RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom.

2010; 24: 16651672 Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.4525

Direct analysis of pharmaceutical tablet formulations using matrix-assisted laser desorption/ionisation mass spectrometry imaging
Caroline J. Earnshaw1, Vikki A. Carolan1, Don S. Richards2 and Malcolm R. Clench1*
1 2

Biomedical Research Centre, Shefeld Hallam University, Howard Street, Shefeld S1 1WB, UK Pzer Global R&D, Ramsgate Road, Sandwich CT13 9NJ, UK

Received 6 January 2010; Revised 24 February 2010; Accepted 24 February 2010

Matrix-Assisted Laser Desorption/Ionisation Mass Spectrometry Imaging (MALDI MSI) has been used to directly analyse a range of tablets in order to assess the homogeneity of the active drug compound throughout the excipients contained within the tablets studied. The information gained from the imaging experiments can be used to improve and gain a greater understanding of the manufacturing process; such knowledge will enable improvements in nished product quality to make safer and more efcacious tablet formulations. Commercially available and prescription tablet formulations have been analysed, including aspirin, paracetamol, sildenal citrate (Viagra1) and a batch of tablets in development (tablet X: placebo; 1 mg; 3 mg and 6 mg). MALDI MSI provides semiquantitative information that is related to ion abundance, therefore Principal Component Analysis (PCA), a multivariate analysis technique, has been used to differentiate between tablets containing different amounts of active drug ingredient. Aspects of sample preparation have also been investigated with regard to tablet shape and texture. The results obtained indicate that MALDI MSI can be used effectively to analyse the spatial distribution of the active pharmaceutical component (API) in pharmaceutical tablet formulations. Copyright # 2010 John Wiley & Sons, Ltd.
Tablets are compacted pharmaceutical dosage forms that contain active drug compounds and excipients such as bulking agents, lubricants and disintegrants. There are many different tablet formulations and pharmaceutical manufacturing techniques that can be used to produce them. The manufacturing process can vary depending on the chemical properties of the active pharmaceutical ingredient (API). It is of extreme importance to the pharmaceutical industry to evaluate the outcomes of the drug formulation process, and in particular for solid dosage formulations, viz. tablets, to study the distribution of constituents within them. Examining the nished pharmaceutical product, for example tablets, is one way to achieve this and it is vital for slow release formulations where the distribution of the active drug ingredient throughout the excipients can affect how effective the tablet is post-administration. The introduction of the process analytical technology (PAT) initiative by the US Food and Drugs Administration (FDA) has made pharmaceutical companies aware of the increasing need to improve manufacturing efciency and nished product quality.1 Imaging techniques have formed part of this initiative and they have already been used to investigate tablet formulations. Techniques used include magnetic resonance imaging (MRI),2 Raman and near infrared (NIR) chemical imaging.35 Raman spectroscopy
*Correspondence to: M. R. Clench, Biomedical Research Centre, Shefeld Hallam University, Howard Street, Shefeld S1 1WB, UK. E-mail: m.r.clench@shu.ac.uk

in particular has had an important role in the analysis of solid pharmaceutical products owing to the minimal sample preparation required. APIs tend to give a good Raman response and there is less spectral overlap in Raman spectra than is observed with NIR. It is possible to use Raman spectroscopy to map APIs in pharmaceutical tablet formulations and this has been successfully demonstrated by Sasic.6 Due to the complexity of the data and the lack of imaging software, this can, however, be a difcult process. These imaging techniques can help to identify compositional or structural defects in tablets that may have an impact on their mechanical stability and drug-releasing properties. Imaging techniques can also assist in pharmaceutical safety strategies, for example in the investigation of cases of counterfeit pharmaceutical products. Nyadong et al. have demonstrated the use of imaging desorption electrospray ionisation to investigate counterfeit antimalarial drugs.7 A range of mass spectrometry techniques has also been utilised in the study of pharmaceutical formulations. Secondary ion mass spectrometry (SIMS) is a surface ionisation technique where secondary ions are generated by focusing a pulsed primary ion beam on the sample surface. The secondary ions that are generated are usually analysed using a time-of-ight (TOF) mass spectrometer. Prestidge et al. have demonstrated the use of TOF SIMS for the characterisation of solid-state pharmaceutical products.8 SIMS can be used with imaging software to study the distribution of the analyte of interest within a sample.
Copyright # 2010 John Wiley & Sons, Ltd.

