European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147

Contents lists available at ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics

journal homepage: www.elsevier .com/locate /e jpb

Research paper

In situ dissolution analysis using coherent anti-Stokes Raman scattering(CARS) and hyperspectral CARS microscopy

0939-6411 � 2013 The Authors. Published by Elsevier B.V.http://dx.doi.org/10.1016/j.ejpb.2013.08.012

⇑ Corresponding author. Optical Sciences Group, Faculty of Science and Technol-ogy, MESA+ Institute of Nanotechnology, University of Twente, P.O. Box 217,7500AE Enschede, The Netherlands. Tel.: +31 534891097; fax: +31 534893511.

E-mail addresses: A.L.Fussell@utwente.nl (A. Fussell), E.T.Garbacik@utwente.nl(E. Garbacik), H.L.Offerhaus@utwente.nl (H. Offerhaus), Kleinebudde@uni-duessel-dorf.de (P. Kleinebudde), clare.strachan@helsinki.fi (C. Strachan).

Open access under CC BY-NC-ND license.

Andrew Fussell a, Erik Garbacik a, Herman Offerhaus a,⇑, Peter Kleinebudde b, Clare Strachan c

a Optical Sciences Group, Faculty of Science and Technology, MESA+ Institute of Nanotechnology, University of Twente, Enschede, The Netherlandsb Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine University, Düsseldorf, Germanyc Faculty of Pharmacy, Division of Pharmaceutical Technology, University of Helsinki, Helsinki, Finland

a r t i c l e i n f o

Article history:Received 13 March 2013Accepted in revised form 20 August 2013Available online 28 August 2013

Keywords:Coherent anti-Stokes Raman scatteringMicroscopyIn situ analysisSolid-state formDissolutionTheophylline

.

a b s t r a c t

The solid-state form of an active pharmaceutical ingredient (API) in an oral dosage form plays an impor-tant role in determining the dissolution rate of the API. As the solid-state form can change during disso-lution, there is a need to monitor the oral dosage form during dissolution testing. Coherent anti-StokesRaman scattering (CARS) microscopy provides rapid, spectrally selective imaging to monitor the oral dos-age form during dissolution. In this study, in situ CARS microscopy was combined with inline UV absorp-tion spectroscopy to monitor the solid-state change in oral dosage forms containing theophyllineanhydrate undergoing dissolution and to correlate the solid-state change with a change in dissolutionrate. The results from in situ CARS microscopy showed that theophylline anhydrate converted to theoph-ylline monohydrate during dissolution resulting in a reduction in the dissolution rate. The addition ofmethyl cellulose to the dissolution medium was found to delay the theophylline monohydrate growthand changed the morphology of the monohydrate. The net effect was an increased dissolution rate fortheophylline anhydrate. Our results show that in situ CARS microscopy combined with inline UV absorp-tion spectroscopy is capable of monitoring oral dosage forms undergoing dissolution and correlatingchanges in solid-state form with changes in dissolution rate.

� 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license

1. Introduction

Active pharmaceutical ingredients (API) commonly exist in var-ious crystalline forms, known as polymorphs, with different molec-ular arrangements and/or conformations. Multi-componentcrystals, where one or more additional compounds are incorpo-rated into the crystal lattice, may also be formed and include salts,co-crystals, and solvates. A widely observed solvate is the hydrate,where water molecules have been incorporated into the API’s crys-tal lattice. Single and multi-component API solids may also exist ina higher energy, disordered amorphous form.

The solid-state form of an API is an important parameter in thedevelopment of oral dosage formulations. It has an effect on thechemical and physical stability, processibility, solubility, dissolu-

tion rate, and potentially bioavailability of the API. In dynamicenvironments, solid-state changes in pharmaceutical materialsare common: there have been numerous reports on solid-stateforms undergoing changes during processing [1–3] and storageand dissolution [4–6].

Solid-state changes that occur during dissolution when a meta-stable form (higher solubility) converts to a stable form (lower sol-ubility) through precipitation from a supersaturated solution arecalled solvent-mediated phase transformations [7]. The kineticsof a solvent-mediated phase transformation are determined bythe relative rates of dissolution and growth of the two phases[7,8]. Solution-mediated transformations of APIs are well docu-mented. Murphy et al. [5] studied the conversion of carbamazepine(CBZ) form III to the dihydrate form in samples undergoing disso-lution testing. They found that the conversion time depended ongrinding and storage conditions of the form III [5]. Savolainenet al. [9] studied the dissolution of amorphous indomethacin(IMC) and compared this to the dissolution of a and c forms ofIMC. As expected, they found that the initial intrinsic dissolutionrate for amorphous IMC was faster than for either crystalline form.However, the dissolution rate decreased as the sample surfaces be-gan to crystallize to a-IMC during dissolution. Aaltonen et al. [10]have reported the conversion of theophylline (TP) anhydrate (TPa)to theophylline monohydrate (TPm) during dissolution testing.

