Cellulose ethers are an important class of cellulose derivatives

Transcription

1 694 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 DRUGS, COSMETICS, FORENSIC SCIENCES Alternative Liquid Chromatographic Method for Determination of the Methoxyl and 2-Hydroxypropoxyl Content in Cellulose Ether Derivatives JOE RASHAN,RAYMOND CHEN, 1 TODD ZELESKY, and SONJA SEKULIC Groton Laboratories, Pfizer Global Research and Development, Pharmaceutical Sciences, Analytical R/D, Groton, CT An alternative liquid chromatographic (LC) method was developed and validated for the simultaneous determination of methoxyl and 2-hydroxypropoxyl substituents in hypromellose and hypromellose acetate succinate. The method uses the hydriodic acid cleavage reaction, catalyzed by adipic acid, of the substituted methoxyl and 2-hydroxypropoxyl groups, which are quantitatively converted to iodomethane and 2-iodopropane. The iodomethane and 2-iodopropane are extracted into xylene, the extract is diluted with methanol, and the analytes are separated and assayed by gradient elution using a reversed-phase C 18 column. The method is selective and sensitive and has good linearity with values of for R 2 and 0.25% for the y-intercept bias from 1.39 to 5.55 mg/ml for iodomethane, and for R 2 and 0.52% for the y-intercept bias from to mg/ml for 2-iodopropane. The relative standard precisions for this LC method were found to be 2.3% for determining methoxyl at the 23.1% (w/w) level, and 3.5% for determining 2-hydroxypropoxyl at the 6.7% (w/w) level. Compared with the current gas chromatographic (GC) compendial (JPE) method, the LC assay method has equal or better precision. It was found that both the standard and sample solutions have limited stability (8 h) after preparation. This limited stability has not been reported previously in the literature and may have an impact on the reported accuracy/precision of the literature data for the GC method. The LC method was proven to be robust with respect to variation in derivatization time and temperature, flow rate, and column temperature. It is well suited for the quality control needed in today s fast-paced pharmaceutical laboratories. Received December 13, Accepted by JM March 28, Author to whom correspondence should be addressed; raymond_chen@groton.pfizer.com. Cellulose ethers are an important class of cellulose derivatives made by etherification of cellulose, the most abundant natural polymer on earth. A variety of cellulose ethers are commercially available that are differentiated by the type and degree of alkoxyl group substitution (1). Each cellulose ether has a unique combination of properties. For example, although ethyl cellulose is one of the water-insoluble polymers frequently used in the pharmaceutical industry, methylcellulose and hypromellose [hydroxypropylmethylcellulose (HPMC) Chemical Abstracts Service Registry No ] are mostly used as water-soluble cellulose derivatives. The availability of the wide choice of properties enables extensive application of many cellulose ethers in the food and pharmaceutical industries as emulsifiers, binders, thickeners, and coating agents. Because the type and degree of alkoxyl substitution are important attributes that enable cellulose ether to have such a wide range of properties, accurate measurement of the alkoxyl substitution content is needed for quality control of the polymers used in food and pharmaceutical applications. A number of analytical techniques are available for the determination of alkoxyl substitution in cellulose ethers. Infrared spectroscopy is fast with the least amount of time needed for sample preparation, but it does not provide acceptable accuracy/precision for a quantitative assay (2). The same is true for proton nuclear magnetic resonance spectrometry, despite the fact that it gives the most detailed structural information (3 5). For routine quality control analysis, the most widely used technique is cleavage derivatization of alkoxyl groups followed by either gas chromatography (GC) or titration analysis (6 9). In The United States Pharmacopeia/National Formulary (USP/NF), there are 3 compendial test methods for the analysis of cellulose ether derivatives for alkoxyl substitution. In the hypromellose monograph, methoxyl and 2-hydroxypropoxyl groups are assayed by a modified Zeisel reaction followed by GC (10). In USP-Chemical Tests <431>, the classical Zeisel reaction followed by titration analysis is used for methoxyl determination (11). A different derivatization reaction (chromic acid oxidation) followed by titration analysis is used to assay 2-hydroxypropoxyl substitution in the monograph for hydroxypropyl cellulose (12).

