, a straight line was obtained, according tothe following equation:

log LND% ¼ 0.001 t(min) + 1.9997

where 0.001 is the photodegradation rate constant, t the time(min) and 1.9997 is the experimental logarithmic value of thestarting concentration percentage (100%).

Aer about 120 min, the forced irradiation induced a morecomplex degradation, with the appearance of a peak at 13.85min, identied as LDC, the decarboxylated by-product of LND.Its content reached the value of 13.5 mg ml�1 aer 600 min,calculated by the relative equation. At the same time, thesimultaneous formation of other two main photoproductspresenting signals at 5.26 (PP1) and 9.01 (PP2) min wasrevealed, both increasing all through the experiment. Simulta-neously, a number of other secondary products were formed buttheir contents remained low or at trace levels. Unfortunately,isolation of these products was very hard and their character-ization was not carried out. The gradual disappearance of LNDwas recorded up to a residue value of about 12 mg ml�1. In Fig. 3the HPLC chromatogram of a LND sample aer 210 min offorced irradiation is shown.

In order to identify the two photoproducts PP1 and PP2,degradation experiments were also performed on the productsICA and IND, assuming that they can be secondary photo-products aer decarboxylation and debenzylation of LND. IND

This journal is ª The Royal Society of Chemistry 2013

Fig. 3 HPLC chromatogram of a 100 mg ml�1 LND solution kept under stressingirradiation for 210 min.

Paper Analytical Methods

samples showed a remarkable stability and aer 240 min thepercentage of residual product was found to be 97% with theformation of some degradation products, but all in low or traceamounts, and none of them coincided with the productsinvestigated. In contrast, degradation of ICA showed theformation of several degradants including two of them withretention times close to those of PP1 and PP2 and similarspectroscopic characteristics. In fact, the DAD signals of boththese degradants presented UV spectra with a clear decrease ofthe absorbance in the zone 275–325 nm. This absorbancesignal is present in the spectra of all the main products in thepresent study, including LND, ICA, IND, IS and LDC. Theunknown degradants from LND irradiation were so hypothe-sized to be secondary products from degradation of ICA, inturn derived from debenzylation of LND.

Fig. 4 Proposed degradation of LND under stressing thermal and irradiation cond

This journal is ª The Royal Society of Chemistry 2013

The second degradation experiments, in which the LNDsamples were treated under irradiation between 320 and800 nm, showed a lower degradation compared to that previ-ously reported, with a LND residual concentration of 90% aer600 min of irradiation. The mean photoproduct formed in theseexperiments was ICA, reaching a concentration of about 5.1 mgml�1. Only traces of LDC and of the same photoproducts PP1and PP2 described in the previous experiment were revealed.

These results demonstrated that LND is sensitive to light,but its degradation is signicant only in the presence of high-energy light, with wavelengths shorter than 320 nm. In normalconditions of manufacturing or storage of the drug, when theproduct is protected from the so-called window-lter, theproduct is quite stable. The adoption of suitable measures inthe processing and packaging of the drug can certainly beeffective in protecting the drug from light degradation andminimizing degradation at insignicant levels. A scheme of thehypothesized degradation of LND under stressing thermal andirradiation conditions is summarized in Fig. 4.

d Real samples

Both thermal and photodegradation tests were directly per-formed on real samples of tablets from three different Dor-idamine� packages (LND 30 mg) and on the tablets in theoriginal rst packaging.

For each package, six tablets were exposed to 40 and 70 �C,respectively, and analyzed by HPLC at 2, 5, 10, 20, 30 days. Thesamples showed no signicant change aer an exposure time of10 days; aer that traces of the decarboxylated product LDC,under the LOD value, were detected.

Six tablets for each experiment were treated with forcedirradiation in the light cabinet under both the conditionsreported above and analyzed by HPLC at 60, 120, 240, 480 and960 min. When irradiation between 320 and 800 nm wasadopted, only traces of ICA were detected but the content ofLND did not show signicant decrease. Under the stressingirradiation with wavelengths below 320 nm, the peaks of the

itions.

Anal. Methods, 2013, 5, 1715–1720 | 1719

Analytical Methods Paper

degradants LDC and ICA were revealed in traces aer 240 minbut their content did not increase signicantly up to 960 min.The content of LND did not change signicantly along theentire test. The photodegradation tests on the tablets in theirpackaging showed in all the tests a complete stability of theformulation.

