Anal. Chem. 2008, 80, 3040-3044
At-Line Quantitative Ion Mobility Spectrometry for Direct Analysis of Swabs for Pharmaceutical Manufacturing Equipment Cleaning Verification Mark A. Strege,* Jessica Kozerski, Nieves Juarbe, and Patrick Mahoney
Lilly Research Laboratories, A Division of Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana 46285
The potential for ion mobility spectrometry (IMS) to provide rapid at-line quantitation of residues on surfaces via direct analysis of swabs is attractive for pharmaceutical manufacturing equipment cleaning verification. In this study, the development of an IMS method to provide acceptable quantitation of active pharmaceutical ingredients and cleaning agents is described. Key modifications to commercially available instrumentation were made to achieve a dynamic range of 5-100 µg per 25 cm2 surface area and acceptable analyte recovery in the presence of ionizable matrix components. The results of this study effectively demonstrate the capability of IMS to serve as an at-line quantitative analytical method. In an ion mobility spectrometry (IMS) instrument, analytes undergo thermal desorption, chemical ionization, and separation as the ions are drawn through atmospheric gas molecules to an ion collector plate. The combination of moderate separation power and nanogram-level sensitivity 1 has led to applications of IMS as a stand-alone monitor for the detection of trace amounts of narcotics, explosives, and chemical warfare agents for security purposes.1-3 The first applications of IMS for the analysis of pharmaceutical compounds were reported approximately 15 years ago4,5 and more recently for the direct analysis of drug formulations.6 Over the past 5 years, the technology has been promoted by instrument manufacturers for use in the pharmaceutical industry as a replacement for high performance liquid chromatography (HPLC) and total organic carbon (TOC) analysis for the testing of swab extracts for the verification and validation of manufacturing equipment cleaning. These applications described the determination of residues corresponding to either active pharmaceutical ingredient (API) or cleaning agents.7-10 The major * To whom correspondence should be addressed. Phone: (317) 276-9116. Fax: (317) 277-5519. E-mail:
[email protected]. (1) Eiceman, G. A.; Karpas, Z. Ion Mobility Spectrometry, 2nd ed.; CRC Press: Boca Raton, FL, 2005; Chapter 1. (2) Hill, H. H.; Siems, W. F.; St. Louis, R. H.; McMinn, D. G. Anal. Chem. 1990, 62 (23), 1201A-1209A. (3) Baumbach, J. I.; Eiceman, G. A. Appl. Spectrosc. 1999, 53 (9), 338A-355A. (4) Lawrence, A. H. Anal. Chem. 1989, 61, 343-349. (5) Eiceman, G. A.; Blyth, D. A.; Shoff, D. B.; Snyder, A. P. Anal. Chem. 1990, 62, 1374-1379. (6) Weston, D.; Bateman, R.; Wilson, I. D.; Wood, T. R.; Creaser, C. S. Anal. Chem. 2005, 77 (23), 7572-7580. (7) Payne, K.; Fawber, W.; Faria, J.; Buaron, J.; DeBono, R.; Mahmood, A. Spectroscopy 2005, 20 (Suppl.), 24-27. (8) Tan, Y.; DeBono, R. Today’s Chemist Work 2004, November, 15-16.