2 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, In this paper, we report a validated analytical method based on the modified Zeisel reaction (7) and analysis by liquid chromatography (LC) for the simultaneous determination of both methoxyl and 2-hydroxypropoxyl substitutions in hypromellose and hypromellose acetate succinate [hydroxypropylmethylcellulose acetate succinate (HPMCAS) Chemical Abstracts Service Registry No ]. The method uses the hydriodic acid cleavage reaction, catalyzed by adipic acid, of the substituted methoxyl and 2-hydroxypropoxyl groups, which are converted quantitatively to their corresponding alkyl iodides iodomethane and 2-iodopropane. The iodomethane and 2-iodopropane formed are extracted into xylene in situ, the extract is diluted with methanol, and the analytes are separated and assayed by gradient elution using a reversed-phase C 18 LC column. LC-based methods are currently the most widely used separation technique in today s fast-paced pharmaceutical quality control laboratories. The modern LC system is very reliable, robust, and capable of achieving high injection precision. Consequently, external standard calibration can be used routinely in today s LC methods. However, an internal standard has to be used in GC methods to compensate for injection variability. In addition, the ability of most of today s analysts to perform LC separation enables easier method transfer to different laboratories. Experimental Chemicals and Solvents (a) Iodomethane (99.5%, Cat. No. 28,956-6), 2-iodopropane (99%, Cat. No. 14,893-8), o-xylene (98%, LC grade, Cat. No. 29,588-4), adipic acid (99+%, Cat. No. 24,052-4), and hydriodic acid (57%, w/w, in water, unstabilized, 99.99%, Cat. No. 21,001-3). Purchased from Aldrich (Milwaukee, WI). (b) LC grade methanol (Cat. No ). Purchased from J.T. Baker (Phillipsburg, NJ). (c) Water. Purified through a Millipore (Bedford, MA) MilliQ system and filtered through a 0.22 m Millipak filter. LC grade water from J.T. Baker was also used in the experiments. (d) Polymers. Various lots of 4 grades of HPMCAS polymer (AQUOAT, LF, MF, HF, and MG) and 1 grade of HPMC polymer were purchased from Shin Etsu Chemical Co. Ltd. (Tokyo, Japan). Three lots of HPMC (Premium LV grades of E3, E5, and E15) were purchased from Dow Chemical (Midland, MI). Preparation of Mobile Phase Mobile Phase A was water methanol ( , v/v), and Mobile Phase B was water methanol ( , v/v). Preparations of Standard Solutions Approximately 2 ml o-xylene was placed in a 10 ml volumetric flask. The flask was capped with a glass stopper and tared on an analytical balance. Approximately 200 L iodomethane was added by using an adjustable micropipet. The flask was capped and weighed again on the balance. The recorded weight of iodomethane should be ca 350 ( 70) mg. The capped flask was tared again, and ca 34 L 2-iodopropane was added by using an adjustable micropipet. The recorded weight of 2-iodopropane should be ca 50 ( 10) mg. The contents of the flask were then brought to volume with o-xylene, and the contents were mixed well by hand. The concentration of the iodomethane was ca 35 mg/ml, and the concentration of the 2-iodopropane was ca 5 mg/ml. Next, mg adipic acid was added to an 8 ml vial. To the vial, 2 ml hydriodic acid was added, and 2 ml of the o-xylene solution prepared above was accurately measured and added. The vial was shaken by hand, and the 2 phases were allowed to separate. Approximately 1.5 ml of the top o-xylene layer was removed by using a disposable glass Pasteur pipet (making sure that the bottom aqueous layer was not disturbed) and placed in a small vial; 1 ml of this solution was accurately measured into a 10 ml volumetric flask and diluted to volume with methanol. The contents were mixed well by hand. A portion of this solution was transferred to an LC vial and labeled Standard 1. The above procedure was repeated with a new weighing of iodomethane and 2-iodopropane, and the solution was labeled Standard 2. Preparation of Sample Solutions A mg portion of test polymer was weighed into a 5 ml Reacti-Vial (Cat. No , Pierce Chemical Co., Rockford, IL). To the same vial, mg adipic acid was added.a2mlportion of hydriodic acid was added to the vial, and2mlo-xylene was accurately measured into the vial. The Table 1. Influence of reaction temperature and time on assay results a Value of variable, C ormin Methoxyl, % (w/w) With temperature change 2-Hydroxypropoxyl, % (w/w) With time change a The polymer used was HPMCAS (MF grade). The experiments were performed and the results were calculated according to the procedure described in the text, except that either the reaction temperature or the time was changed as indicated. The results, in % (w/w), were corrected for the value of loss on drying. The values for methoxyl and 2-hydroxypropoxyl substituents in the manufacturer s certificate of analysis are 23.4 and 7.0%, respectively.