Conclusions

Application of a HPLC-DAD method to the thermal and photo-degradation experiments allowed us to describe the degrada-tion pathway of the anticancer lonidamine. The drugdegradation process was shown to proceed with two differentmechanisms when exposed to light or high temperature. Majordegradation of the drug substance was found to occur in solu-tion under forced irradiation to form photoproducts fromdebenzylation and decarboxylation whereas the heating of thedrug solutions caused mainly decarboxylation. The resultscarried out from the validation of the HPLC method wereprecise and accurate and limits of detection of degradationproducts were sufficiently low. Analysis of the LND pure powderand tablet formulations demonstrated good stability and onlytraces of two degradants were revealed aer 240 min understressing conditions. The results reported in this work could beuseful to identify precautionary measures needed inmanufacturing or in formulation of the LND products and theHPLCmethod can be used for routine analysis of LND in qualitycontrol laboratories.

Acknowledgements

This research was supported by grants from M.I.U.R. (Italy) andEuropean Commission, FSE fund and Calabria Region.

References

1 B. Silvestrini, Semin. Oncol., 1991, 18, 2–6.2 E. Calvi~no, M. C. Esta~n, G. P. Simon, P. Sancho, M. delCarmen Boyano-Adanez, E. de Blas, J. Breard and P. Aller,Biochem. Pharmacol., 2011, 82, 1619–1629.

1720 | Anal. Methods, 2013, 5, 1715–1720

3 D. Amadori, G. L. Frassineti, A. De Matteis, G. Mustacchi,A. Santoro, S. Cariello, M. Ferrari, O. Nascimben, O. Nanni,A. Lombardi, E. Scarpi and W. Zoli, Breast Cancer Res.Treat., 1998, 49, 209–217.

4 R. Rosso, D. Amoroso, G. Gardin, L. Miglietta, T. Guido,M. Pace, L. Repetto and P. Pronzato, Semin. Oncol., 1991,18, 62–75.

5 C. Y. Cheng, M. Mo, J. Grima, L. Saso, B. Tita, D. Mruk andB. Silvestrini, Contraception, 2002, 65, 265–268.

6 S. Di Cosimo, G. Ferretti, P. Papaldo, P. Carlini, A. Fabi andF. Cognetti, Drugs Today, 2003, 39, 157–174.

7 G. Robustelli della Cuna and P. Pedrazzoli, Semin. Oncol.,1991, 18, 18–22.

8 M. T. Gatto, B. Tita, M. Artico and L. Saso, Contraception,2002, 65, 277–278.

9 G. Corsi, G. Palazzo, C. Germani, P. S. Barcellona andB. Silvestrini, J. Med. Chem., 1976, 19, 778–783.

10 V. Ram, G. Kher, K. Dubal, B. Dodiya and H. Joshi, Pharm. J.,2011, 19, 159–164.

11 A. Mohammadi, I. Haririan, N. Rezanour, L. Ghiasi andR. B. Walker, J. Chromatogr., A, 2006, 1116, 153–157.

12 I. Ivana, Z. Ljiljana and Z. Mira, J. Chromatogr., A, 2006, 1119,209–215.

13 ICH, Q1A(R2). Stability testing of new drug substances andproducts, International Conference on Harmonization,IFPMA, Geneva, 2003.

14 R. Leclaire, J. G. Besner, P. Band, S. Mailhot, P. Gervais, A. DeSanctis and M. Deschamps, J. Chromatogr., Biomed. Appl.,1983, 277, 427–432.

15 C. Bottalico, G. Micelle, A. Guerrieri, F. Palmisano,V. Lorusso and M. De Lena, J. Pharm. Biomed. Anal., 1995,13, 1349–1353.

16 C. Vetuschi, G. Ragno, L. Baiocchi and P. Ridol, Spectrosc.Lett., 1989, 22, 375–385.

17 C. Vetuschi and G. Ragno, Anal. Lett., 1989, 22, 1685–1692.18 Product monograph in http://chemyq.com/En/xz/xz7/630

26kyhqo.htm.19 D. L. Massart, B. Vandeginste, L. Buydens, S. De Jong, P. Lewi

and J. Smeyers-Verbeke, Handbook of Chemometrics andQualimetrics: Part A, Elsevier, Amsterdam, 1997.

This journal is ª The Royal Society of Chemistry 2013