3040 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
feature of IMS touted within these reports has been the speed of analysis. However, reduced analysis time alone is not a significant factor in reducing the time required for equipment release or routine monitoring for most pharmaceutical manufacturers. The traditional manufacturing equipment cleaning verification process consists of surface swabbing, submission of the swabs to a testing laboratory, and extraction of the swabs followed by analysis, concluding with the release of the data via a laboratory information management system (LIMS) system. This entire process can take as long as 24-48 h. In contrast, surface sampling followed by testing without solvent-based extraction is termed “direct swab analysis”. Direct swab analysis uses thermal desorption to extract sample molecules directly from the swab matrix into a gaseous phase from which they can be analyzed as ions. One benefit of direct swab analysis concerns detection sensitivity, because dilutions taking place via extraction in a liquid medium are avoided. Most importantly, direct swab analysis can enable sampling, testing, and decision to release or reclean equipment in a matter of minutes on the floor of the manufacturing suite. In this scenario, at-line testing replaces traditional laboratory analyses, and the financial incentives achievable via a reduction in equipment downtime and resourcing are significant. Several challenges must be addressed, however, prior to the enablement of IMS for at-line analysis. Commercial IMS instruments offer a limited (1-2 orders of magnitude) dynamic range of nanogram to low microgram levels.1 Applications for cleaning verification within the pharmaceutical industry, however, require determination of residue mass up to 4 µg/cm2 (i.e., 100 µg/swab for a 25 cm2 surface area), a level generally accepted within the industry to correspond to the limit of visual cleanliness.11 Additionally, charge exchange reactions and competitive ionization may result in signal suppression of one or more analytes when a mixture of components (such as API, excipients, and cleaning agents) is present.12 Finally, acceptable surface recoveries and quantitative direct swab analyses have yet to be demonstrated through the use of high temperature-resistant swabs. In this investigation, effective measures to address the issues associated with dynamic range, competitive ionization, and surface (9) Walia, G; Davis, M.; Stefanou, S.; DeBono, R. Pharm. Technol. 2002, April, 72-78. (10) Walia, G.; Davis, M.; Stefanou, S. Pharma. Process. 2003, September, 2022. (11) Fourman, G. L.; Mullen, M. V. Pharm. Technol. 1993, April, 54-60. (12) Eiceman, G. A.; Karpas, Z. Ion Mobility Spectrometry, 2nd ed.; CRC Press: Boca Raton, FL, 2005; Chapter 3. 10.1021/ac702264f CCC: $40.75
© 2008 American Chemical Society Published on Web 03/12/2008
Figure 1. IMS plasmagrams of 25 µg of duloxetine obtained using the conditions described in the Experimental Section and PTFE nozzles with aperature diameters corresponding to (A) 0.044 and (B) 0.022 in.
recovery are reported. IMS quantitation via direct swab analyses is demonstrated for the determination of duloxetine hydrochloride, the active pharmaceutical ingredient (API) in the drug product Cymbalta, and the surfactant component of the cleaning agents used in the associated manufacturing process. EXPERIMENTAL SECTION Instrumentation. A Kaye Validator ion trap mobility spectrometer was purchased from GE Sensing (Billerica, MA). Unless otherwise indicated, instrument desorber and detector temperatures were 249 and 205 °C, respectively, and a default sample acquisition time of 180 s (15 scans acquired per second, alternating positive and negative ion mode) was utilized. Negative ion mode separation profiles (“plasmagrams”) were created by the instrument software through summation and signal averaging across the total acquisition time. The drift gas was partially dried atmospheric air. Default gas flow settings were 500 cm3/min in the sample flow chamber and 250 cm3/min on the detector flow. Custom-built desorber nozzles were constructed from hightemperature resistant polytetrafluoroethylene (PTFE) on-site at Eli Lilly and Company. Materials. Samples were prepared using duloxetine hydrochloride API and formulated drug product provided by Eli Lilly and Company (Indianapolis, IN). Cleaning agents (CIP-100 and CIP-200) and a proprietary amphoteric surfactant were provided by Steris, Inc. (St. Louis, MO). The swabs used in this study were constructed from a compressed, high temperature-resistant polyimide fiber material (2 cm × 5 cm strips) and were provided by GE Sensing. Polished 316-L stainless steel plates were provided by Eli Lilly and Company for use in surface recovery studies. Methods. Samples of duloxetine and surfactant were prepared in methanol and water, respectively, at a concentration of 1.0 mg/
mL, and aliquots of appropriate volume were spiked directly onto swabs or onto the surfaces (5 cm × 5 cm) to be swabbed. All samples were prepared and analyzed in triplicate. For stainless steel surface recovery experiments, the swabs were wet via the application of 100 µL of methanol to the swab approximately 1 cm from the contact end. All samples were air-dried prior to analysis. Raw data generated by the IMS instrument were in the form of integrated area counts based on detector signal response. The data from the instrument were transferred into an Excel (Microsoft Corp, Redmond, WA) spreadsheet for conversion of sample response into units of µg of analyte recovered. RESULTS AND DISCUSSION Optimization of Dynamic Range for Duloxetine. Following its exposure to Cymbalta drug product, process equipment is subjected to a series of automated rinses with cleaning agents and water. Within the drug product composite containing excipients, the API is both the least water-soluble and most biologically potent component. For these reasons, the API is the analytical target for cleaning verification. In the negative ion IMS plasmagram obtained for duloxetine displayed in Figure 1A, peaks corresponding to water cluster reactant ions (the predominant chemical species generated by the 63Ni ionization source), duloxetine proton-abstracted monomer, and proton-abstracted dimer are present in order of migration (the identity of the small peak present at approximately 6 ms was unknown but duloxetine-related). The presence of the three major components observed for duloxetine is typical for pharmaceutical compounds analyzed by IMS.4,5 As dictated by the dynamics of charge conservation, the signal intensity of the reactant ions was observed to decrease as the duloxetine concentration increased, Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
3041
Figure 2. Duloxetine standard response curves obtained over the range of 5-100 µg using PTFE nozzles with the following aperature diameters (2) 0.044, ([) 0.022, and (9) 0.011 in.