1142 A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147

Decreased dissolution rates due to solid-state transformationsduring dissolution testing have led to investigations into methodsfor preventing or reducing such transformations. A number ofstudies have found that some excipients are capable of inhibitingor delaying the hydrate formation. Katzhendler et al. [11] first re-ported hydroxypropylmethylcellulose (HPMC) inhibiting thetransformation of CBZ to CBZ dihydrate on tablets in aqueoussolutions. They suggest that hydroxyl groups from HPMC may at-tach to CBZ at the site of water binding, thereby inhibiting thegrowth of the dihydrate form. More recently, Wikstrom et al.[12] investigated the effect of pharmaceutical excipients on TPmformation during wet granulation. They found that the majorityof excipients tested did not change the transformation kineticsof TP. However, polymeric binders such as methyl cellulose(MC) and HPMC could significantly inhibit the conversion of TPato TPm during wet granulation. Wikstrom et al. [12] suggestedthat the inhibitory polymers adsorb to fast-growing surfaces ofthe hydrate crystal inhibiting crystal growth and causing morpho-logical changes.

Dissolution testing is an important part of oral solid dosageform development since the dissolution behavior of a drug is akey determinant of therapeutic efficacy for many drugs. This isparticularly important for poorly soluble drugs, as drugs mustdissolve first before being absorbed [13]. Standard pharmacopoe-ial methods require the immersion of the drug (e.g., as a compact)in a flowing dissolution medium with samples of the dissolutionmedium being removed over a series of time intervals to be ana-lyzed for drug concentration in solution using UV absorptionspectroscopy or HPLC [14]. Analyzing solution concentration pro-vides information about how much drug is dissolved. However, itdoes not give any direct information about physical changes onthe surface of the dissolving dosage form, including solid-state changes, which can affect dissolution behavior. Direct anal-ysis of the solid dosage form changes during dissolution testingcan therefore provide improved understanding of dissolutionbehavior.

In situ analysis of the solid drug or dosage form during disso-lution testing places a number of restrictions on the suitable ana-lytical techniques. Firstly, the technique must be nondestructiveto the sample. Secondly, the technique must be able to obtaindata in the presence of dissolution medium with a sufficient tem-poral resolution (on the order of seconds) to observe rapidchanges in the sample. Finally, the technique must not interferewith the dissolution process. Common solid-state techniques suchas X-ray powder diffraction (XRPD), scanning electron microscopy(SEM), near infrared (NIR), and infrared (IR) spectroscopy all havelimiting factors preventing their use in situ for dissolution. Theradiation used in NIR and IR imaging is strongly absorbed bywater limiting their use in aqueous environments [15,16]. Scan-ning electron microscopy has been used to characterize solid-state changes on the surfaces of dosage forms after dissolution[10,17]. X-ray powder diffraction has been used to depth profilephase changes on samples undergoing dissolution related solid-state changes [17]. However, both SEM and XRPD are unsuitablefor in situ analysis of dissolution due to sample preparationrequirements. Spontaneous Raman spectroscopy has been shownto be suitable for in situ analysis of solid-state transformationsduring dissolution [9,10]. Spontaneous Raman spectroscopy hasthe advantage that it can generate full vibrational spectra in a rel-atively short period of time. Coupled with a flow through cell andUV flow through absorption spectroscopy, it has been used in situto monitor various solid-state conversions and their effects ondissolution, including transformation from TPa to TPm, and thecrystallization of amorphous IMC and CBZ. However, spontaneousRaman spectroscopy gives no spatial information, meaning that itis not capable of identifying where solid-state conversions are

occurring during dissolution and techniques developed to mapRaman intensity are slow (minutes to hours), precluding in situanalysis during dissolution testing.

Instead, we use coherent anti-Stokes Raman scattering (CARS)microscopy as a tool for in situ analysis during dissolution. Asummary of CARS microscopy is provided in [18]. Briefly, CARSmicroscopy is capable of rapid spectrally- and spatially-resolvedimaging allowing the visualization of different solid-state formsof drugs based on their Raman vibrational spectra. A narrowbandCARS setup typically utilizes two synchronized pulsed lasers, oneof which is tuneable in wavelength. The two laser beams are tem-porally and spatially overlapped before being focused on the sam-ple. If the frequency difference between the two laser beamsmatches a Raman active vibrational mode, an anti-Stokes (blue-shifted with respect to input beams) signal is produced. Ramanvibrational modes are specific for compounds or groups of com-pounds providing chemically specific images. As the CARS signalis produced only within the focal volume of the lasers, the processis inherently confocal, allowing resolution down to the diffractionlimit in three dimensions.