3 696 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 vial was capped with a Mininert valve (Cat. No , Pierce Chemical Co.) and weighed before heating. The vial was then placed on a Reacti-Therm III heating block (Cat. No , Pierce Chemical Co.) at 150 C. The vial was shaken after 5 and 30 min of heating, removed from the block after 1 h of heating, and allowed to cool to ambient temperature. The vial was weighed again. If the weight loss was >10 mg, the mixture was discarded, and another reaction solution was prepared. Otherwise, approximately 1.5 ml of the top o-xylene layer was removed with a disposable glass Pasteur pipet and placed in a small glass vial (making sure that the bottom aqueous layer was not disturbed). Next, 1 ml of the o-xylene layer that was removed was accurately measured into a 10 ml volumetric flask and diluted to volume with methanol. The contents of the flask were mixed well. A portion of this solution was transferred to an LC vial and labeled with sample identification. Each lot of test polymer was assayed in triplicate. Instrumentation and Operating Conditions Two HP1100 LC instruments from Agilent Technologies, Inc. (Palo Alto, CA) were used for all experiments. One LC instrument consisted of a degasser (G1322A), a QuatPump (G1311A), a thermostated autosampler (G1329A and G1330B), a column heater (G1316A), and a variable wavelength UV detector (G1314A). The other LC instrument consisted of the same types of degasser, QuatPump, thermostated autosampler, and column heater, but it had a diode array UV detector (G1315A). An AQUASIL column (5 m, C Å, mm; Cat. No ) from ThermoHypersil-Keystone Scientific, Inc. (Bellefonte, PA) was used. The column was maintained at 30 C, and the sample tray was maintained at 5 C. No guard column was used in the separation, and the detection wavelength was set at 254 nm. The UV flow cell was a standard cell with a 10 mm light path and a4nmslit. The injection volume was 10 L. The flow rate was 1.0 ml/min, and the total run time was 20 min with the following gradient profile: at 0.00 min, 70% Mobile Phase A, 30% Mobile Phase B; at 8.00 min, 40% A, 60% B; at min, 15% A, 85% B; at min, 15% A, 85% B; and at min, 70% A, 30% B. Results and Discussion Derivatization Reaction Two types of derivatization reactions are commonly used for cleaving the ether linkage in cellulose ethers. The classical Zeisel reaction (13) is the basis of the methoxyl determination in USP-Chemical test <431> (11). However, it is not suitable for the determination of hydroxyethyl and hydroxypropyl ethers because formation of ethylene and propylene would reduce the recovery and lead to lower measured substitution values for these groups (14). Hodges et al. (7) modified the Zeisel reaction to solve the recovery problem and thus expanded the usage of the reaction to the determination of methoxyl, ethoxyl, hydroxyethoxyl, and hydroxypropoxyl substitution in mixed or homogeneous cellulose ethers. Another commonly used derivatization reaction is based on chromic acid oxidation (8), which is used in American Society for Testing and Materials (ASTM) standard method D-2363 (9) and the USP assay method for the hydroxypropoxyl group in the monograph of hydroxypropyl cellulose (12). Compared with the modified Zeisel method, the chromic acid oxidation method has inferior precision and reproducibility (7). We, therefore, chose the modified Zeisel method for the cleavage derivatization and developed an LC method for the separation and quantification of the iodomethane and 2-iodopropane formed in the HPMC and HPMCAS polymer reaction solutions. The derivatization procedure is essentially the same as described by Hodges et al. (7) except that no internal standard (toluene) is used in the o-xylene for extraction. We observed that the use of other organic acids, such as oxalic, benzenesulfonic, glutaric, or malonic acid, had the same catalytic effect for this reaction. This is consistent with the observation of Hodges et al. (7), who observed the same catalytic effect for the derivatization reaction when one of the following was used: adipic, succinic, formic, acetic, citric, or valeric acid. The influence of reaction temperature and time on the assay results was studied, and the results are shown in Table 1. It can be seen that a change in the reaction temperature between 130 and 160 C has minimal influence on the assay results, and a reaction time within 60 to 90 min is appropriate for the complete cleavage of both the methoxyl and 2-hydroxypropoxyl groups and the quantitative conversion to iodomethane and 2-iodopropane, respectively. LC Method Optimization Because of its simplicity, isocratic elution was first considered for the separation and quantitation of iodomethane and 