Perspective Direct qualitative and quantitative characterization of a radiosensitizer, 5-iodo-2A-deoxyuridine within biodegradable polymeric microspheres by FT-Raman spectroscopy Annabelle Geze,a Igor Chourpa,*b Franck Boury,a Jean-Pierre Benoita and Pierre Duboisb a UPRES EA 2169, Facult�e de Pharmacie, Universit�e d’Angers, 16 Boulevard Daviers, 49100 Angers, France b Laboratoire de Chimie Analytique, Facult�e de Pharmacie ‘Philippe Maupas’, Universit�e de Tours, 31 Avenue Monge, 37200 Tours, France Received 25th August 1998, Accepted 29th October 1998 Non-destructive qualitative and quantitative characterization of a radiosensitizer, 5-iodo-2A-deoxyuridine (IdUrd), incorporated within injectable microspheres of a biodegradable polymer, poly(d,l-lactide-co-glycolide) (PLGA), was performed using Fourier transform (FT) Raman spectroscopy.Raman spectra of IdUrd, free and entrapped in microspheres, were recorded under fluorescence-free conditions, described and assigned.For the Raman bands of the PLGA microspheres, assignments with preferential localization of the corresponding vibrations at lactic or glycolic units were proposed. No evidence for drug–polymer interactions in microspheres was found. This allowed the FT-Raman spectra to be used for the quantification of the IdUrd content in the samples. For the microspheres with IdUrd loadings varying from 2 to 27% of the total weight, the methodology used provided good reproducibility and precision (1%).Within the sensitivity of the technique, samples exposed to sterilization doses (27 kGy) of g-radiation did not exhibit marked changes in the drug structure. Nowadays, Raman spectroscopy has become an easily used technique with a very wide range of applications. For the resolution of numerous analytical problems, the molecularspecific information obtained in a non-destructive way seems to be irreplaceable. Moreover, Raman spectroscopic data can be successfully used for quantitative measurements.Owing to some particular advantages, such as fluorescence-free conditions, Fourier transform (FT) Raman spectroscopy with excitation in the infrared region is especially useful in analytical studies of raw samples. In this study, we aimed to apply FT-Raman spectroscopy to the non-destructive qualitative and quantitative characterization of a drug incorporated within a polymeric matrix. The polymers concerned are bioresorbable aliphatic polyesters based on copolymers of lactic (LA) and glycolic acid (GA).These biocompatible polymers degrade hydrolytically in the body with the formation of non-toxic products.1 Poly(d,l-lactide-coglycolide) (PLGA) polymers have been studied as materials for osteosynthesis, sutures and prosthetic devices.1 They are widely used for therapeutic purposes especially to make sustained drug release delivery systems.2 We focused on PLGA-based delivery systems as used for the sustained release of 5-iodo-2A-deoxyuridine (IdUrd).This molecule is a thymidine analog and is a powerful radiosensitizer. 3 This halogenated pyrimidine competes with thymidine in the biosynthesis of DNA. The treatment of malignant brain tumors based on conventional radiotherapy is far from satisfactory. Therapeutic IdUrd concentrations within the tumor during the time course of radiotherapy might improve treatment results, by increasing the lethal effects of radiation on the tumor cells having incorporated the radiosensitizer.The intracranial implantation of IdUrd-loaded microparticles in the vicinity of the cancer cells can meet these requirements.4 The study involved the preparation of PLGA microspheres loaded with different amounts of IdUrd, by using a phase separation technique. Intact IdUrd-loaded microspheres were characterized qualitatively and quantitatively by FT-Raman spectroscopy. Since the microspheres are intended to be administered in vivo into the brain, they need to be sterile. The sterilization of biodegradable drug delivery systems is often carried out by g-irradiation.The stability of the drug entrapped in the microspheres5,6 after exposure to g-radiation was determined from the respective Raman spectra. Experimental Chemicals IdUrd (99% pure, odorless, white, crystalline powder, slightly soluble in water: 2 mg ml21) was obtained