CARS is a third order non-linear optical technique that probesthe same molecular vibrational frequencies as spontaneous Ramantechniques. This means that CARS spectra are comparable to butnot the same as Raman spectra. Coherent Raman techniques suchas CARS have about 100 times faster imaging speed when com-pared to spontaneous Raman mapping techniques [19]. Spontane-ous Raman techniques collect information over a wide spectralrange, while narrowband coherent Raman techniques collect infor-mation from only a single Raman shift. Broadband coherentRaman techniques are able to probe a wider spectral region(600–3200 cm�1) but with a reduced spectral resolution (about10 cm�1) and a slower imaging speed (50 ms/pixel) when com-pared to narrowband coherent Raman techniques [20]. As CARSproduces anti-Stokes shifted signal (blueshifted with respect toexcitation pulses), it is free from single photon fluorescence, whichhampers spontaneous Raman measurements. Unlike spontaneousRaman where the anti-Stokes scattering is much weaker than theStokes scattering, the CARS process actively drives molecules intoa specific vibrational mode and therefore generates significantlymore signal with reduced temperature sensitivity.

CARS microscopy has been used to image a few pharmaceuticalsystems during drug release. Kang et al. [21–23] published workwhere they imaged in situ release of paclitaxel from polymericfilms in a static medium (phosphate buffer) using CARS micros-copy. In the first work focusing on orally administered drugs anddosage forms, Windbergs et al. [24] and Jurna et al. [25] used CARSmicroscopy to image the distribution of TP in lipid dosage formsand monitored the release of TP during dissolution in a flowthrough cell setup. They were able to image both drug releaseand conversion from TPa to TPm in real time.

We have developed a new analytical technique to record thedissolved drug concentration and simultaneously monitor solid-state changes on the surface of the oral solid dosage form undergo-ing dissolution. Furthermore, we have applied hyperspectral CARSmicroscopy for improved solid-state form characterization. Weillustrate the use of these techniques using the model drug theoph-ylline (TP) in different dissolution media.

2. Materials and methods

2.1. Materials

USP grade theophylline (TP, 1,3-dimethyl-7H-purine-2,6-dione)anhydrate and monohydrate were gifted from BASF (Ludwigsha-fen, Germany). Methyl cellulose (MC) (Methocel A4C premium)was gifted from Colorcon GmbH (Idstein, Germany).

A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147 1143

2.1.1. Compact preparationWeighed amounts of TPa and TPm (0.49 g) were directly com-

pressed using a force feed tablet press (IMA Kilian Pressima, Italy).The upper punch had a pre-compression height of 9.22 mm and afinal compression height of 3.02 mm using a compaction force ofabout 13 kN, resulting in compacts which had a diameter of12.02 mm and a thickness of about 3 mm. The compression didnot result in changes in the solid-state form, which we confirmedusing hyperspectral CARS microscopy.

Fig. 2. Hyperspectral imaging process illustrated using a sample with two differenttypes of plastic beads (reproduced from [26] with permission of Wiley). (Forinterpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

2.2. Methods

2.2.1. CARS microscopyThe CARS microscopy system is illustrated in Fig. 1 and is de-

scribed in more detail elsewhere [26]. A Nd:YVO4 picosecondpulsed laser (Coherent Inc., USA) operating at a fundamental wave-length of 1064 nm was frequency doubled to pump an opticalparametric oscillator (OPO) (APE Berlin GmbH, Germany), whichproduced two dependently tunable laser beams. The fundamentallaser beam was combined with the signal beam from the OPO anddirected into an inverted laser-scanning microscope (OlympusIX71/FV300, Japan) where they were focused onto the sampleusing a 20�/0.5 NA objective. The backscattered CARS signal wascollected by the focusing objective, spectrally filtered to removethe excitation wavelengths, and detected with a photomultipliertube (Hamamatsu R3896, Japan). The CARS microscope systemhad an axial spatial resolution of about 10 lm and a lateral spatialresolution of about 1 lm.