2-iodopropane. An example of isocratic separation is shown in Figure 1; the chromatographic conditions were the same as for the final gradient procedure described in the Experimental section under Instrumentation and Operating Conditions,except that isocratic elution with a mobile phase of methanol water ( ) was used, and the run time was 15 min. The sample solution was prepared according to the procedure described in the Experimental section, but the standard solution was prepared by directly diluting iodomethane and 2-iodopropane with methanol. This 15 min isocratic elution gave very good separation for both iodomethane and 2-iodopropane in the standard solution. However, it did not give acceptable separation for the iodomethane peak in the chromatogram for the sample solution because a front shoulder peak appeared under the iodomethane peak (Figure 1b). Further method development work involving varying the organic solvent and mobile phase composition, assisted by computer simulation by DryLab (LC Resources, Inc., Walnut Creek, CA), demonstrated that several small components with the potential to interfere generally eluted between iodomethane and 2-iodopropane. These components had a very different selectivity change in response to the change in mobile phase composition. In order to satisfactorily separate these small peaks from both the iodomethane and the 2-iodopropane peaks, an isocratic elution would require a very

4 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, Before sample analysis in each run sequence, the following injections were made: 1 blank injection of methanol used for sample preparation to determine the peaks due to the solvent or the chromatographic system, 5 consecutive injections of Standard 1, and 1 injection of Standard 2. The peak areas for iodomethane and 2-iodopropane were measured in the chromatograms for the standards. The criteria set forth below established chromatographic conditions that ensure the system was operating in a manner suitable to carry out the analytical procedure: (1) The relative standard deviation of the mean peak area for each compound was 2.0% for the 5 injections of Standard 1. (2) The number of theoretical plates (N) , and the peak tailing factor (T) was between 0.9 and 2.0 when the iodomethane peak from the 5 replicate injections of Standard 1 was used; N and T are as defined in USP <621> (15). 3) The difference between the standard response factor (see calculation below) for Standard 2 and the average response factor for Standard 1 was 2% for both iodomethane and 2-iodopropane. After the system suitability criteria were met, the actual sample analysis was begun. The following sample assay pattern was used for the sample analysis: methanol (blank), standard, sample, sample, standard, sample, sample, standard, etc., methanol (blank). Either Standard 1 or Standard 2 was used in the analysis, but the standard chosen was the same for the bracketed samples. Furthermore, there were always 6 samples between standard brackets. Calculation of Substituent Content (a) The standard response factor (RF) for iodomethane is calculated according to the following equation: RF = (A )(DF)(V) Std (W )(PF) Stock Figure 1. Representative chromatograms obtained by isocratic elution of (a) standard solution [C final (= C stock /DF) for iodomethane and 2-iodopropane are 3.46 and mg/ml, respectively] and (b) the HPMCAS polymer solution (W polymer = mg). Retention times (min) were 3.34 for iodomethane, 5.61 for 2-iodopropane, and 8.76 for o-xylene. where A Std = peak area obtained for iodomethane in the standard solution; DF = dilution factor (DF for iodomethane is 10, i.e., 10 ml/1 ml); V = volume of o-xylene used for preparing the standard, i.e., 10 ml; W Stock = weight, in mg, of iodomethane used for preparing the standard; and PF = purity factor of the iodomethane. The RF for 2-iodopropane is calculated with the same equation by using the corresponding values of A Std,V,W Stock, long elution time for a late eluter such as o-xylene. It was then decided that a gradient elution procedure would be developed. The optimized gradient elution procedure is described in the Experimental section, and an example is shown in Figure 2. This 20 min gradient run used 2-step gradients and gave satisfactory separation of iodomethane and 2-iodopropane in both standard and sample solutions. LC System Suitability Check and Run Procedure Figure 2. Representative chromatograms obtained by gradient elution of (a) the standard solution [C final (= C stock /DF) for iodomethane and 2-iodopropane are 3.57 and mg/ml, respectively], (b) the HPMCAS polymer solution (W polymer = mg), and (c) the methanol blank. Retention times (min) were 8.10 for iodomethane, for 2-iodopropane, and for o-xylene.