1666 C. J. Earnshaw et al.

Another mass spectrometry technique that has been used for the analysis of pharmaceutical drug formulations is desorption electrospray ionisation mass spectrometry (DESI MS). In DESI MS charged droplets created by an electrospray source are concentrated onto a solid sample, resulting in surface interaction creating secondary ions that can be detected using MS. DESI MS has been used to study pharmaceutical drug formulations9,10 and for the screening of illicit Ecstasy tablets.11 Whilst DESI requires no sample preparation, is relatively fast and is a soft ionisation technique, the spatial resolution available for imaging experiments is only of the order of 0.25 mm12 and selective extraction based on the desorption solvent used would appear to be a signicant issue in DESI experiments. Matrix-assisted laser desorption/ionisation (MALDI) MS, a soft ionisation technique, utilises a chemical matrix coated onto the sample of interest to act as an energy transfer intermediate between the laser and the sample.13,14 The role of the matrix is primarily to absorb at the laser wavelength and transfer energy to the analyte in order to promote efcient ionisation. The most commonly used lasers in MALDI MS experiments are UV lasers nitrogen (l 337 nm) and Nd:YAG (l 355 nm and l 266 nm) and, since the majority of MALDI MS experiments that have been published to date have been conducted in positive ion mode, the most common matrices employed in such studies are strongly UV-absorbing organic acids. The most commonly used organic acid matrices include a-cyano-4-hydroxycinnamic acid (a-CHCA), dihydroxybenzoic acid (DHB) and sinapinic acid (3, 5-dimethoxy4-hydroxycinnamic acid). Matrix-assisted laser desorption/ionisation mass spectrometry imaging (MALDI MSI)14,15 is a relatively new technique that offers all of the benets of conventional mass spectrometry such as rapid analysis time, specicity and sensitivity. It can be used for the analysis of a range of samples. Data is obtained by recording mass spectra as a sample is moved by set increments under a stationary laser. By plotting the spatial dimensions of the x and y axes against the abundance/intensity of selected ion/ions, images can be generated. Mass spectrometry permits the parallel detection of multiple analytes thus providing a wealth of information about the sample of interest and this is why MALDI MSI has been used in a wide range of biomedical applications. MALDI MS has been applied previously to the analysis of street tablets containing prohibited substances.16 The tablets were ground into a ne powder and dissolved in methanol in order to permit conventional MALDI analysis to be performed using a combination of a-CHCA and cetyltrimethylammonium bromide (CTAB) in a water/ acetonitrile solution (50:50 v/v) as matrix. The work presented herein demonstrates the use of MALDI MSI to study whole tablets for the rst time. The aim of this work was to map the distribution of the active drug throughout all the excipients contained within tablets. MALDI MSI has been used to directly image the distribution of the active drug throughout the tablet, and also to provide some quantitative information in relation to the manufactured dose of the tablet. The following tablets were obtained for MALDI MSI analysis: an unknown tablet formulation, tablet X (placebo,
Copyright # 2010 John Wiley & Sons, Ltd.

Figure 1. Photograph of tablet X showing the curved surface of the tablet and the position at which the horizontal section was obtained using a commercially available tablet cutter.