from Sigma-Aldrich Chimie (St. Quentin Fallavier, France). PLGA 50/50 was purchased from Boehringer Ingelheim (RG 506, B.I.Chimie, Paris, France). The composition of the chains includes 25% llactic units, 25% d-lactic units and 50% glycolic units. The mass- and number-average molecular masses were 75 000 and 48 000, respectively. These values were determined in tetrahydrofuran by size exclusion chromatography (SEC Waters, St. Quentin en Yvelines, France), referred to polystyrene standards. Methylene chloride, silicone oil (Rhodorsil, viscosity 300 cSt, relative density 0.97) was obtained from Prolabo (Paris, France).Heptane was purchased from Verbi`ese (Wasquehal, France) and dimethyl sulfoxide (DMSO) from Carlo Erba (Val de Reuil, France). IdUrd crystal milling Grinding of IdUrd crystals was performed with a Pulverisette 7 planetary micro-mill (Fritsch, Idar-Oberstein, Germany). A 800 mg amount of IdUrd was milled for 10 min at a rotation speed of 2500 rpm. Analyst, 1999, 124, 37–42 37Microsphere preparation The coating polymer PLGA (250 mg) was dissolved in methylene chloride to reach a concentration of 1.3% m/m.Various amounts of milled IdUrd crystals (18 ± 3 mm, SD of different mean size values) were then dispersed in the organic phase with sonication for 3 min. A separation phase inducer (silicone oil, 8 g) was added to the mixture and stirred magnetically at room temperature for 2 min in order to precipitate the polymer around the drug particles. The resulting dispersion (semi-formed microparticles or coacervates) was poured into 400 ml of heptane (hardening agent), stirred at 600 rpm (Heidolph RGH500, Prolabo, Paris, France).After 30 min of agitation, the solidified microparticles were filtered on a 0.45 mm filter (HV type, Millipore, Maurepas, France) and washed with heptane (50 ml). The resulting microspheres were dried under reduced pressure for 60 h at 35 °C. They were stored at 6 °C shielded from light. Microsphere size distribution analysis The average size of the microparticles was determined using a Coulter Multisizer (Coultronics, Margency, France) after dispersion of the microparticles in a conducting liquid (Isoton II, Coultronics).Crystal size distribution The average size of the crystals was determined using a Mastersizer S (Malvern Instruments, Malvern, Orsay, France). Size measurements were performed in the liquid phase, using a 300 RF lens, in the size range 0.05–880 mm and an MSI module as a manual liquid sampler.A 30 mg amount of milled powder was suspended in 3 ml of cyclohexane with sonication and immediately analyzed. g-Irradiation of microspheres IdUrd-loaded microspheres were accurately weighed into 100 mg samples, transferred to 2.5 ml glass vials and sealed. The vials were irradiated at a dose of 26.7 kGy 60Co source, (Ionisos, Dagneux, France). This was done in triplicate. IdUrd content determination This was achieved using two methods, spectrophotometry and Raman spectroscopy.The former method provides direct access to the drug concentration via molar absorptivity (7.9 3 103 l mol21 cm21 at 287 nm) but was destructive, since the microspheres (6–8 mg) needed to be dissolved in DMSO. The latter method, allowing non-destructive quantification of IdUrd, was developed using the Raman spectra obtained from intact microspheres. These spectra contained a contribution of the polymer vibrations which could be useful as an internal standard.The IdUrd/PLGA peak area ratio of the Raman bands allowed the calculation of the corresponding ratio of the concentrations of these molecules. This is discussed below. Spectroscopic instrumentation The absorption of IdUrd was measured with a Uvikon 922 UV spectrophotometer (Kontron Instruments, St. Quentin en Yvelines, France). The IdUrd solutions were shielded from light. The Raman spectra were excited with 1.06 mm radiation from an ADLAS Nd:YAG laser and recorded with a Bruker (Wissembourg, France) RFS100/D418-S FT-Raman spectrometer.The laser power at the sample was about 140 mW. No laser-induced sample degradation was noted during the experiments. All the measurements were repeated at least three times. The data appeared to be well reproducible. The data obtained were analyzed with Labspec software (DILOR, Lille, France), which permitted the very easy and rapid treatment of the spectra (baseline correction, peak area analysis, normalization, etc.) and statistical analysis of any spectral parameters.In addition, this software package allowed us to treat simultaneously and in exactly the same manner all the sets of the recorded spectra. Results and discussion In order to analyze the Raman spectra of the drug-containing microspheres, the model Raman spectra of the blank PLGA 50/50 microspheres and of IdUrd crystals (Fig. 2) were recorded. FT-Raman spectra of IdUrd crystals In the FT-Raman spectra of IdUrd crystals (Fig. 2 and Table 1), both the bands of iodo-substituted aromatic nucleus (5-iodouracil) and those of the deoxyribose moiety (Fig. 1) can be observed. As a result, the spectra appeared very rich in vibrations, especially within the 1400–1800 cm21 region. Since the Raman spectra of IdUrd have not been reported previously, we describe them in detail. Our discussion is limited to only the more intense bands between 1800 and 700 cm21. The CH region bands above 2800 cm21 and the lower wavenumber weak deformational bands were less interesting with respect to the main task of the present study. Fig. 1 Structural formulae of IdUrd and PLGA. Fig. 2 FT-Raman spectra of IdUrd crystals (bottom) and the blank PLGA microspheres (top). 38 Analyst, 1999, 124, 37–42As expected for the IdUrd molecule, the very characteristic stretching bands of the non-conjugated and conjugated C =O groups were clearly observed at 1696 and 1676 cm21, respectively (Fig. 2 and Table 1).In their neighborhood, at 1611 cm21, one could observe another strong band, that of the n(C = C) vibration. The region of the CH2 deformation motions was represented by a weak doublet at 1460/1444 cm21 encircled with two even weaker bands which are not discussed here (see Table 1). A strong band at 1350 cm21 had the characteristic wavenumber of H–N–C = O stretching (amide III). A group of weak bands located at lower wavenumbers, down to 1230 cm21, was due to CH deformation and CH2 twisting.The bands within the 1200–1140 and 1105–1020 cm21 regions were assigned to asymmetric and symmetric COC stretching, respectively. The vibrations observed between 1010 and 850 cm21 were attributed to the C–C stretching and CH2 rocking. The very strong band at 779 cm21 was assigned to a ring breathing mode of the 5-iodouracil moiety. The Raman bands of IdUrd located near 750 cm21 (dC = O) and lower were those of the various deformational motions of the C = O, CCO, etc., groups.The FT-Raman spectra of IdUrd after mechanical milling (see Experimental) were devoid of any detectable changes (data not shown). Based on this and on the data from X-ray diffraction (not shown), it was concluded that no polymorphic forms were present in detectable amounts in these samples. Therefore, the crystallinity of IdUrd was preserved after milling. FT-Raman spectra of the blank PLGA 50/50 microspheres FT-Raman spectra of blank microspheres (Fig. 2 and Table 2) were analyzed in comparison with previously reported Raman spectra (in the visible region) of poly(d,l-lactide) (PLA)7 and poly(glycolide) (PGA)8 polymers. The band positions and shapes in the spectra of PLGA microspheres indicated an amorphous form of polymer.9,10 In general, the PLGA spectra contained both the PLA and PGA Raman bands, the former being more pronounced (see Table 2). With respect to this comparison, we proposed a temptative attribution of the observed PLGA bands to vibrational modes of lactic (LA) or/and glycolic (GA) units (Table 2).Whereas both PLA and PGA spectra have already been well documented,7-10 to our knowledge, there has been no report on the Raman spectra of PLGA. For this reason, only some particular features in the Raman patterns which differentiate the PLGA samples from PLA and PGA (Table 2), are discussed. Table 1 Major Raman wavenumers (cm21) and their tentative assignments for IdUrd in pure crystals and when in PLGA microspheresa Microspheres with IdUrd Blank microspheres Assignment Crystals of IdUrd Assignment 3080 w 3080 m nasCH 3010 w 3010 m nsCH 3002 3002 s nasCH3 2965 sh 2968 vs nasCH2 2953 s 2956 s nasCH2 2947 vs 2947 vs nsCH3 + nasCH2 2936 m 2936 m nasCH2 2908 sh 2908 m nCH 