5 698 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 Table 2. Summary of experimental results for method accuracy test Iodomethane 2-Iodopropane Theoretical, mg/ml Measured, mg/ml a Recovery, % Theoretical, mg/ml Measured, mg/ml a Recovery, % Overall mean b RSD, % c Mean b for 80% level RSD, % c Mean b for 100% level (3.466 mg/ml) RSD, % c Mean b for 120% level RSD, % c a b c Each measured concentration was calculated from the peak area of the corresponding compound and compared with the average response factor for that compound from the 2 injections of the standard solution that bracketed a given sample set. Mean is the average of the individual values in the specified group. RSD is the relative standard deviation of the mean. and PF for 2-iodopropane. The value of DF for 2-iodopropane is 10, i.e., 10 ml/1 ml. (b) The assay of iodomethane (CH 3 I) in mg for each reacted test polymer sample is given by the following equation: W = Area DF V RFAvg where W = experimentally determined weight, in mg, of iodomethane in the reacted test polymer sample; Area = peak area obtained for iodomethane in the test solution; DF = dilution factor (DF for iodomethane is 10, i.e., 10 ml/1 ml); V = volume of o-xylene used for preparing the sample, i.e., 2 ml; and RF Avg = average RF for iodomethane from the 2 injections of the standard solution that brackets a given sample set. The weight of 2-iodopropane (C 3 H 7 I) is calculated with the same equation by using the corresponding values of Area and RF Avg for 2-iodopropane. The dilution factor for 2-iodopropane is 10, i.e., 10 ml/1 ml. (c) The % (w/w) of methoxyl in the test polymer sample is given by the following equation: Methoxyl, % (w / w) = W W Polymer where W = experimentally determined weight, in mg, of iodomethane in the test solution, and W Polymer = weight, in mg, of the test polymer used for preparing the test solution. (d) The % (w/w) of 2-hydroxypropoxyl in the test polymer sample is given by the following equation: 2-Hydroxypropoxyl, % (w/w) = W W Polymer where W = experimentally determined weight, in mg, of 2-iodopropane in the test solution, and W Polymer = weight, in mg, of the test polymer used for preparing the test solution. The content of both the methoxyl and the 2-hydroxypropoxyl substitutions can be corrected for the value of the loss on drying (LOD) by substituting W Poly - mer (1 LOD) for W Polymer in the equation, where LOD in % (w/w) can be measured by the USP method (16).

6 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, Table 3. Summary of experimental results for injection precision for iodomethane Sample ID No. of samples No. of injections per sample Mean SD a RSD, % b Standard c Method precision for iodomethane Standard d Intermediate precision for iodomethane Standard e 1.3 (97.9, 102.1) f 1.3 Injection precision for 2-iodopropane Standard c Method precision for 2-iodopropane Standard d Intermediate precision for 2-iodopropane Standard e 1.0 (98.0, 101.6) f 0.99 a SD = Standard deviation. b RSD = Relative standard deviation. c Mean area. d Mean response factor. e Mean recovery (%). f The recoveries in parentheses are the low and high extremes. Method Validation The following aspects of the method were studied in order to establish method validation. Selectivity and sensitivity. The assay was shown to be specific for both iodomethane and 2-iodopropane. Typical chromatograms for standard and sample solutions are shown in Figure 2. Analysis of the methanol blank confirmed that there are no interfering peaks due to solvent or chromatographic system peaks in the retention time range of interest. The sensitivity of the method is not an issue in this application because it is an assay method of category I under the definition of USP <1225>-validation of compendial methods (17). However, it is of general interest to determine the sensitivity of the method. The limits of detection (at a signal-to-noise ratio [S/N] of 3) were determined to be 1 and 0.2 g/ml for iodomethane and 2-iodopropane, respectively. The limits of quantitation (at an S/N of 9) were determined to be 2 and 0.8 g/ml for iodomethane and 2-iodopropane, respectively. Linearity, accuracy, and precision. The linearity and concentration range were studied. Thirteen standard solutions Table 4. Summary of experimental results for solution stability test Area t /area 1 for iodomethane a Area t /area 1 for 2-iodopropane Time, h Standard Sample Standard Sample a Area t = peak area at t h; area 1 = peak area at 1 h.