2.2.2. Hyperspectral CARS imagingHyperspectral CARS imaging provides a method to rapidly and

visually confirm the solid-state form on the surface of an oral dos-age form, both pre- and post-dissolution. Hyperspectral CARSimages were obtained by rapidly imaging the sample while slowlysweeping the wavelength of the OPO in discrete steps, so that eachframe in the image stack corresponds to a different vibrational fre-quency [26]. A color look up table was then applied to the imagestack, with a separate color applied to each frame in the image.Finally, the frames were projected together, resulting in a singletwo-dimensional image wherein each material appears with a un-ique color. This process is illustrated as a diagram in Fig. 2. In thisstudy, 512 � 512 pixel hyperspectral images were collected over arange of 100 cm�1 with each hyperspectral image taking approxi-mately 2 min to record. CARS spectra shown in this article are pixelintensity profiles across the vibrational frequencies and wereextracted from the hyperspectral image data. Further informationabout the collection of CARS spectra can be found in Garbaciket al. [26].

Fig. 1. Schematic illustrating CARS for dissolution setup. (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version ofthis article.)

2.2.3. In situ CARS dissolution imagingIn situ CARS images (512 � 512 pixels) covering 350 � 350 lm

were recorded every 1.12 s (roughly 4.3 ls/pixel dwell time) forthe duration of the dissolution experiments (15 min). All in situCARS images recorded during dissolution testing were recordedat 2952 cm�1 and were false colored green. This peak has been as-signed to antisymmetric C–H stretching in the methyl groups [27]and provided a strong CARS signal for both TPa and TPm.

2.2.4. DissolutionA deuterium light source (DT-MINI-2, Ocean Optics, The Neth-

erlands) was connected by an optical fiber to a Z-shaped flow cell(FIA-Z-SMA, Ocean Optics, The Netherlands) with a 10 mm pathlength An optical fiber connected the Z-shaped flow cell to a CCDspectrometer (USB2000+, Ocean Optics, The Netherlands). Openloop channel flow through intrinsic dissolution was conductedusing a peristaltic pump (Reglo, ISMATEC, Germany), whichpumped dissolution medium (distilled water or methyl cellulose0.45% w/v) through the custom built CARS microscopy dissolutionflow cell and through the Z-shaped UV flow cell at a rate of 5 mL/min. UV spectra were collected at 290 nm every 30 s. Dissolutionwas conducted multiple times on each sample to check forconsistency.

3. Results and discussion

3.1. Hyperspectral CARS analysis of TPa and TPm

CARS spectra of the C–H stretch region were collected prior todissolution experiments on pure TPa and TPm to identify anappropriate vibrational frequency at which to record CARS imagesduring dissolution experiments and for comparison to the beforeand after dissolution hyperspectral scans of the compacts. TheC–H stretch region was chosen as TP has an intense signal in thatregion for both solid-state forms (Fig. 3). Both TPa and TPm fea-tured a peak at around 2950 cm�1, which has been assigned toantisymmetric C–H stretching in the two methyl groups (masC(1,3)-

H3) [27]. There was a peak shift between the two forms in the C–H stretching region of the spectra at a higher Raman shift, withthe TPa peak at around 3120 cm�1 and the TPm peak at around3105 cm�1. This peak has been assigned to the imidazole ring

Fig. 3. CARS C–H stretch region spectra between 2900 and 3200 cm�1 for TPa (blackline), and TPm (red dashed line). (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

Fig. 5. CARS spectra between 3050 and 3150 cm�1 for TPa compacts beforedissolution (black line) and after dissolution where water was used as dissolutionmedium (red dashed line). (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

1144 A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147

C–H stretching (mC(8)–H), and the redshift is due to C(8)–H� � �Ointermolecular hydrogen bonding in the TPm form [27,28]. Thepeak shift allowed us to visualize the change in anhydrate tomonohydrate using hyperspectral imaging. However, the shiftingpeak was not suitable for single-frequency CARS dissolution imag-ing because it was not possible to simultaneously image the TPmcrystal growth on the surface of a TPa compact.

Since both TPa and TPm produce a strong signal at 2952 cm�1,single-frequency CARS images were recorded at this Raman shiftduring dissolution experiments to allow visualization of both TPaand TPm simultaneously. Additionally, at 2952 cm�1, there is verylittle interference due to the presence of water.

Hyperspectral images were recorded before and after dissolu-tion experiments to allow a rapid visual confirmation of the solid-state conversion on the surface of the compact which would beevident as a change in color. Fig. 4A shows the pre-dissolutionhyperspectral image for a TPa compact, while Fig. 4B shows thepost-dissolution hyperspectral image for the same TPa compactrecorded after the duration of one dissolution experiment(15 min) using water as dissolution medium. The color changebetween Fig. 4A and B is due to the mC(8)–H peak shift in CARSspectra, indicating that the TP on the surface has converted toTPm form.

