Analysis of pharmaceutical formulations using atmospheric pressure ion mobility spectrometry combined with liquid chromatography and nano-electrospray ionisation

Abstract

The hyphenation of liquid chromatography with atmospheric pressure ion mobility spectrometry is reported using a custom-made dynamic nano-electrospray ionisation (nano-ESI) interface. The analysis of pharmaceutical actives is described, including beta blocker (timolol), antidepressant (paroxetine), analgesic (paracetamol) and opiate (codeine) preparations. On-line ultraviolet diode array (UV) spectroscopic detection was used prior to sample ionisation, to evaluate chromatographic and nano-ESI interface performance. Active drug responses were characterised by chromatographic retention time and electrophoretic ion mobility drift time, and selected ion mobility responses were used to evaluate method performance. Limits of detection for active drugs were in the low-nmol to pmol range. Quantitative responses were investigated using a series of standard solutions of caffeine, showing good linearity (R² = 0.9982, n = 6) and reproducibility (RSD = 2.3 %, n = 6). The analysis of an over the counter pharmaceutical formulation demonstrates the potential of ion mobility spectrometry combined with liquid chromatography and nano-electrospray ionisation for the rapid determination of active drugs, as a result of the electrophoretic separation and selectivity afforded by IMS.

---

Introduction

Ion mobility spectrometry (IMS) is a gas-phase electrophoretic technique in which ions are separated on the millisecond timescale on the basis of their relative mobility, determined by reduced mass, charge and collision cross-section. The underlying theory and detailed applications of this technique have been presented in a number of reviews.^1–3 The speed and selectivity of atmospheric pressure ion mobility spectrometry (AP-IMS) has shown utility in many areas including monitoring of chemical warfare agents⁴ and in security enforcement.⁵ In general, for most AP-IMS instruments, a ⁶³Ni radiation source¹ has been used to generate sample ions. However, as a result of safety and legislative requirements, and the need to extend the range of analytes amenable to AP-IMS, alternative sources have been developed, including coronaspray,⁶ thermionic,⁷ laser,⁸ surface⁹ and photoionisation.¹⁰ Electrospray ionisation (ESI), first attempted by Dole in 1968,¹¹ was reported in combination with IMS a few years later¹² and has since been used by several groups.^13–15 However, seminal work by Hill and co-workers, showing successful desolvation of gas phase ions prior to IMS detection, allowed the first measurement of reduced mobilities from ESI-generated ions.^6,16–18 The miniaturization of the ESI technique, giving rise to nano-electrospray (nano-ESI), was theoretically described and experimentally demonstrated by Wilm and Mann in conjunction with mass spectrometry,^19,20 allowing lower sample consumption and higher sensitivity compared to ESI. Previous work by our laboratory has shown the coupling of AP-IMS to a static nano-ESI source²¹ for the study of amino acids and peptides.

The direct analysis of pharmaceutical drug formulations has been performed using IMS combined with ESI,²² and desorption electrospray ionisation (DESI) with time-of-flight mass spectrometric detection,²³ and the coupling of IMS with solid phase microextraction (SPME-IMS)²⁴ was used for determination of parabens in pharmaceutical products (cream, lotion, solution and ointment). The hyphenation of AP-IMS with liquid chromatography (LC) has been reported by Hill and co-workers.²⁵ The authors evaluated the system by studying a mix of peptides, showing improved separation using the combined LC/IMS approach. Recently, Valentine et al.²⁶ showed the potential of liquid chromatography combined with low pressure ion mobility mass spectrometry for the analysis of complex mixtures of peptides, allowing one dimensional (1D) or two dimensional (2D) LC/IMS/MS measurements. However, the analysis of pharmaceutical actives and formulations by LC/IMS has not been reported.

This paper describes the novel coupling of liquid chromatography and on-line UV detection with AP-IMS using a custom-made dynamic nano-ESI interface. The application of LC/AP-IMS to the selective analysis of pharmaceutical actives and an over-the-counter pharmaceutical formulation is reported. The potential to resolve compounds with similar LC retention time (RT) using IMS drift time, and vice versa, shows the utility of the orthogonal separation provided by the LC/AP-IMS system.