7 700 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 Table 5. Summary of experimental results for assay precision and robustness in analyses of HPMCAS polymer a Conditions b Methoxyl, % (w/w) 2-Hydroxypropoxyl, % (w/w) Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Rxn time, 30 min Rxn time, 45 min Rxn time, 75 min Rxn time, 90 min Rxn temp., 130 C Rxn temp., 140 C Rxn temp., 160 C Flow rate, 0.9 ml/min Flow rate, 0.9 ml/min Flow rate, 1.1 ml/min Flow rate, 1.1 ml/min Column temp., 25 C Column temp., 25 C Column temp., 25 C Column temp., 35 C Column temp., 35 C Column temp., 35 C Overall mean c Standard deviation RSD, % d Mean (at normal conditions) Standard deviation RSD, % Mean (at variations) Standard deviation RSD, % a b c d All values, in % (w/w), were corrected for the value of loss on drying. The values for methoxyl and 2-hydroxypropoxyl substituents in the manufacturer s certificate of analysis are 23.4 and 7.0%, respectively. Normal means at the optimized conditions specified in the Experimental section. If a specific condition is listed, only that condition was changed for the robustness experiments. The values after represent halfwidth of the 95% confidence interval. RSD %, was calculated from the standard deviation divided by the mean.

8 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, Table 6. Summary of experimental results for assay accuracy in analyses of HPMC and HPMCAS polymers Methoxyl Hydroxypropoxyl Polymer Manufacturer CofA, LC assay, CofA, LC assay, % (w/w) a % (w/w) b RD, % c % (w/w) a % (w/w) b RD, % c HPMCAS-LF Shin Etsu HPMCAS-MF Shin Etsu HPMCAS-HF Shin Etsu HPMCAS-MG Shin Etsu HPMC Shin Etsu HPMC-E3 Dow Chemical HPMC-E5 Dow Chemical HPMC-E15 Dow Chemical a b c The values of substitution content measured by the compendial test method (10) from the manufacturer s certificate of analysis. The values of substitution content measured by this LC assay method. The results, in % (w/w), were corrected for the value of loss on drying. RD = Relative difference between the value from the LC assay and the value in the certificate of analysis. were prepared with concentrations ranging from 1.39 to 5.55 mg/ml for iodomethane and from to mg/ml for 2-iodopropane. The linear regression analysis of the 13 standards demonstrated good linearity for iodomethane with values of for R 2 and 0.25% for the y-intercept bias; for 2-iodopropane the values were for R 2 and 0.52% for the y-intercept bias. The 13 standards were also assayed against an individually prepared Standard 1, and recoveries (the experimentally determined weight divided by the theoretical weight) of both iodomethane and 2-iodopropane were calculated for each solution. The results, shown in Table 2, demonstrate that the method can accurately measure the concentrations of both iodomethane and 2-iodopropane in solution against an individually prepared external standard. The precision of the method was also evaluated, and the results are summarized in Table 3. The injection precisions for both iodomethane and 2-iodopropane are typical for an LC instrument. The retention times for both iodomethane and 2-iodopropane were found to be highly reproducible. In one instrument with a dwell volume of 2.0 ml, the average retention times standard deviation (SD), calculated from 55 injections of standards and samples over 3 days, were and min for iodomethane and 2-iodopropane, respectively. In another instrument with a dwell volume of 1.3 ml, the average retention times SD, calculated from 46 injections of standards and samples over 4 days, were and min for iodomethane and 2-iodopropane, respectively. The method precision was calculated from 5 replicate preparations of Standard 1 by an analyst and represented the influence of the variation in experimental bench work. The intermediate precision was calculated from a set of historical data of recoveries of Standard 2 versus Standard 1. Two analysts performed the experiments on 2 instruments during a period of >3 months. The intermediate precision indicates how well the method would perform from analyst to analyst using different instruments over time. The fact that both method and intermediate precisions are relative standard deviations (RSDs) of <2.0% indicates the analyses by the method are highly reproducible. Solution stability. It was found that both the standard and the sample solutions had unacceptable stability (assayed values of substitutions for sample solution or recoveries of standard were reduced by >2%) after 1 day of storage, even at 5 C. The stability was assessed versus freshly prepared (daily) standard solutions. If the aged sample solution was assayed versus the aged standard solution after 1 day, it was found that the assayed values of substitutions were consistent with the values obtained at Day 0 when both solutions were freshly prepared. Nonetheless, it was decided that a 9 h stability test would be performed; the results are shown in Table 4. On the basis of these experimental results, it was determined that all standard and sample solutions should be analyzed within 6 8 h after preparation. Assay accuracy, precision, and robustness. The assay precision and robustness were evaluated by repeated analyses of a lot of HPMCAS polymer for the content of methoxyl and 2-hydroxypropoxyl substitutions. The results obtained by 2 analysts using 2 LC instruments at various conditions are summarized in Table 5. Of the 29 measurements, 12 were made at the optimized conditions. Seventeen measurements were made under conditions in which only one parameter was deliberately varied from the optimized conditions. The relative standard deviations expressed as a percentage for this LC method under all measurement conditions were 2.31% for methoxyl determination, and 3.45% for 2-hydroxypropoxyl determination. In comparison, the relative standard precision for the GC assay method using the same derivatization reaction ranged from 44 to 2.68% for methoxyl determination, and from.96 to 6.02% for 2-hydroxypropoxyl determination (7). This comparison demonstrated that the LC assay method has equal or better precision for determining the content of both methoxyl and 2-hydroxypropoxyl substitutions.