The CARS spectra were collected before and after each dissolu-tion experiment for comparison with the reference spectra (Fig. 3)

Fig. 4. Hyperspectral CARS image recorded between 3050 and 3150 cm�1 of aTPa compact before (A) and after (B) dissolution where water was used asdissolution medium. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

and to confirm the solid-state conversion observed in the dissolu-tion images. Fig. 5 shows the pre-dissolution (black line) and post-dissolution (red dashed line) CARS spectra for a TPa compact afterdissolution using water as the dissolution medium. The CARS spec-tra confirm the observed shift in the peak from around 3120 cm�1

(before) to 3105 cm�1 (after), indicating the conversion from TPa toTPm on the surface of the compact.

3.2. In situ analysis of TP during dissolution testing

3.2.1. Theophylline in water dissolutionSingle-frequency CARS images (512 � 512 pixels) were re-

corded at 2952 cm�1 approximately every second for the durationof the dissolution experiments (15 min). Fig. 6 shows snapshots ofthe dissolution imaging from dissolution conducted using water asdissolution medium. From Fig. 6, it is apparent that the TPm nucle-ation and crystal growth begin almost immediately after the begin-ning of the dissolution experiment with TPm crystals (needleshape) growing outwards from two nuclei on the surface of thecompact. After 2 min, about half of the field of view is covered inTPm crystals, and it appears that most of the field of view is com-pletely covered in TPm after about 4 min.

Ultraviolet spectra were collected every 30 s during the dissolu-tion experiment to determine the dissolution rate profiles for TPaand TPm in our channel flow cell system. Fig. 7 shows the dissolu-tion profiles for TPa (solid lines) and TPm compacts (dashed lines).From Fig. 7, it can be seen that TPa initially increases to peak valuesof between 150 and 190 lg/mL, while the TPm reaches concentra-tions of between 70 and 80 lg/mL. Subsequently, there is a sharpdrop in the first few minutes of the TPa dissolution that is not seenfor the TPm dissolution. This change in dissolution behavior is dueto a solvent-mediated transformation wherein the dissolving TPa(solubility 12 mg/mL at 25 �C [29]) reaches supersaturation whichcauses precipitation and growth of the more stable but less solubleTPm (solubility 6 mg/mL at 25 �C [29]) crystals that grow on thesurface of the TPa compacts during dissolution. The surface growthof TPm on TPa samples undergoing dissolution has also been ob-served in other studies, using offline XRPD analysis [17] and inlinespontaneous Raman spectroscopy [10,30].

The UV data shown in Fig. 7 correlate well with the CARSimages (Fig. 6) that were recorded during the dissolution

Fig. 6. Single-frequency CARS images (2952 cm�1) of TPm crystals growing on the surface of a TPa compact undergoing dissolution using water as the dissolution medium.Still images from a dissolution video show progressive crystal growth. Video is available in Supplementary material. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

Fig. 7. Channel flow through cell dissolution profiles for TPa (solid lines) and TPm(dashed lines) compacts using water as dissolution medium. (For interpretation ofthe references to color in this figure legend, the reader is referred to the web versionof this article.)

A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147 1145

experiments. The dissolution rate peaked after about 2 min whichrelated to about half of the microscope field of view covered in TPmneedle-shaped crystals. After about 5 min, the dissolution ratereached a plateau at the same time the crystal growth appearedto completely cover the field of view. Fig. 7 shows that the TPmdissolution rate quickly reached a steady state after around1 min and remained there for the duration of the experiment.The steady-state dissolution rates were calculated to be360 ± 37 lg/min/cm2 and 320 ± 12 lg/min/cm2 for the compactsprepared from TPa and TPm, respectively. The slightly higher disso-lution rate (not statistically significant) for the compacts originallycomposed of TPa after surface conversion to TPm can be attributedto the TPm needle growth resulting in a larger surface area.

3.2.2. Theophylline in methyl cellulose solution dissolutionIn situ CARS dissolution imaging identified delayed TPm crystal

growth on the surface of TPa compacts undergoing dissolutionusing a MC solution (0.45% w/v) as the dissolution medium.Fig. 8 shows in situ single-frequency CARS snapshots taken froma dissolution video. The TPm crystal growth was delayed as itwas first observed after approximately 300 s (5 min), and the sur-face coverage with TPm was incomplete after the duration of theexperiment (15 min). Additionally, the TPm crystals were of a dif-ferent morphology than previously seen when using water as thedissolution medium. Instead of the thin needle-like structure seengrowing in water, there was a broad almost sheet-like growthalong the surface of the compact.