indicating the saturation of the capacity of the ionization source at approximately 25 µg. The response of duloxetine dimer over the range of 5-100 µg is displayed as the top curve in Figure 2. In the Validator instrument, sample molecules are drawn forward from the thermal desorber chamber through a 0.044 in. aperture at the center of a PTFE nozzle prior to passage into the 63Ni ionization source. To achieve the dynamic range of interest, saturation of the ionization source was avoided by limiting the amount of sample molecules passing through the ionization chamber per unit time through decreasing the diameter of the PTFE nozzle aperture. To evaluate the influence of a decrease in nozzle aperture upon the dynamic range, nozzles with apertures 50% (0.022 in.) and 25% (0.011 in.) of the diameter of the standard nozzle aperture were custom-built. With each of these nozzles, duloxetine was analyzed at levels corresponding to 5, 25, 50, 75, and 100 µg. When both the 0.022 and 0.011 in. aperture nozzles were employed, the duloxetine plasmagram indicated the presence of monomer only, and the reactant ion peaks maintained intensity across all sample concentrations (see Figure 1B). These effects were a result of the significant decrease in sample flow (dimers are known to form only at high sample concentrations).12 The merger of the multiple duloxetine components into the monomer alone provided benefits in sensitivity. The results, based upon quantitation of the negative ion monomer peak for the 0.022 and 0.011 in. apertures, were plotted in Figure 2 and demonstrated that the reduction in aperture diameter resulted in a decrease in signal response so that the ionization source saturation point was extended to >100 µg. Although the extension of the dynamic range was most dramatic in the data set generated using the 0.011 in. aperture, it was observed that the signal for duloxetine at 5 µg was very near the limit of detection under these conditions. Therefore, the 0.022 in. aperture diameter was chosen as the preferred nozzle configuration for further optimization. The influence of sample acquisition time (during which the analyte response is summed) upon dynamic range is displayed in Figure 3 as plots of the summed response vs duloxetine level across a range of 60-360 s. When a 0.022 in. nozzle aperature diameter was employed, the sample appeared to require at least 360 s to completely pass from the desorption chamber into the detector. Because 360 s provided the optimal dynamic response range, this setting was utilized for method validation. Analysis of Cleaning Agents. The CIP product line of cleaners consists of a mixture of water, inorganic and organic acids or bases, and relatively low concentrations of surfactants and 3042 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
Figure 3. Duloxetine standard response curves obtained over the range of 5-100 µg using a 0.022 in. aperture diameter PTFE nozzle and the following sample acquisition times ([) 60, (9) 120, (b) 180, (2) 240, ([) 300, and (9) 360 s. A logarithmic fit was performed on the 360 s data set.
wetting agents13 and is utilized by Eli Lilly and Company within the Cymbalta manufacturing process. Because the surfactants are the least water-soluble components, they are expected to be the last constituents of the cleaner to leave the surface during the water rinse steps of the equipment cleaning procedure and therefore may serve as target compounds for the demonstration of effective rinsing. According to the vendor, the surfactants used in the CIP cleaners are a mixture of four proprietary amphoteric surfactants.14,15 Within the surfactant plasmagram, four components were visible and the largest peak (at 4 ms) was chosen for quantitation. Analysis of a mixture of the surfactant and duloxetine is displayed in Figure 4. A plot of instrument response vs mass of surfactant revealed a response curve similar to that achieved for duloxetine. Validation of Quantitation. According to the ICH guidelines, a quantitative analytical method must be validated for linearity, specificity, range, limits of detection (LOD) and quantitation (LOQ), precision, and accuracy.16 For IMS, because the response over the range of interest is logarithmic, the validation of “linearity” is considered to be represented by the logarithmic fits of the response curves for surfactant and duloxetine (Figure 3) which displayed correlation coefficients (0.9819 and 0.9973, respectively) acceptable for this application. Specificity was demonstrated by the fact that all of the peaks corresponding to the reactant ions, duloxetine, and surfactant were spatially resolved within the IMS plasmagram, as was evident in Figure 4. The range was demonstrated through linearity, precision, and accuracy/recovery, as described below. The LOD and LOQ were estimated by determining the standard deviation of the response from a series of six replicate injections of 5 µg standards and then multiplying that number by 3 and 10, respectively. Through this experiment, the LODs for duloxetine and surfactant were 0.9 and 3.2 µg, respectively, and the LOQs were 3.0 and 10.6 µg, respectively. Swab-to-swab precision repeatability and day-to-day intermediate precision were determined by triplicate analyses of swabs (13) (14) (15) (16)
CIP product information literature provided by Steris Corp. Steris Corp., personal communication. Lomax, E. G. Soap Cosmet., Chem. Spec. 1972, 48 (11), 29-32, 66. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Guidelines. Validation of Analytical Procedures: Text and Methodology; 2007; Section Q2.