9 702 RASHAN ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 The robustness of the assay method was assessed by statistically comparing the results measured under normal condition with those measured under the varied conditions. First, a statistical F-test demonstrated that there was no statistically significant difference between the variances of the 2 sets of means at the 95% confidence level. Then a statistical t-test assuming equal variances at the 95% confidence level showed no statistically significant difference for the means obtained from the 2 groups for both methoxyl and 2-hydroxypropoxyl determination. Thus, the assay was proved to be robust in regard to variation in derivatization time or temperature, flow rate, or column temperature. To assess the accuracy of the LC method, 4 lots of HPMC polymer and 4 lots of HPMCAS polymer were assayed by the method, and the results were compared with those listed in the certificate of analysis from the manufacturer. The results are shown in Table 6. Most of the values obtained by the 2 methods are in agreement with each other with an approximately 2% relative difference. Two pairs have about a 3% relative difference, and 2 pairs for the 2-hydroxypropoxyl determination have about a 4.5% relative difference. Because the GC method standard precisions could be as large as 2.68% for methoxyl and 6.02% for 2-hydroxypropoxyl, the overall agreement between the 2 assay methods is considered good. Conclusions An alternative LC assay method was developed and validated for the simultaneous determination of the content of methoxyl and 2-hydroxypropoxyl substitutions in HPMC and HPMCAS. The method was proven to be selective and sensitive, and has good linearity with values of for R 2 and 0.25% for the y-intercept bias from 1.39 to 5.55 mg/ml for iodomethane, and for R 2 and 0.52% for the y-intercept bias from to mg/ml for 2-iodopropane. The relative precisions for this LC method were found to be 2.31% for determining methoxyl at the 23.1% (w/w) level, and 3.45% for determining 2-hydroxypropoxyl at the 6.7% (w/w) level. Compared with the current GC compendial test method, the LC assay method has equal or better precision. Both the standard and sample solutions were found to have limited stability (6 8 h) after preparation. This limited stability has not been reported previously in the literature and may have an impact on the reported accuracy/precision of the literature data for the GC method. The LC method has been proven to be robust with respect to variation in derivatization time or temperature, flow rate, or column temperature. It is well suited for quality control in today s fast-paced pharmaceutical laboratories. References (1) Doelker, E. (1993) Adv. Polym. Sci. 107, (2) Zhokhova, F.A., & Zharkov, V.V. (1981) USSR Plast. Massy. 1, (3) Tezuka, Y. (1999) Jpn. Cellul. Comm. 6, (4) Tezuka, Y., Imai, K., Oshima, M., & Ito, K. (1991) Carbohydr. Res. 222, (5) Tezuka, Y., Imai, K., Oshima, M., & Chiba, T. (1991) Polym. J. (Tokyo) 23, (6) Cobler, J.G., Samsel, E.P., & Beaver, G.H. (1962) Talanta 9, (7) Hodges, K.L., Kester, W.E., Wiederrich, D.L., & Grover, J.A. (1979) Anal. Chem. 51, (8) Lemieux, R.V., & Purves, C.B. (1947) Can. J. Res. Sect. B 25, (9) ASTM (2000) American Society for Testing and Materials, Philadelphia, PA, Method D (10) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp (11) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp (12) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp (13) Zeisel, S. (1885) Monatsh. 6, (14) Merz, W. (1967) Fresenius Z. Anal. Chem. 232, (15) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp (16) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp (17) The United States Pharmacopeia (2002) 25th Ed., U.S. Pharmacopeial Convention, Inc., Rockville, MD, pp