The delayed onset of crystal growth and different morphologiessuggests that the polymer affects both nucleation and crystalgrowth. It has been suggested that the polymer adsorbs to the sur-face of the TPa particles, and this leads to inhibition of heteroge-neous nucleation, while the changed crystal morphology is likelyto be due to preferential adsorption to specific crystal faces ofTPm, which, in turn, affects their relative growth rates [12].

The delayed TPm crystal growth seen in the CARS dissolutionimaging (Fig. 8) was expected to affect the TPa dissolution rateand Fig. 9 shows that this was the case. Fig. 9 shows the dissolutionprofiles for TPa and TPm compacts undergoing dissolution usingMC solution as the dissolution medium. From Fig. 9 it can be seenthat the characteristic decrease in dissolution rate associated withTPm growth on the surface of TPa compacts (Fig. 7) is no longerseen. Instead the TPa compacts reach a concentration of about150 lg/mL and remain there for the duration of the experiment.The dissolution behavior of the TPm compacts appear minimallyaffected by the use of the MC dissolution medium as they reacha concentration of about 80 lg/mL and remain there for the dura-tion of the experiment. This concentration is the same as was ob-served for water without the polymer, revealing that thesolubility of the drug is not affected by the polymer in solution,and therefore the different dissolution profiles obtained with andwithout polymer solution are not solubility mediated.

Fig. 8. Single-frequency CARS images (2952 cm�1) of TPm crystals growing on the surface of a TPa compact undergoing dissolution using MC solution (0.45% w/v) asdissolution medium. Still images from a dissolution video show delayed crystal growth and different crystal morphologies. Video is available in Supplementary material. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 9. Channel flow through cell dissolution profiles for TPa (solid lines) and TPm(dashed lines) using MC solution as dissolution medium. (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version ofthis article.)

1146 A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147

The steady-state intrinsic dissolution rates were calculated tobe 700 ± 130 lg/min/cm2 and 350 ± 40 lg/min/cm2 for the com-pacts prepared from TPa and TPm respectively (assuming bothcompacts had a perfectly planar surface). Since the solubility ofthe TPa is twice that of TPm in water at 25 �C [29], a two-fold in-crease in the dissolution rate of the compact prepared from TPawould theoretically only be expected if there were no conversionto the monohydrate. However, an increase in surface area of thecompacts prepared from TPa after TPm formation was observedin water which affected the dissolution behavior, and therefore asurface area increase can also be expected to affect the dissolutionprofiles in the polymer solution. Additionally, from Fig. 9 there arenoticeable fluctuations in the steady-state concentrations (when

compared to Fig. 7) of both TPa and TPm this is attributed to bub-bles in the dissolution medium not removed by sonication.

The inherent confocality, chemical specificity, and speedprovided by CARS microscopy increases the spatial and chemicalresolution of the system providing advantages over existingapproaches including traditional optical microscopy and Ramanmicroscopy based on spontaneous Raman scattering. The biggestadvantage when compared to traditional optical microscopy isthe fact that the detected signal is generated when the excitationbeams match the Raman vibrational mode for the chemical ofinterest this provides chemical selectivity. If the excitation laserfrequencies are incorrect or the sample is the wrong chemical thenno resonant signal is generated. The inherent confocal naturemeans that the morphological information is resolved to approxi-mately 10 lm in the depth direction, a resolution that is not ob-tained with traditional optical microscopy.

4. Conclusions

We used CARS microscopy to image in situ solid-state conver-sions of samples during dissolution in real time. The combinationof CARS microscopy with flow through UV absorbance spectros-copy allowed us to correlate the visualized polymorphic conver-sion with changes in dissolution rates. Additionally the inhibitionof TPm crystal growth due to the presence of MC was correlatedwith changes in dissolution rate for TPa compacts. HyperspectralCARS microscopy provided a rapid visual technique to confirmthe polymorphic conversion that occurred during dissolution. Thecombination of the rapid analysis and chemical selectivity of CARSand hyperspectral CARS with UV absorption spectroscopy has thepotential to allow improved characterization of solid dosage formsundergoing dissolution. CARS with UV absorption spectroscopy al-lows further in depth analysis on dosage forms exhibiting unex-pected dissolution profiles, including failed dissolution tests.Improved characterization of solid dosage forms undergoing disso-

A. Fussell et al. / European Journal of Pharmaceutics and Biopharmaceutics 85 (2013) 1141–1147 1147

lution will help in the development of formulations where dissolu-tion profiles are especially important. Formulations such as thosecontaining a poorly soluble APIs and controlled release formula-tions, where bioavailability is dissolution- or release-rate limitedwill benefit from improved characterization.