Experimental

Chemicals

HPLC grade methanol and analytical reagent (AR) grade glacial acetic acid were purchased from Fisher Scientific (Loughborough, UK). Distilled water was obtained in-house using a Triple Red water purification system (Triple Red, Long Crendon, UK). Timolol maleate, caffeine and paracetamol were donated by AstraZeneca, and ortho-fluoro analogue of paroxetine was donated by GlaxoSmithKline. Analyte stock solutions were prepared in 49.5 : 49.5 : 1 (v/v/v) methanol–water–acetic acid at 1 mg mL⁻¹ (timolol: 2.66 mmol L⁻¹, ortho-fluoro analogue of paroxetine: 2.31 mmol L⁻¹, caffeine: 6.62 mmol L⁻¹ and paracetamol: 5.15 mmol L⁻¹) and diluted accordingly, using the same solvent composition, to give a series of solutions with final concentrations in the nmol L⁻¹ to µmol L⁻¹ range. HPLC solvents were degassed on-line, prior to mixing, using a DGU-14A membrane degasser (Shimadzu, UK).

Instrumentation

A schematic representation of the LC/UV/nano-ESI/AP-IMS instrumentation is shown in Fig. 1. Reverse phase liquid chromatographic separation and on-line UV detection was carried out prior to nano-ESI ionisation and AP-IMS detection, using a modular LC-10AVP LC system (Shimadzu, Milton Keynes, UK), consisting of two binary pumps (LC-10ADVP), gradient controller (SCL-10AVP), UV-DAD unit (SPD-M10AVP), autosampler (SIL-10ADVP) and column oven (CTO-10ACVP), which was computer controlled using ClassVP software (Version 5.032, Shimadzu, Milton Keynes, UK). Separation was carried out by injecting aliquots of sample solution (20 µL) into an isocratic solvent flow 49.5 : 49.5 : 1 (v/v/v) methanol–water–acetic acid (500 µL min⁻¹), directed through a SymmetryShield^™ RP₁₈ (C₁₈) analytical column (150 × 4.6 mm × 5 µm, Waters, Manchester, UK) held isothermally at 30 °C in the column oven. UV spectra were acquired from 190 to 800 nm over the analytical LC run (10.0 minutes, 2 s scan⁻¹).

Fig. 1 Schematic diagram of LC/AP-IMS system, showing the nano-ESI interface.

IMS experiments were carried out using a high temperature ion mobility spectrometer (HTIMS, Smiths Detection, Watford, UK), modified to accept a custom-made nano-ESI interface, with a dynamic nanospray emitter (120 mm × 75 µm ID, 10 µm tip orifice, New Objectives, CA, USA) mounted in a flow-through tee-piece (1/16″ × 1/16″ × 0.050″, SGE, Milton Keynes, UK) to form a liquid junction, made liquid tight by graphite ferrules (072623 or 072626, SGE, Milton Keynes, UK) and fixed to a mechanical x-y-z manipulator (Coherent Technologies, Kingston-Upon-Thames, UK). The LC eluent was directed to waste through the tee-piece, with a small flow (320 nL min⁻¹) being sampled by the dynamic nanospray emitter. A spray voltage (1.8–2.5 kV) was applied to the tee-piece liquid junction using a high voltage DC power supply (476R, Brandenburg Limited, Dudley, UK). The tip of the nano-spray emitter was positioned inside the inlet of the IMS cell, close to the gate screen, at an optimised distance to afford both sensitivity and adequate desolvation of spray formed by the emitter. Ions were introduced into the drift region via a Bradbury–Nielsen gate²⁷ (pulse width 50 µs, repetition period 20 ms). A unidirectional flow of nitrogen (200 mL min⁻¹, cylinder supply of >99.5% purity) was used as the drift gas (Air Products, UK) and an electric field gradient of 200 V cm⁻¹ was applied to the drift cell (4.2 cm length), operated at an optimised temperature of 200 °C. Mobility responses and performance of the AP/IMS unit were monitored using a digital real-time oscilloscope (Tektronix TDS-210, SJ Electronics, Coggershall, Essex, UK) with scan-averaging facility enabled (2 s per acquisition). Mobility spectra were acquired (sampling rate of 20 kHz, 128 scans averaged) using a National Instruments PCI-1200 DAQ card (National Instruments, Newbury, Berkshire, UK) and processed with Trimscan Link 32 software (Graseby Dynamics, Smiths Detection, Watford, UK), which allowed representation of real-time (waterfall) mobility spectra and further exportation (Microsoft Excel) as selected ion mobility chromatograms, showing intensity vs. time (s) for a selected drift time (ms) window.