1 mg, 3 mg and 6 mg dosage forms); sildenal citrate (Viagra 25 mg); paracetamol (500 mg); aspirin (75 mg); Anadin Extra (a mixture of aspirin (300 mg), paracetamol (200 mg) and caffeine (45 mg) and Solpadeine (a mixture of paracetamol (500 mg) and caffeine (65 mg)). Principal component analysis (PCA) is a form of multivariate analysis that can be used to simplify multidimensional datasets. In previous published work, PCA has been incorporated into the interpretation of MALDI MSI data to discriminate between different regions in mammalian tissue sections1719 and to evaluate data from the analysis of porcine skin that had been treated with hydrocortisone.20 PCA is commonly used to reduce the dimensionality of multivariate data sets while retaining most of the original information content. Discriminant analysis (DA) is often used to dene a classication from an observation of pre-dened groups. PCA-DA is a two-stage supervised statistical technique where the number of variables in the data set is rst reduced by the use of PCA before a second discriminant analysis is performed. PCA-DA takes into account user-supplied information about external variables (group information) when reducing the dimensionality of the datasets in the rst stage. Therefore, it is better suited than PCA for clustering analysis.21 Here we have demonstrated how PCA-DA can be used to differentiate between the active drug component and the excipients contained within the tablet and also how PCADA can be used to provide semi-quantitative information.

EXPERIMENTAL (i) Materials

a-Cyano-4-hydroxycinnamic acid (a-CHCA), ethanol and triuoroacetic acid (TFA) were purchased from Sigma Aldrich (Poole, UK). The following tablets were purchased commercially: paracetamol 500 mg (Wallis Laboratory Ltd., Luton, UK), aspirin 75 mg (M & A Pharmachem Ltd., Bolton, UK), Anadin Extra (Wyeth Consumer Healthcare, South Taplow, UK), and Solpadeine (GlaxoSmithKline Ltd., Dungarvan, Ireland). Sildenal citrate (Viagra) 25 mg dosage form and tablet X in placebo, 1 mg, 3 mg and 6 mg dosage forms were supplied by Pzer (Sandwich, UK). The chemical
Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

Direct analysis of tablet formulations using MALDI-MSI

1667

Copyright # 2010 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

1668 C. J. Earnshaw et al.

Copyright # 2010 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

Direct analysis of tablet formulations using MALDI-MSI

1669

Figure 2. (i) Full scan representative spectra for tablet X: (a) placebo; (b) 1 mg; (c) 3 mg; and (d) 6 mg, respectively. The matrix peaks are clearly visible in the spectra; and have been indicated with an asterisk (); however, as the active drug concentration increases ([M H] at m/z 507), suppression of the matrix peaks is observed. (ii) Representative images for tablet X: (a) 1 mg; (b) 3 mg; and (c) 6 mg. Images shown are for [M H] at m/z 507. Images have been normalised against the corresponding matrix peak [M H] at m/z 190. Images intensities were normalised against the corresponding protonated matrix peak (m/z 190.05) and indicate that the active drug component (compound X) is relatively distributed throughout the tablet at the different concentrations: (a) 1 mg; (b) 3 mg; and (c) 6 mg. Note the maximum intensity for m/z 507 in the placebo data was 3 counts; hence, no image was observed for the distribution of this m/z following normalisation to m/z 190 and setting of the intensity scale to 1 104. (iii) Principal component analysisdiscrimant analysis (PCA-DA) of tablet X: placebo, 1 mg, 3 mg and 6 mg tablet formulations. Five random locations from each tablet obtained from the image data were selected for PCA-DA. The scores plot shows signicant grouping for the three different dose tablets and a separate grouping for the placebo. From the scores plot it can be seen that the more active drug the tablet contains, the higher the value of PC2 and the lower the value of PC1. As the placebo does not contain any active drug, these samples give the lowest value of PC2 and the highest value of PC1.

matrix (a-CHCA) was made to a concentration of 25 mg/mL in EtOH containing 0.1% TFA.