2876 s 2876 s nCH 1769 s 1769 s nC =O ~ 1760 sh ~ 1760 sh nC =O 1696 m 1696 m nC = O, non conj. 1676 s 1676 s nC = O, conj. 1611 s 1611 s nC = C 1458 s 1452 s dasCH3 + dCH2 1460 mw dCH2 1442 sh 1445 w dCH2 1427 m 1427 m dCH2 1390 w 1384 w dsCH3 1395 w dCH2 1350 s 1345 vw d1CH + dsCH3 1350 s Amide III + dCH 1300 w 1303 w d2CH 1296 w dCH 1274 vw twCH2 1268 w 1267 mw twCH2 1252 vw twCH2 1239 w 1239 mw twCH2 1199 mw 1198 m nasCOC 1145 w 1146 w nasCOC 1130 m 1130 m rasCH3 1102 vw nsCOC 1093 w 1093 w nsCOC 1096 w nsCOC 1075 vw nsCOC 1055 vw nsCOC ~ 1048 sh nC–CH3 1032 w 1032 w nsCOC 1028 w nsCOC 990 w 989 w nC–C + rCH2 961 w 961 w nC–C + rCH2 953 vw nC–C + rCH2 + rCH3 920 vw nC–C + rCH2 913 w nC–C + rCH2 892 m nC–C + rCH2 885 m 886 m nC–C + rCH2 874 s 874 s nC–COO 878 m nC–C + rCH2 849 sh 847 mw rCH2 855 mw rCH2 779 vs 779 vs Ring breathing 746 w 750 vw dC =O 756, 749 w dC =O 706 vw 706 vw gC = O + dC =O a Abbreviations: v = very; s = strong; m = medium; w = weak; sh = shoulder; subscript s = symmetric; subscript as = asymmetric.Analyst, 1999, 124, 37–42 39The wavenumbers that appeared significantly shifted in PLGA are given in italics in Table 2.This particularly concerned the stretching (3002 and 2947 cm21) modes of the CH3 (LA) and CH2 (GA) groups (Table 2). The interesting observation was that the copolymer formation seemed to affect the dCH2 (1427 cm21) more than the dCH3 (1384 cm21) wavenumbers and the d1CH (1345 cm21) more than the d2CH (1303 cm21) wavenumbers. The perturbations of the nC–C and rCH2 modes (both 920 and 892 cm21) were less predictable than those for dC = O (750 and 706 cm21). FT-Raman spectra of the PLGA microspheres loaded with IdUrd These logically contained both drug and polymer bands (Fig. 3 and Table 1). We subtracted from these spectra the model spectra of the pure drug and blank microspheres (Fig. 4). The difference spectra, even obtained from those with a significant contribution of the polymer or drug to the Raman bands, did not reveal any noticeable changes as compared with their respective model Raman patterns. Therefore, no indication either of structural modification of IdUrd and/or PLGA matrix or of their interactions could be discerned.It was also noted that unchanged spectra (in wavenumbers and relative intensities) could be considered as the sum of the model spectra. This allowed the quantitative considerations discussed in the following section. Determination of Raman IdUrd content in the IdUrd-loaded PLGA microspheres The methodology used with incorporation of a phase separation technique yielded PLGA microspheres of nearly regular size in the range 40 ± 5 mm (SD on mean size of different microsphere batches) as established by particle sizing.The first step in the current work was to establish the calibration curve, i.e. the function describing the Raman spectral parameters as a function of the drug/polymer relative concentrations. For this purpose, microspheres with different IdUrd loadings, ranging from 2 to 27% [IdUrd (mg)/microspheres (mg) 3 100: incorporation ratio, determined using spectrophotometry] were analysed.Comparison of the Raman spectra of the drug-loaded microspheres with the spectra of the pure IdUrd crystals and blank PLGA microspheres (Fig. 3) revealed several regions devoid of superposed bands of these molecules. In particular, in the region between 1600 and 1820 cm21, the drug and polymer vibrations were in a close neighborhood but well separated. This made these vibrations usable as an internal standard. Three strong Raman bands of IdUrd, at 1696, 1676 and 1611 cm21, and a very large band of PLGA with a maximum at about Table 2 Major Raman wavenumbers (cm21) and their tentative assignments for PLGA 50/50 microspheres compared with PLA and PGA polymersa Raman (lex 514.5 nm)b,c FT-Raman (lex 1064 nm): PLA amorphousb PGA amorphousc PLGA50/50 microspheres Localizationd Assignment 2997 s 3002 s LA nasCH3 2942 vs 2947 vs LA + GA nsCH3 + nasCH2 2954 vs 2877 m 2876 s LA nCH 1769 s 1769 s LA nC =O 1760 s ~ 1760 sh GA nC =O 1455 s 1450 sh 1452 s LA + GA dasCH3 + dCH2 1423 s 1427 GA dCH2 1400 sh — wCH2 1386 m 1384 w LA dsCH3 1365, 1355 m 1345 w LA d1CH + dsCH3 1296, 1300 s 1303 w LA d2CH 1274 m 1274 w GA twCH2 1128 s 1130 m LA rasCH3 