Max Hecht (Presenter)
University of Tartu
Bio: Currently, I am PhD student in the field of analytical chemistry at the University of Tartu in Estonia. My research focused on direct ionization techniques and how to make them work in a clinical setup. I came to Tartu for a Master program called Applied Measurement Science in 2014 and decided to stay for the PhD, which offered me the opportunity I was looking for: developing hardware for mass spectrometer to be used for rapid analysis of clinical samples. I learned a lot about MS during my 4 years working as a student in the clinical chemistry lab in Leipzig, Germany. Alongside that, I was studying biochemistry. Now I am eager to share my findings about sponge spray ionization that I have developed and introduced.
Authorship: Max Hecht(1), Hanno Evard(1,2), Kalev Takkis(1), Rūta Veigure(1), Rudolf Aro(1), Rünno Lõhmus(3), Koit Herodes(1), Ivo Leito(1), Karin Kipper(1,4)
(1) University of Tartu, Institute of Chemistry, 14a Ravila Street, 50411 Tartu, Estonia (2) Faculty of Pharmacy, Division of Pharmaceutical Chemistry and Technology, University of Helsinki, Viikinkaari 5, 00790 Helsinki, Finland (3) University of Tartu, Institute of Physics, W. Ostwaldi Street 1, 50411, Tartu, Estonia (4) Analytical Services International, St George’s University of London, Cranmer Terrace, London, SW17 0RE, United Kingdom
Sample preparation for the analysis of clinical samples with the mass spectrometer (MS) can be extensive and expensive. With the novel sampling and MS analysis technique we call sponge spray, we can circumvent this and directly analyse blood, plasma and urine. A volumetric absorption microsampling (Neoteryx Mitra) device is used to take up an exact amount of sample and from that same tip an electrospray can be directed into a mass spectrometer. We demonstrated that although with significant matrix effects, quantitation of penicillin G, a common antimicrobial, is possible in plasma and urine.
Paper spray ionization (PSI) for mass spectrometry (MS), in the context of biological sample analysis, first debuted as a sample introduction/ionization technique for the analysis of dried blood spots (DBS) . Since then, PSI has been applied to a range of bioanalyses [2-5], although to date none have resulted in widespread adoption in clinical laboratories.
The primary limitations of quantitative analysis using PSI are (i) poor spray reproducibility, caused in part by the non-uniform natural cellulose fibre composition of paper [6,7], and (ii) interferences caused by the biological sample matrix  deposited on the paper. Robust, reliable quantitation is essential for the majority of clinical diagnostic tests or in therapeutic drug monitoring where interpretation of results is carried out against reference concentration ranges. The interferences caused by matrix components are inherent to direct ionization techniques, as any kind of clean up or separation is missing. The most common solution for overcoming the interferences is by using isotopically labelled internal standard (ILIS). ILIS is essential for achieving reliable quantification and ideally it should be pre-mixed with the liquid biological sample prior to spotting and drying . Alternatively, quantitative methods have been reported in which the ILIS was successfully added to the paper and allowed to dry, followed by volumetric addition of liquid biological samples by precise pipetting . In both cases, the requirement to collect a liquid sample, or pipette the sample onto the paper offsets the initial simplicity of the PSI approach.
Similarly, DBS are well-established for qualitative applications such as new-born screening [10,11] due to the less-invasive nature and simplicity of sampling, and the ease of transporting dried specimens. Typical, non-PSI methods for the analysis of DBSs involve extracting analytes from the DBS cards (e.g. by punching discs from the DBS cards), and subsequent analysis of the extracts by flow-injection or liquid chromatography-tandem mass spectrometry. However, it is well documented that quantitative DBS analysis, whether using PSI or more traditional analytical methods, is limited by the between-sample haematocrit bias  which significantly affects the amount of blood collected for a given DBS punch size.
A promising solution to the haematocrit bias observed for DBS collection onto cards was developed and introduced in 2014[12-15] MitraTM (Neoteryx, Torrance, CA) is a volumetric absorptive micro-sampling (VAMS) device. VAMS device absorbs accurately and precisely (coefficient of variation 4 % for 10 µL blood) a fixed volume of blood following a fingertip lancet stab  without the need of a pipette or a syringe. Although a skin puncture is still required, VAMS eliminates most of the remaining disadvantages associated with dried blood spot sampling. Primarily, VAMS has been shown to overcome the issues associated with variations in blood haematocrit [13-15]. The Mitra VAMS devices are composed from a macroporous, highly hydrophilic material, hereafter referred to as the ‘sponge’.