(ii) Tablet preparation

Tablet sections approximately 1 mm in thickness were created using a tablet cutter to fracture the tablet, thus exposing the surface of the tablet. (This procedure was not necessary for the paracetamol and aspirin tablets as they already had at surfaces). The underside of the tablet was led down in order to ensure that it would t into a recessed MALDI target plate. The exposed surface of the tablet was coated with a chemical matrix (a-CHCA 25 mg/mL in EtOH containing 0.1% TFA) prior to MALDI analysis by an airspray technique. The shape of the tablet can dictate the sample preparation to some extent. For example, if a tablet had a convex surface this tablet would need to be sectioned in a way that would produce a relatively at surface this was required in order to perform MALDI analysis on the tablet X dosage forms. A photograph of one of the tablets studied is shown in Fig. 1 indicating the direction of fracture and showing the curved nature of the surface.
Copyright # 2010 John Wiley & Sons, Ltd.

Since MALDI is a surface ionisation technique, tablet coating (examples include hypromellose and triacetin) can prove to be problematic when conducting imaging experiments, although if the tablet coating is of a consistent thickness over the tablet it may be possible to establish a method that could ablate through the coating layer by ring the laser for a longer period of time in each location. The shape and texture of tablets such as Solpadeine and Anadin Extra made it very difcult to obtain good quality sections; therefore, images could not be obtained. However, with such tablets it is possible to carry out proling experiments to detect the APIs using a simple spot analysis procedure, where the tablet is dissolved in a suitable solvent, combined with the matrix and spotted onto a MALDI spot target plate for analysis.

Instrumentation Mass spectrometric analysis

All analyses were performed in positive ion mode on a QStar Pulsar-i hybrid quadrupole time-of-ight mass spectrometer (Applied Biosystems/MDS Sciex, Concord,
Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

1670 C. J. Earnshaw et al.

Figure 3. (i) MALDI MS images of: (a) Viagra [M H]; (b) paracetamol [M H]; and (c) aspirin [M Na] in tablet formulations. (ii) MALDI mass spectra for (a) Viagra; (b) paracetamol; and (c) aspirin.

Copyright # 2010 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

Direct analysis of tablet formulations using MALDI-MSI

1671

Ontario, Canada) with an orthogonal MALDI source and a high repetition Nd:YAG laser (1 kHz). A bre optic delivers the Nd:YAG laser to the sample target with an elliptical spot size of 100 150 mm. The laser was scanned over the sample in a raster pattern and acquired data at 100 mm increments via a discrete spot-to-spot method. The laser was red at a power of 30% (3.2 mJ) at each location for 2 s.

Visualisation software
MALDI MS images were generated using oMALDI Server software (version 5.0; Applied Biosystems/MDS Sciex).

Principal component analysis (PCA)

PCA was performed using the Markerview software package (Applied Biosystems/MDS Sciex). Five random locations from the tablet image data were taken and these data entered into Markerview. PCA-DA was performed using Pareto scaling. This type of scaling calculates the square root of the standard deviation and uses this information as a scaling factor.22 This in turn reduces any major intensity changes that are sometimes observed within the sample and also with the matrix peaks.

RESULTS AND DISCUSSION Analysis of tablet formulations

The drugs studied can be clearly observed both in the spectra and in the images as protonated molecules, or as sodium adducts, such as with aspirin. As can be seen in the tablet images (Figs. 2(ii) and 3(i)), the top edge of the tablets studied has an apparently increased concentration of the active ingredient. This we attribute to the laser position and the angle at which the laser ablates the tablet surface. Thus, as a consequence, the top edge of the tablet is sampled in addition to the tablet surface. The top edge of the tablet is not coated so extensively with matrix during the sample preparation procedure and hence normalisation of the data to a matrix peak could articially enhance the signal from the active drug in this region. The shape distortion at the top of each tablet supports the proposed theory, in that the images acquired were not perfectly circular (for paracetamol and aspirin) or rhomboid (for Viagra) in shape whereas the tablets were circular. Figure 2(i) shows a collection of representative spectra for the analysis of tablet X: (a) placebo; (b) 1 mg tablet; (c) 3 mg tablet and (d) 6 mg tablet. In Figs. 2(i)bd the protonated molecule ([M H] at m/z 507) of the active drug is clearly visible as is the sodium adduct ([M Na]at m/z 529). The full scan spectra shows a relative increase in ion counts (that have been normalised to the protonated matrix molecule at m/z 190) when the concentration of the active drug increases. The ion counts for the matrix peaks decrease as the ion intensity for the active drug increases. With higher concentrations of this particular drug the matrix peaks are suppressed. This is because the active drug, in this case, is easily ionisable. Figure 2ii shows the corresponding images based on the distribution of m/z 507 throughout the tablet sections. These have been normalised against the corresponding matrix peak ([M H] at m/z 190). The intensity
Copyright # 2010 John Wiley & Sons, Ltd.