1092 s 1090 w 1093 w LA + GA nsCOC 1042 s ~ 1048 LA nC–CH3 1029 m 1032 GA nsCOC 953 sh 950 m 953, 920 LA + GA rCH3 + nC–C + nC–C + rCH2 885 s 892 m GA nC–C + rCH2 873 vs 874 s LA nC–COO 845 s 847 w GA rCH2 740 m 750 vw LA dC =O 700 vw 720 w 706 vw LA + GA gC = O + dC =O a Abbreviations: b As referred in ref. 7. c As referred in ref. 8. d With respect to the d, l lactic (LA) or glycolic (GA) units. v = very strong; s = strong; m = medium; w = weak; sh = shoulder; subscript s = symmetric; subscript as = asymmetric. Values in italics are wavenumbers shifted for PLGA compared with those of PLA and/or PGA. Fig. 3 Low-wavenumber region of the FT-Raman spectra: (a) IdUrd crystals; (d) blank PLGA microspheres; (b) and (c) PLGA microspheres containing different amounts of the drug. To emphasize the comparison, the spectra have been normalized using nC = O bands (see Table 1) of the drug [(a) and (b)] or of the polymer [(c) and (d)]. 40 Analyst, 1999, 124, 37–421769 cm21 were observed (for assignments, see Tables 2 and 3).Therefore, this spectral region was selected for quantitative measurements. Hereafter, the peak area ratio of the 1586–1723/1723–1815 cm21 spectral regions, representing the drug/polymer Raman spectral ratio, Rd/Rp, was considered.Rd/Rp plotted as a function of the respective incorporation ratio (%) is presented in Fig. 5. These experimental points fitted well a second-order polynomic curve (Table 3). In the range of incorporation ratios used for clinical applications (16–23%) the calibration curve was quasi-linear. This could be used to simplify the quantitative calculations for the commonly used incorporation ratios. The largest deviations of the observed Rd/Rp value from the fitted curve gave higher deviations of IdUrd incorporation within ±1% (see the shaded area in Fig. 5). Therefore, the reproducibility of the data was good and the approach allowed the determination of the IdUrd content in ‘non-calibrated’ microspheres from their Raman spectra with a precision of at least 1%. Effects of exposure to g-rays on IdUrd-loaded PLGA microspheres Two sets of microspheres, loaded with 20 and 27% of IdUrd, were exposed to 27 kGy of ionizing radiation. A dose of 25 kGy was considered to be the minimum necessary for sterilization specified by the European Pharmacopoeia.Strong ionizing radiation is known to be destructive for the polymeric matrix. Several reports have commented on g-ray irradiation-induced chain scission and cross-linking11-13 in polylactide and polyglycolide. However, concerning the active agent, the influence of irradiation on the drug structure remained to be defined. This point was essential since microspheres must be radiosterilized before implantation. In the present study, we focused on the irradiation effect on the sole IdUrd structure.As followed from both spectrophotometric and Raman data, the irradiated PLGA microspheres exhibited nearly the same (±1%) IdUrd content as before the irradiation process. The FTRaman bands of the irradiated IdUrd crystals, free or when incorporated within the microspheres, also remained unchanged, both in wavenumers and intensities (Fig. 6). This was also supported by the infrared absorption spectra (data not shown) of the same samples.Therefore, no drug degradation was found. This was a promising observation with respect to clinical application. Conclusion We interpreted each of the FT-Raman spectra of IdUrd-loaded PLGA microspheres in both qualitative and quantitative ways. Fig. 4 (a) Model FT-Raman spectrum of IdUrd crystals; (b) and (c) difference spectra obtained by subtraction of the polymer model spectrum from those of the microspheres loaded with (b) 27% and (c) 4% of IdUrd; (d) model FT-Raman spectrum of the blank PLGA microspheres; (e) and (f) difference spectra obtained by subtraction of the drug model spectrum from those of the microspheres loaded with (e) 4% and (f) 27% of IdUrd.The difference spectra have been normalized for comparison with corresponding model spectra. Table 3 Results of fitting of the Rd/Rp data to a second-order curve: y = a + bx + cx2 Parameter Value Standard error 95% confidence interval A (fixed constant) 0.0 B 0.028141 0.003231 0.0202343– 0.0360485 C 0.001818 0.000141 0.00147298– 