Results and discussion

Optimisation of the LC/AP-IMS system

Temperature. A mixture of timolol (2.31 mmol L⁻¹) and ortho-fluoro analogue of paroxetine (2.66 mmol L⁻¹) was used to optimise the nano-ESI interface and IMS detector. The LC/UV chromatogram (wavelength 283 nm) in Fig. 2a shows a typical analysis for this test mixture. The ion mobility spectra recorded for chromatographic peaks with retention times (RT) of 6.5 min (timolol) and 7.8 min (ortho-fluoro analogue of paroxetine) are shown in Fig. 2b and 2c, respectively. Drift times for timolol (10.65 ms, Fig. 2b) and ortho-fluoro analogue of paroxetine (11.00 ms, Fig. 2c) were observed at an IMS temperature of 200 °C. Manipulation of exported IMS data (Microsoft Excel) allowed compounds with characteristic drift time to be further interrogated by producing a selected ion mobility chromatogram, showing the evolution of IMS intensity over the chromatographic run time for a particular drift time window (ms timescale). Fig. 2d shows the overlaid selected ion mobility chromatograms for timolol (drift time of 10.65 ms, blue trace) and ortho-fluoro analogue of paroxetine (11.00 ms, red trace) over the 10.0 minute LC run, illustrating the selectivity afforded by the IMS separation. A small contribution in each drift time window from both analytes arises as a result of their small differences in drift time. The shoulder observed on the timolol peak may be explained by the existence of an interferent, with characteristics similar to timolol.

Fig. 2 (a) LC/UV-DAD trace (283 nm) for the mixture of timolol and ortho-fluoro analogue of paroxetine, (b) ion mobility spectrum of timolol recorded at LC RT of 395 s, (c) ion mobility spectrum of ortho-fluoro analogue of paroxetine recorded at LC RT of 473 s, and (d) selected ion mobility chromatograms for timolol (drift time 10.65 ms) and ortho-fluoro analogue of paroxetine (11.00 ms) over LC run time of 10.0 min.

The effect of the temperature of the IMS cell (100–200 °C) was studied to determine optimised conditions for mobility separation of gas-phase ions, as well as those for assisting desolvation in the nano-ESI process. The results of this experiment are shown in Fig. 3. The temperature of the IMS cell was varied between 100 and 200 °C (in steps of 25 °C) for a series of replicate LC injections of the mixture of timolol and ortho-fluoro analogue of paroxetine. Real-time (waterfall) IMS spectra (0–25 ms) were recorded over the LC run (10.0 min), to investigate IMS resolution and sensitivity for the two-analyte mixture. Overlaid mobility spectra are shown for the mixture of timolol and ortho-fluoro paroxetine at temperatures of 100 °C (drift time of 12.80 and 13.75 ms, respectively), 150 °C (11.65 and 12.20 ms) and 200 °C (10.65 and 11.00 ms) in Fig. 3a, b, and c, respectively, along with corresponding overlaid selected ion mobility chromatograms (Fig. 3d, e, and f). Ion mobility separation of the two analytes improved with decreasing temperature, but at 100 °C (Fig. 3a), desolvation and stability of the nano-ESI spray was very poor, shown by noise along the baseline of the selected ion mobility chromatogram (Fig. 3d). Increasing the temperature to 150 °C gave a slight increase in sensitivity and a minor improvement in baseline noise (Fig. 3b and 3e), as a result of making the desolvation process more efficient, although mobility resolution of the two analytes decreased slightly. A further, small decrease in ion mobility separation was observed at 200 °C, together with a marked improvement in baseline noise, afforded by greatly improved desolvation of the nano-ESI sample spray (Fig. 3f). A temperature of 200 °C was chosen for future experiments, including those for optimisation of the nano-ESI interface.