Acknowledgements

AF is supported by the Dutch Technology Foundation STW,which is the applied science division of NWO, and the TechnologyProgram of the Ministry of Economic Affairs (STW OTP 11114). EGis supported by a NWO VICI grant to Professor Jennifer Herek. BASFis acknowledged for the generous donation of theophylline anhy-drate and monohydrate. Colorcon is acknowledged for the gener-ous donation of methyl cellulose. We thank Coherent Inc. for thePaladin laser and APE Berlin GmbH for the Levante Emerald OPO.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ejpb.2013.08.012.

References

[1] R. Hilfiker, Polymorphism in the Pharmaceutical Industry, WILEY-VCH VerlagGmbH, Weinheim, 2006. pp. 414.

[2] K.R. Morris, U.J. Griesser, C.J. Eckhardt, J.G. Stowell, Theoretical approaches tophysical transformations of active pharmaceutical ingredients duringmanufacturing processes, Advanced Drug Delivery Reviews 48 (2001) 91–114.

[3] A.C. Jorgensen, C.J. Strachan, K.H. Pollanen, V. Koradia, F. Tian, J. Rantanen, Aninsight into water of crystallization during processing using vibrationalspectroscopy, Journal of Pharmaceutical Sciences 98 (2009) 3903–3932.

[4] Y. Kobayashi, S. Ito, S. Itai, K. Yamamoto, Physicochemical properties andbioavailability of carbamazepine polymorphs and dihydrate, InternationalJournal of Pharmaceutics 193 (2000) 137–146.

[5] D. Murphy, F. Rodríguez-Cintrón, B. Langevin, R.C. Kelly, N. Rodríguez-Hornedo, Solution-mediated phase transformation of anhydrous to dihydratecarbamazepine and the effect of lattice disorder, International Journal ofPharmaceutics 246 (2002) 121–134.

[6] F. Tian, N. Sandler, K.C. Gordon, C.M. McGoverin, A. Reay, C.J. Strachan, D.J.Saville, T. Rades, Visualizing the conversion of carbamazepine in aqueoussuspension with and without the presence of excipients: a single crystal studyusing SEM and Raman microscopy, European Journal of Pharmaceutics andBiopharmaceutics 64 (2006) 326–335.

[7] R.J. Davey, P.T. Cardew, D. McEwan, D.E. Sadler, Rate controlling processes insolvent-mediated phase transformations, Journal of Crystal Growth 79 (1986)648–653.

[8] H. Wikström, J. Rantanen, A.D. Gift, L.S. Taylor, Toward an understanding of thefactors influencing anhydrate-to-hydrate transformation kinetics in aqueousenvironments, Crystal Growth & Design 8 (2008) 2684–2693.

[9] M. Savolainen, K. Kogermann, A. Heinz, J. Aaltonen, L. Peltonen, C. Strachan, J.Yliruusi, Better understanding of dissolution behaviour of amorphous drugs byin situ solid-state analysis using Raman spectroscopy, European Journal ofPharmaceutics and Biopharmaceutics 71 (2009) 71–79.

[10] J. Aaltonen, P. Heinänen, L. Peltonen, H. Kortejärvi, V.P. Tanninen, L.Christiansen, J. Hirvonen, J. Yliruusi, J. Rantanen, In situ measurement ofsolvent-mediated phase transformations during dissolution testing, Journal ofPharmaceutical Sciences 95 (2006) 2730–2737.

[11] I. Katzhendler, R. Azoury, M. Friedman, Crystalline properties ofcarbamazepine in sustained release hydrophilic matrix tablets based on

hydroxypropyl methylcellulose, Journal of Controlled Release 54 (1998) 69–85.

[12] H. Wikstrom, W.J. Carroll, L.S. Taylor, Manipulating theophylline monohydrateformation during high-shear wet granulation through improvedunderstanding of the role of pharmaceutical excipients, PharmaceuticalResearch 25 (2008) 923–935.

[13] M. Ku, Use of the biopharmaceutical classification system in early drugdevelopment, The AAPS Journal 10 (2008) 208–212.

[14] United States Pharmacopeial Convention, The United States Pharmacopeia, in,Rockville, 2009.