Antimicrobials such as penicillin G form an essential part of the treatment of critically ill patients, yet there is often very limited evidence to support the dose regimens used routinely in intensive care units (ICU) . Identifying the optimum dose is key to improving therapeutic outcomes in ICU and to reduce toxicity . Since the antibiotics used in ICUs are predominantly penicillin derivatives such as beta-lactams, they are also amongst the ones most affected by antimicrobial resistance . The development of a quick beta-lactam antimicrobial therapeutic drug monitoring has been proven to be beneficial to avoid under- and overdosing [19,20]. We have developed and tested a direct ionization MS technique from VAMS devices (‘sponges’) using penicillin G as a model compound.
Mitra™ microsampling devices (10 µL and 20 µL) were from Neoteryx (Torrance, CA, USA). Penicillin G and penicillin G-D7, LC-MS grade acetonitrile (MeCN), isopropanol (iPrOH), ammonium acetate, 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) and formic acid (FA) were from Sigma Aldrich (Missouri, USA). Water was purified [18.2 MΩ·cm at 25 °C and a total organic carbon value 2 – 3 ppb] in-house using a Millipore Advantage A10 system from Millipore (Bedford, USA). Blood, plasma and urine were donated by healthy volunteers and/or purchased from the Blood Bank of Tartu University Hospital (Tartu, Estonia).
SSI was carried out using an MSD Trap XCT MS system (Agilent Technologies, Santa-Clara, CA, USA). The ionisation source setup consisted of an Agilent APCI corona needle, mounted on top of an XYZ 3-Axis Trimming Platform Linear Stages Bearing Tuning Sliding Table 60 × 60 mm micromanipulator, positioning the sponge 23.0 mm from the MS inlet shield centre. Both items were attached to the instrument with an open frame (built in-house). The APCI corona needle, set to maximally provide 4 mA, functioned as a high voltage supply for the VAMS device used to generate SSI. A shortened disposable pipette tip was inserted into the rear of the sponge, coupled using PEEK tubing (1/16 in. × 0.005 in, O.D. x I.D.) to carry eluent to the tip. No nebulising gas was used and the MS parameters were optimised for m/z 335→160 and 342→160 for penicillin G and penicillin G-D7 respectively. A Digital Microscope B006 of Genesys Logic (New Taipei City, Taiwan) was used to visualise the sponge spray. Blood, plasma and urine samples were fortified with penicillin G from 0.5 to 100 mg/L. Selected concentration range reflects penicillin G concentrations in the real samples to treat different infections caused by various bacterial species . Penicillin G-D7 (12.5 mg/L in methanol) was used as the internal standard (IS), and was pre-deposited on the VAMS devices and left to dry for 2 h. VAMS were then used to collect the biological samples (10 µL and 20 µL) and were allowed to dry (ambient temperature: 20 ± 3 °C, 2 h). Data analysis was carried out using Data Analysis for LC/MSD Trap Version 3.2 (Bruker Daltonik GmbH) and Microsoft Excel.
Sponge spray has the potential to be used on various sample matrices. In the current study standard solutions, plasma and urine samples were investigated. Plasma was used as the primary matrix for the evaluation of quantitation of penicillin G. Whole blood samples were also tested using the approach, and produced chronograms similar to those from plasma samples. It was observed that the optimal eluent composition for SSI was matrix-dependent. Of the various eluent compositions tested, iPrOH gave the best response as the organic component of the eluent. A solution of 50 % (v/v) iPrOH in deionised water containing 0.1 % (v/v) HFIP, 0.1 % (v/v) FA and 1 mM ammonium acetate was suitable to generate SSI for standard solutions and urine, but no spray was observed using this solvent composition for plasma or whole blood samples. For plasma and for whole blood, greater organic content was required to generate a stable spray [75 % and 90 % (v/v) iPrOH for plasma and whole blood, respectively]. These latter eluent compositions were also found to be suitable for ionisation of standard solutions and urine samples.
The time necessary to form a spray was found to be dependent on (i) the size of the VAMS device (10 or 20 µL) and (ii) the flow-rate of the eluent. When eluent was applied, it was observed that the devices first became saturated (observed as a darkening of the colour of the device) prior to generation of a Taylor cone and successful spray from the excess solvent which formed a layer on the outer surface of the VAMS devices. Spray generated from a 10 µL VAMS devices with an eluent flow-rate of 25 µL/min, which resulted in sponge spray being formed within 1 min. For a 20 µL device with a flow rate of 20 µL/min, spray formation took approximately 3 min. Due to the eluent evaporating in the hot drying gas stream from the MS more eluent is needed, than the VAMS can take up.