Fig. 3 Real-time ion mobility spectra from the LC/AP-IMS analysis of the mixture of timolol and ortho-fluoro analogue of paroxetine recorded at (a) 100 °C, (b) 150 °C, and (c) 200 °C. Selected ion mobility chromatograms for timolol (blue) and ortho-fluoro analogue of paroxetine (pink) at (d) 100 °C, (e) 150 °C, and (f) 200 °C, respectively.

Nano-ESI interface design. Optimisation of the flow-through nano-ESI interface was important, as the design had to allow efficient sampling of a very low flow to the nano-ESI tip (320 nL min⁻¹), without LC band broadening, whilst splitting most of the LC eluent (500 µL min⁻¹) to waste. Injection of a caffeine solution (257 µmol L⁻¹, 20 µL) was carried out to compare the LC peak width at the base for caffeine by LC/UV detection (wavelength 283 nm, retention time 489 s) with the peak width at the base for the selected ion mobility chromatogram for caffeine (drift time 8.00 ms, retention time 497 s). These experiments showed no significant broadening or tailing of the LC peak for caffeine (data not shown), when comparing the LC/AP-IMS data to relevant LC/UV data, which indicated efficient sampling by the nano-emitter and adequate refreshing of the low dead-volume in the tee-piece liquid junction.

Quantitative analysis by LC/AP-IMS

A quantitative study was performed using caffeine (drift time 8.00 ms) to assess the dynamic range of the LC/AP-IMS method. Analyte solutions were prepared (5.0 to 500 µmol L⁻¹, 20 µL injected) and analysed using optimised parameters (vide infra). A calibration curve was produced using the intensity of the selected ion mobility response for caffeine against concentration, which shows two distinct regions (Fig. 4). A linear region is observed for the range 5 to 64 µmol L⁻¹, with a correlation coefficient of R² = 0.9982 (n = 6). At higher concentrations (Fig. 4, inset) there is a rapid transition to saturation of the IMS, which provides a clear upper limit to the linear range of the detector. The limit of detection (LOD) for caffeine in these analyses was 900 nmol L⁻¹ (18 pmol on column), corresponding to a signal-to-noise ratio of 3 : 1. Replicate injections of a caffeine solution (50 µmol L⁻¹, n = 6) gave an RSD of 2.3%, highlighting the good reproducibility of the hyphenated technique.

Fig. 4 Calibration curve for the LC/AP-IMS analysis of standard solutions of caffeine (5 to 64 µmol L⁻¹, n = 6), with inset panel showing non-linearity due to saturation for extended concentration range.

A study of the effect of concentration on IMS performance was also carried out using paracetamol. Analyte solutions were prepared (66 to 660 nmol L⁻¹, 20 µL injected) and analysed using optimised parameters (vide infra). At higher analyte concentration (660 nmol L⁻¹), the IMS spectrum (LC retention time approx. 460 s) displayed two IMS responses in addition to the solvent peak (denoted by *), shown in Fig. 5a. The IMS peak at 7.25 ms is assigned to the protonated monomer of paracetamol (i.e. [M + H]⁺), whereas the response at 10.40 ms is assigned to the dimeric species (i.e. [2M + H]⁺). A similar observation of a paracetamol dimer by electrospray ionisation has been reported in recent work by Williams et al. in a study of various drugs by three different ionisation techniques (desorption electrospray ionisation (DESI), direct analysis in real time (DART), and desorption atmospheric pressure chemical ionisation (DAPCI)).²⁸ At a lower analyte concentration (132 nmol L⁻¹) the intensity of both peaks are reduced (Fig. 5b) and, finally, at 66 nmol L⁻¹ the dimer is no longer observed (Fig. 5c).