Figure 4. IMS plasmagram of a mixture of 25 µg of duloxetine and 100 µg of surfactant obtained using a 0.022 in. PTFE nozzle. Peaks are labeled to indicate (R) reactant ions, (D) duloxetine, and (S) surfactant. Table 1. Repeatability of Triplicate Swab Analyses of Duloxetine and Surfactant Standards Performed on Three Consecutive Daysa mass of residue [µg] 5 25 50 75 100 a
duloxetine % RSD 2.6-6.8 (8.9) 2.4-4.6 (4.6) 0.8-2.7 (3.1) 1.0-1.3 (3.0) 1.5-3.8 (3.6)
surfactant % RSD 9.3-34.3 (24.3) 7.5-26.4 (20.3) 8.0-13.2 (11.0) 3.0-7.4 (11.1) 3.4-8.3 (14.9)
Day-to-day intermediate precision values are indicated in parentheses.
spiked with duloxetine or surfactant at five levels obtained over the course of 3 days (see Table 1). For duloxetine, swab-to-swab repeatability was 25 µg resulted in suppression of surfactant signal recovery to Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
3043
levels of 40-50%. To validate the recovery of duloxetine from a matrix containing excipients, 60 mg of Cymbalta drug product pellets (containing approximately 20% API w/w) were solublized in 50 mL of methanol and then loaded on swabs at levels corresponding to 5, 25, 50, and 100 µg of total residue weight. At these levels, mean recoveries of API corresponding to 95, 75, 78, and 56% were achieved. For surface recovery experiments, stainless steel plates were spiked with duloxetine and surfactant solutions and then evaporated to dryness to deposit residues corresponding to 5, 25, and 100 µg per 25 cm2. Care was taken during application of the swab to the coupons to avoid exposure of the swab fabric to the surface beyond the defined 2 cm contact region extracted in the desorber. The results of these experiments demonstrated recoveries ranging from 65 to 118% for duloxetine and 52-54% for surfactant, with the response for the surfactant recovery at 5 µg < the LOD. Recoveries greater than or equal to 50% are generally accepted within the pharmaceutical industry for the demonstration of successful surface sampling.
for future pharmaceutical applications include an adjustable sample flow controller between the desorber and the ionization chamber and also an ionization source that may provide greater capacity. Evident in this study was the importance of a quantitative assessment of the effects of ionization suppression upon the analysis of swabs containing multiple components. For pharmaceutical process equipment cleaning validation, it is an accepted practice to assess the quantitative recovery of target components from the surfaces of interest so that a “surface recovery factor” can be implemented to correct the final calculations for the loss of target compounds to the surface during swab sampling. In analogous fashion, for the use of IMS for the analysis of swabs potentially containing a mixture of components, an “instrument recovery factor” may be considered. For example, the results of this current investigation suggest that an instrument recovery factor of 2.5 (based on a worst-case signal recovery of 40%) be employed for concurrent quantitation of duloxetine and surfactant across the range of 5-100 µg.
CONCLUSIONS The benefits of using a specific method for swab analysis include both a reduction in the risk of false positive results and the efficiency gained by testing for both API and cleaner concurrently from one swab sample. The use of custom-built PTFE nozzles enabled the GE Validator instrument to provide a range of 5-100 µg for both the duloxetine API and the surfactant component of the CIP cleaners. Modifications to the current commercially available instrument that would be of great interest
ACKNOWLEDGMENT The authors wish to acknowledge Douglas Classen, Neeraj Maddiwar, and William Cleary of Eli Lilly and Company and Derek Brand of GE Sensing for technical assistance.
3044
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
Received for review November 2, 2007. Accepted February 5, 2008. AC702264F