[15] G. Reich, Near-infrared spectroscopy and imaging: basic principles andpharmaceutical applications, Advanced Drug Delivery Reviews 57 (2005)1109–1143.

[16] C.A. Coutts-Lendon, N.A. Wright, E.V. Mieso, J.L. Koenig, The use of FT-IRimaging as an analytical tool for the characterization of drug delivery systems,Journal of Controlled Release 93 (2003) 223–248.

[17] S. Debnath, P. Predecki, R. Suryanarayanan, Use of glancing angle X-raypowder diffractometry to depth-profile phase transformations duringdissolution of indomethacin and theophylline tablets, PharmaceuticalResearch 21 (2004) 149–159.

[18] C.L. Evans, X.S. Xie, Coherent anti-Stokes Raman scattering microscopy:chemical imaging for biology and medicine, Annual Review of AnalyticalChemistry (Palo Alto Calif) 1 (2008) 883–909.

[19] M.N. Slipchenko, H. Chen, D.R. Ely, Y. Jung, M.T. Carvajal, J.-X. Cheng,Vibrational imaging of tablets by epi-detected stimulated Raman scatteringmicroscopy, Analyst 135 (2010) 2613–2619.

[20] S.H. Parekh, Y.J. Lee, K.A. Aamer, M.T. Cicerone, Label-free cellular imaging bybroadband coherent anti-Stokes Raman scattering microscopy, BiophysicalJournal 99 (2010) 2695–2704.

[21] E. Kang, H. Wang, I.K. Kwon, J. Robinson, K. Park, J.-X. Cheng, In situvisualization of paclitaxel distribution and release by coherent anti-StokesRaman scattering microscopy, Analytical Chemistry 78 (2006) 8036–8043.

[22] E. Kang, J. Robinson, K. Park, J.-X. Cheng, Paclitaxel distribution inpoly(ethylene glycol)/poly(lactide-co-glycolic acid) blends and its releasevisualized by coherent anti-Stokes Raman scattering microscopy, Journal ofControlled Release 122 (2007) 261–268.

[23] E. Kang, H. Wang, I.K. Kwon, Y.-H. Song, K. Kamath, K.M. Miller, J. Barry, J.-X.Cheng, K. Park, Application of coherent anti-Stokes Raman scatteringmicroscopy to image the changes in a paclitaxel–poly(styrene-b-isobutylene-b-styrene) matrix pre- and post-drug elution, Journal ofBiomedical Materials Research A 87A (2008) 913–920.

[24] M. Windbergs, M. Jurna, H.L. Offerhaus, J.L. Herek, P. Kleinebudde, C.J.Strachan, Chemical imaging of oral solid dosage forms and changes upondissolution using coherent anti-Stokes Raman scattering microscopy,Analytical Chemistry 81 (2009) 2085–2091.

[25] M. Jurna, M. Windbergs, C.J. Strachan, L. Hartsuiker, C. Otto, P. kleinebudde, J.L.Herek, H.L. Offerhaus, Coherent anti-Stokes Raman scattering microscopy tomonitor drug dissolution in different oral pharmaceutical tablets, Journal ofInnovative Optical Health Sciences 2 (2009) 37–43.

[26] E.T. Garbacik, J.L. Herek, C. Otto, H.L. Offerhaus, Rapid identification ofheterogeneous mixture components with hyperspectral coherent anti-StokesRaman scattering imaging, Journal of Raman Spectroscopy 43 (2012) 651–655.

[27] M.M. Nolasco, A.M. Amado, P.J.A. Ribeiro-Claro, Computationally-assistedapproach to the vibrational spectra of molecular crystals: study of hydrogen-bonding and pseudo-polymorphism, ChemPhysChem 7 (2006) 2150–2161.

[28] P.J.A. Ribeiro-Claro, P.D. Vaz, Towards the understanding of the spectroscopicbehaviour of the C–H oscillator in C–H� � �O hydrogen bonds: the effect ofsolvent polarity, Chemical Physics Letters 390 (2004) 358–361.

[29] N. Rodríguez-Hornedo, D. Lechuga-Ballesteros, W. Hsiu-Jean, Phase transitionand heterogeneous/epitaxial nucleation of hydrated and anhydroustheophylline crystals, International Journal of Pharmaceutics 85 (1992) 149–162.

[30] P. Lehto, J. Aaltonen, P. Niemelä, J. Rantanen, J. Hirvonen, V.P. Tanninen, L.Peltonen, Simultaneous measurement of liquid-phase and solid-phasetransformation kinetics in rotating disc and channel flow cell dissolutiondevices, International Journal of Pharmaceutics 363 (2008) 66–72.