scale on each of the images represents ion abundance for m/z 507. The scale maximum has been set to 1 104 counts in each image to allow direct comparison; thus the deeper the colour, the more of the active drug there is in that particular location. Note that the maximum intensity for m/z 507 in the placebo data was 3 counts; hence no image was observed for the distribution of ions of this m/z value following normalisation to m/z 190 and setting the intensity scale to 1 104. PCA-DA was used to explore how the four tablets (placebo, 1 mg, 3 mg and 6 mg) differ statistically. Figure 2(iii) shows the results from the PCA-DA analysis and shows the following information. The scores plot shows signicant grouping for the three different dose tablets and a separate grouping for the placebo. The loadings plot conrms that the more active drug the tablet contains, the more it groups towards m/z 507. As the placebo does not contain any active drug, it groups furthest away from m/z 507. These data were used to construct a calibration graph which exhibited good linearity with an R2 value of 0.9963. The matrix peaks were not excluded from this analysis, in order to illustrate further that when the tablets contain a readily ionisable active drug, the signal from it can suppress the matrix signal. As the placebo does not contain a component that is readily ionisable the matrix peaks suppress the excipient peaks. Therefore, it is an interesting observation that although one would expect the grouping for the placebo to be furthest away from the active drug m/z 507, this is also where the matrix peaks are located in the loadings plot. As MALDI MSI can provide semi-quantitative information, the PCA-DA results show that the more of the active drug is contained within a tablet the more it will group towards the corresponding mass in the loadings plot. Thus, PCA-DA supports the semi-quantitative aspect of our investigation and it may also provide information on matrix suppression.

Sildenal citrate (Viagra 25 mg)

Figure 3(i)A shows the distribution of m/z 475 (the sildenal base [M H] ion) throughout a section of a Viagra tablet. These data have been normalised against the corresponding matrix peak ([M H] at m/z 190). The full scan spectrum shows that the sildenal base dominates the spectrum, suppressing the ions associated with the matrix. The Na adduct is also present at m/z 497.

Paracetamol (500 mg) and aspirin (75 mg)

Figure 3(i)B shows the distribution of the protonated paracetamol ([M H] at m/z 152) throughout a tablet section. These data have been normalised against the corresponding matrix peak ([M H] at m/z 190). The image and the corresponding ion intensity scale show that the distribution of paracetamol is consistent and appears homogenous throughout the section, although the edges of the tablet seem to be brighter. Figure 3(i)C shows the distribution of the aspirin sodium adduct ([M Na] at m/z 203) throughout a tablet section. These data have been normalised against the corresponding matrix peak ([M Na] at m/z 212). The sodium adduct rather than the lower intensity protonated molecule
Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm

1672 C. J. Earnshaw et al.

([M H] at m/z 181) was imaged. The image shows that the aspirin appears to be located on the outer parts of the tablet as the ion counts for the middle part of the tablet section indicate that there is less aspirin in the middle of the tablet. The high concentration of active drug ingredient at the tablet surface may be due to greater compression at the surface. Such compression may cause an increase in density at the surface when the tablet is pressed. Increased tablet density would increase the spatial concentration of the API without increasing its actual concentration.

Acknowledgements
Funding for Dr C. J. Earnshaws PhD from EPSRC/RSC is gratefully acknowledged.