Fig. 5 Ion mobility spectra of paracetamol at a concentration of (a) 660 nmol L⁻¹, (b) 132 nmol L⁻¹ and (c) 66 nmol L⁻¹.

Analysis of a pharmaceutical formulation by LC/AP-IMS

A Prontalgine tablet (Boehringer Ingelheim, France), containing paracetamol, caffeine, and codeine as active drug compounds, was analysed using the optimised LC/AP-IMS method. Tablet samples were prepared for analysis by dissolving a crushed tablet in methanol–water–acetic acid (v/v/v) 49.5 : 49.5 : 1.0. Aliquots of these solutions (20 µL) were injected for analysis by LC/AP-IMS.

Analysis of the Prontalgine tablet (0.1 mg mL⁻¹, 20 µL injected) gave LC/UV-DAD and LC/AP-IMS traces displaying a strong chromatographic peak (RT = 458 s) for paracetamol (71.4% (w/w) of the tablet). The associated IMS spectrum (Fig. 6a) shows both monomeric and dimeric ion species and is similar to Fig. 5b, where the paracetamol was analysed without the presence of other actives and excipients. The LC peak for caffeine (8.9% (w/w) of the tablet) has a retention time of 489 s, and the IMS spectrum is shown in Fig. 6b, with a single characteristic IMS response at 8.00 ms. A smaller LC peak was observed (RT 250 s) for the codeine active drug (Fig. 6c), present at a much lower relative concentration (3.6% (w/w) of the tablet). As the codeine response was weaker, higher concentrations of tablet solutions were injected to improve sensitivity for this particular active drug. Selected ion mobility chromatograms showing LC responses for codeine (10.1 ms), caffeine (8.00 ms) and paracetamol (7.35 ms) at tablet concentrations of 1.0 mg mL⁻¹, 0.1 mg mL⁻¹, and 0.02 mg mL⁻¹ are shown in Fig. 7. The injected amount of each compound was 2.41 nmol for codeine, 0.92 nmol for caffeine, and 1.88 nmol for paracetamol at their optimal concentration conditions. In the case of the codeine trace (Fig. 7a), the presence of the paracetamol can be observed at longer retention time because of the similar drift times for codeine and the paracetamol dimer.

Fig. 6 Ion mobility spectra recorded for the analysis of a Prontalgine tablet at the LC peak maxima under optimised concentration conditions for (a) paracetamol (0.02 mg mL⁻¹ solution), (b) caffeine (0.1 mg mL⁻¹ solution), and (c) codeine (1 mg mL⁻¹ solution).

Fig. 7 Selected ion mobility chromatograms showing LC peak responses for (a) the codeine (10.1 ms, 1.0 mg mL⁻¹ solution), (b) the caffeine (8.00 ms, 0.1 mg mL⁻¹ solution), and (c) the paracetamol (0.02 mg mL⁻¹ solution, monomer at 7.35 ms and dimer at 10.45 ms).

These results show the potential of hyphenation between liquid chromatography and atmospheric pressure ion mobility spectrometry, using nano-ESI ionisation, as a novel approach for the rapid and selective analysis of pharmaceutical actives and preparations. In addition, these analyses show that, much like analyses by conventional mass spectrometric methods using ESI-based ionisation, consideration must be given to working dynamic range and the concentration-dependent formation of multimeric ion species.

Conclusion

The novel hyphenation of liquid chromatography with atmospheric pressure ion mobility spectrometry using dynamic nano-electrospray ionisation has been applied to the rapid analysis of pharmaceutical tablet formulations. Selective detection of active drugs, characterised by ion mobility drift time and LC retention time, has been demonstrated for sub-nmol concentrations of active drugs. The quantitative performance of the method has been investigated showing excellent linearity and reproducibility over a useable working range, and key reproducible parameters for optimisation of the LC/AP-IMS systems have been established.

Acknowledgements

This research was carried out under the UK National Initiative in Ion Mobility Spectrometry (NIIMS), a consortium supporting research into the development and application of ion mobility spectrometry at Nottingham Trent University and the University of Manchester, supported by AstraZeneca, GlaxoSmithKline, and Waters Corporation. The authors wish to acknowledge Paul Arnold (Smiths Detection, Watford, UK) for his assistance.