0.00216377 Degrees of freedom 6 r2 0.998181 Absolute sum of squares 0.007489 Syx 0.0353 No.of x values 10 No. of y replicates (mean analyzed) 5 Total no. of values 8 No. of missing values 42 Fig. 5 Drug/polymer Raman peak area ratio, Rd/Rp, versus IdUrd incorporation ratio (%). Fig. 6 Comparison of the FT-Raman pattern of IdUrd in PLGA microspheres (a) before and (b) after irradiation with 27 kGy of g-rays.These are difference spectra obtained by subtraction of the model spectra of the blank microspheres from the spectra of the microspheres loaded with 20% of IdUrd. Analyst, 1999, 124, 37–42 41The simultaneous access to both qualitative and quantitative information illustrated the major advantages of the proposed analytical approach. The qualitative information included the detailed analysis of the spectral shape, i.e., band position and relative intensity.This allowed the molecular characterization of the samples and therefore detection of probable structural changes of the drug and/or polymer matrix induced by interaction, degradation, etc. The quantitative analysis included the peak area ratio calculation for the Raman bands. The analytical method presented several important advantages. First, the evaluation of the drug content was nondestructive and the assayed microspheres could be used for further in vitro investigations. Second, this information can be obtained rapidly: about 1 min is necessary to record a Raman spectrum and the computer-assisted mathematical treatment can be almost instantaneous.Compared with more conventional methods, including spectrophotometric measurements, the Raman approach brought an additional facility and economy of time due to elimination of extended sample preparation (source of errors): for instance, no accurate weighing operation was required. Compared with certain reported quantitative Raman measurements of drugs within polymeric implants,14 the advantage of the proposed method was the use of the band peak area instead of intensity and of relative instead of absolute values of the spectral parameters, which eliminated numerous experimental errors and complicated data manipulations.The use of the Raman band of the polymer as an internal intensity standard made the measurements completely independent of the overall intensity of the Raman spectra.Hence control of the laser power, focusing and other instrumental conditions was unnecessary. The methodology therefore had a more general analytical character, i.e., not related to particular experimental conditions (instrumental parameters, polymer content, etc.). Finally, the proposed approach provided reproducible results. The precision of 1% obtained in determining the IdUrd incorporation ratio makes it possible to study the in vitro drug release kinetics by assaying, by Raman spectroscopy, the remaining drug amounts in the microspheres as a function of time. References 1 S. Li and M. Vert, in Degradable Polymers, ed. G. Scott and D. Gilead, Chapman and Hall, London, 1995, ch. 4, pp. 43–87. 2 J. P. Benoit, H. Marchais, H. Rolland and V. Vande Velde, in Microencapsulation. Methods and Applications, ed. S. Benita, Marcel Dekker, New York, 1996, ch. 3, pp. 35-71. 3 B. Djordjevic and W. Szylbalski, J. Exp. Med., 1960, 112, 509. 4 P. Menei, J. P. Benoit, M. Boisdron-Celle, D. Fournier, P. Mercier and G. Guy, Neurosurgery, 1994, 34, 1058. 5 A. G. Hausberger, R. A. Kenley and P. Deluca, Pharm. Res., 1995, 12, 851. 6 G. Spenlehauer, M. Vert, J. P. Benoit, F. Chabot and M. Veillard, J. Controlled Release, 1988, 7, 217. 7 G. Kiester, G. Cassanas, M. Vert, B. Pauvert and A. T�erol, J. Raman Spectrosc., 1995, 26, 307. 8 G. Kiester, G. Cassanas and M. Vert, Spectrochim. Acta, Part A, 1997, 53, 1399. 9 G. Cassanas, M. Morssli, F. Fabr`egue and L. Bardet, J. Raman Spectrosc., 1991, 22, 11. 10 G. Kiester, G. Cassanas, F. Fabr`egue and L. Bardet, Eur. Polym. J., 1992, 28, 1273. 11 K. M�ader, A. Domb and H. M. Swartz, Appl. Radiat. Isot., 1996, 47, 1669. 12 A. G. Hausberger, R. A. Kenley and P. P. DeLuca, Pharm. Res., 1995, 12, 851. 13 M. C. Gupta and V. G. Deshmukh, Polymer, 1983, 24, 827. 14 P. Milne, S. Gautier, J. M. Parel and V. Jallet, Proc. SPIE, 1997, 2971, 137. Paper 8/06678G 42 Analyst, 1999, 124, 37&ndas