REFERENCES
1. Gowen AA, ODonnell CP, Cullen PJ, Bell SEJ. Eur. J. Pharm. Biopharm. 2008; 69: 10. 2. Richardson JC, Bowtell RW, Mader K, Melia CD. Adv. Drug Delivery Rev. 2005; 57: 1191. 3. Sasic S. Appl. Spectrosc. 2007; 61: 239. 4. Hilden LR, Pommier CJ, Badawy SIF, Friedman EM. Int. J. Pharm. 2008; 353: 283. 5. Wolff J-C, Lopes MB. Anal. Chim. Acta 2009; 633: 149. 6. Sasic S. Pharm. Res. 2007; 24: 58. 7. Nyadong L, Harris GA, Balayssac S, Galhena AS, MaletMartino M, Martino R, Parry RM, Wang MD, Fernandez FM, Gilard V. Anal. Chem. 2009; 81: 4803. 8. Prestidge CA, Barnes TJ, Skinner W. J. Pharm. Pharmacol. 2007; 59: 251. 9. Weston DJ, Bateman R, Wilson ID, Wood TR, Creaser CS. Anal. Chem. 2005; 77: 7572. 10. Williams JP, Scrivens JH. Rapid Commun. Mass Spectrom. 2005; 19: 3643. 11. Leuthold LA, Mandscheff JF, Fathi M, Giroud C, Augsburger M, Varesio E, Hopfgartner G. Rapid Commun. Mass Spectrom. 2006; 20: 103. 12. Wisemann JM, Ifa DR, Song Q, Cooks RG. Angew. Chem. Int. Ed. 2006; 45: 7188. 13. Tanaka K, Waki H, Ido Y, Akita S, Yoshida Y, Yoshida T. Rapid Commun. Mass Spectrom. 1988; 2: 151. 14. Karas M, Hillenkamp F. Anal. Chem. 1988; 60: 2299. 15. Caprioli RW, Farmer TB, Gile J. Anal. Chem. 1997; 69: 4751. 16. Su AK, Liu JT, Lin CH. Talanta 2005; 67: 718. 17. McCombie G, Staab D, Stoeckli M, Knochenmuss R. Anal.Chem. 2005; 77: 6118. 18. Trim PJ, Atkinson SJ, Princivalle AP, Marshall PS, West A, Clench MR. Rapid Commun. Mass Spectrom. 2008; 22: 1503. 19. Djidja M-C, Carolan V, Loadman PM, Clench MR. Rapid Commun. Mass Spectrom. 2008; 22: 1615. 20. Prideaux B, Atkinson SJ, Carolan VA, Morton J, Clench MR. Int. J. Mass Spectrom. 2007; 260: 243. 21. Hoogerbrugge R, Willig SJ, Kistemaker PG. Anal Chem. 1983; 55: 1710. 22. Van den Berg RA, Hoefsloot HCJ, Westerhuis JA, Smilde AK, van der Werf MJ. BMC Genom. 2006; 7: 142.

CONCLUSIONS
A novel approach utilising MALDI MSI for the direct analysis of tablets has been demonstrated in this study. The sensitivity, specicity and rapid analysis time, together with the imaging capabilities, make MALDI MSI a valuable technique in many biomedical disciplines. The introduction of MALDI MSI has been invaluable in pharmaceutical analysis both in animal models of drug distribution and in solid pharmaceutical formulations. The importance of being able to visualise the spatial localisation of a compound of interest within a sample, in this case tablet formulations, has been discussed. Tablet shape and texture can dictate the extent to which this technique can be applied. Images have been obtained for tablet X, Viagra, paracetamol and aspirin; however, due to the crumbly nature of the Solpadeine and Anadin Extra tablets, it was very difcult to obtain intact tablet sections of $1 mm in thickness. Therefore, images could not be obtained for these tablets. The observed edge effects suggest that the position of the laser in relation to the tablet section is a critical factor in the success of tablet imaging experiment. Future studies could include re-adjustment of the laser/sample position to try to combat this issue.

Copyright # 2010 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2010; 24: 16651672 DOI: 10.1002/rcm