References

1. G. A. Eiceman and Z. Karpas, Ion Mobility Spectrometry, 2nd edn, CRC Press, Boca Raton, FL, USA, 2004.
2. G. A. Eiceman, Trends Anal. Chem., 2002, 21, 259.
3. C. S. Creaser, J. R. Griffiths, C. J. Bramwell, S. Noreen, C. A. Hill and C. L. P. Thomas, Analyst, 2004, 129, 984.
4. R. B. Turner and J. L. Brokenshire, Trends Anal. Chem., 1994, 13, 281.
5. G. A. Eiceman and J. A. Stone, Anal. Chem., 2004, 76, 390A.
6. C. B. Shumate and H. H. Hill, Anal. Chem., 1989, 61, 601.
7. G. E. Spangler, J. P. Carrico, Jr. and D. N. Campbell, U.S. Pat. 4928033, May 22, 1990.
8. D. M. Lubman and M. N. Kronick, Anal. Chem., 1982, 54, 1546.
9. C. Wu, H. H. Hill, U. Kh. Rasulev and E. G. Nazarov, Anal. Chem., 1999, 71, 273.
10. C. S. Leasure, M. E. Fleischer, G. K. Anderson and G. A. Eiceman, Anal. Chem., 1986, 58, 2142.
11. M. Dole, L. L. Mack, R. L. Hines, R. C. Mobley, L. D. Ferguson and M. B. Alice, J. Chem. Phys., 1968, 49, 2240.
12. D. Teer and M. Dole, J. Polym. Sci., 1975, 13, 985.
13. L. W. Beegle, I. Kanik, L. Matz and H. H. Hill, Anal. Chem., 2001, 73, 3028.
14. C. A. Srebalus, J. Li, W. S. Marshall and D. E. Clemmer, Anal. Chem., 1999, 71, 3918.
15. R. W. Purves, D. A. Barnett, B. Ells and R. Guevremont, Rapid Commun. Mass Spectrom., 2001, 15, 1453.
16. R. W. Hallen, C. B. Shumate, W. F. Siems, T. Tsuda and H. H. Hill, J. Chromatogr., A, 1989, 480, 233.
17. D. Wittmer, Y. H. Chen, B. K. Luckenbill and H. H. Hill, Anal. Chem., 1994, 66, 2348.
18. D. G. McMinn, J. A. Kinzer, C. B. Shumate, W. F. Siems and H. H. Hill, J. Microcolumn Sep., 1990, 2, 188.
19. M. Wilm and M. Mann, Anal. Chem., 1996, 68, 1.
20. M. Wilm, G. Neubauer and M. Mann, Anal. Chem., 1996, 68, 527.
21. C. J. Bramwell, M. L. Colgrave, C. S. Creaser and R. Dennis, Analyst, 2002, 127, 1467.
22. C. J. Bramwell, M. L. Colgrave, C. S. Creaser and R. Dennis, Proceedings of the 50th American Society of Mass Spectrometry, Orlando, USA, June 2002.
23. D. J. Weston, R. Bateman, I. D. Wilson, T. R. Wood and C. S. Creaser, Anal. Chem., 2005, 77, 7572.
24. J. K. Lokhnauth and N. H. Snow, Anal. Chem., 2005, 77, 5938.
25. L. M. Matz, H. M. Dion and H. H. Hill, J. Chromatogr., A, 2002, 946, 59.
26. S. J. Valentine, X. Liu, M. Plasencia, A. E. Hilderbrand, R. Kurulugama, S. L. Koeniger and D. E. Clemmer, Expert Rev. Proteomics, 2005, 2, 553.
27. N. E. Bradbury and R. A. Nielsen, Phys. Rev., 1936, 49, 388.
28. J. P. Williams, V. J. Patel, R. Holland and J. H. Scrivens, Rapid Commun. Mass Spectrom., 2006, 20, 1447.

---