Only when excess solvent forms a layer on the outside surface of the tip an electrospray can be achieved. The surface tension for the iPrOH (18.69 mN/m at 50 °C)  is low enough for maintaining a stable sponge spray, while the eluent flow-rate had to be sufficient to maintain the solvent excess in the presence of the hot drying gas stream. For methanol, the surface tension (20.21 mN/m at 50 °C)  and lower boiling point meant that solvent evaporation was too fast to maintain the solvent layer without excessive flow-rates. For iPrOH, minimum flow-rates of 10 and 15 µL/min were required to initially generate SSI from 10 and 20 µL VAMS devices, respectively.
Once an initial spray had been generated (after saturation of the devices), it was observed that the flow-rate could be reduced by approximately 5 µL/min whilst still maintaining a stable spray. There are benefits to reducing the flow-rate as far as possible (e.g. to form a nano-spray) in terms of sensitivity and reduction of matrix effects . However, such low flow-rates dramatically increase the total analysis time, and the rate of elution of analytes from the VAMS devices. Therefore, for the investigation of quantitation of penicillin G in plasma using SSI, a flow-rate of 25 µL/min was used to provide a reasonable total analysis time whilst maintaining signal quality.
Direct analysis is inherently rapid and simple, since sample preparation is negated, and the instrumentation is less complex (no requirement for an LC, for example). However, the presence of matrix components can dramatically affect signal intensity. In this study, a 40-fold decrease in signal for penicillin G was observed when comparing the SSI response for standard solutions with those from fortified urine and plasma samples. Additionally, the time profile for elution of the analyte was considerably different between the standard solution and the biological matrices. For standard solution, analyte signal was observed as soon as the spray was formed (after 3 min), and elution of the analyte from the VAMS device was complete after approximately 10 min. For urine samples, the elution profile was similar to that of standard solutions (i.e. elution was complete after approximately 10 min), except that there was significant signal suppression observed when the spray was first formed, presumably due to co-elution of urine salts from the VAMS device. For dried plasma samples, the penicillin G signal was initially low, presumably due to similar ion suppression effects as observed for urine samples. However, the signal then increased gradually over the remaining 30 min of data acquisition, and did not return to a baseline level. This suggested that even after 30 min of SSI, not all the penicillin G had been eluted from the VAMS device. To evaluate whether the protein content of dried plasma was the cause of this analyte retention, a portion of plasma containing penicillin G was precipitated using MeCN, and the supernatant applied to a VAMS device. The observed chronogram showed a similar signal intensity and elution profile to that of the ‘neat’ dried plasma sample, suggesting proteins did not interfere with the SSI process. Another possible cause of the observed retention of penicillin G in dried plasma samples is the high phospholipid content. It may be the case that phospholipids form a layer which covers the surface of the VAMS polymer when the sample is collected, therefore slowing the dissolution process when the eluent is applied to the device.
A window of each chronogram was selected to integrate the SSI signal for quantitation. For urine samples, the integration window was selected to begin after 7.5 min, and to last for 2.5 min. For plasma samples, the integration window was selected to begin after 10 min, and to last for 10 min, thus avoiding the region of the chronogram where (i) the analyte signal was lower and (ii) there were likely greater ion suppression effects from co-eluting matrix components. By starting the integration window later, this noise can be avoided. The same time window was used for integration of the penicillin G-D7 extracted chronograms, which displayed similar elution profiles to penicillin G in all cases. For urine and for plasma samples despite incomplete and variable analyte elution, linear calibration was achieved over the concentration range selected (0.5 – 100 mg/L for both matrices, r² = 0.991 and 0.986 for urine and plasma, respectively.
Precision (% RSD) of the penicillin G-D7 signal for dried plasma samples was used to assess inter-device SSI reproducibility. The IS solution (12.5 mg/L) was added to each VAMS device and allowed to dry prior to adding the fortified matrix sample. The chronogram areas were integrated for penicillin G-D7 for 19 successive VAMS devices subjected to SSI (total 6 h analysis time). The RSD of the signals integrated from the chronogram windows was 13.9 %. This was comparable with previously reported PSI results 6, and falls within the requirements for bioanalytical method validation guidelines for precision of 15 % .
Conclusions & Discussion
In this proof-of-concept study, we report ‘sponge-spray’ ionisation for the first time, and demonstrate that ionisation of a drug can be achieved in different matrices directly from a VAMS device. In a simple ‘collect-and-spray’ approach, penicillin G was determined at clinically relevant concentrations in biological matrices. Furthermore, the initial data presented in this study suggest that quantitation is possible when an isotopically-labelled IS is pre-deposited onto the collection devices. The advantages of VAMS devices, especially for quantitative applications, are extremely significant. The volumetric sample collection allows consistent and reproducible IS pre-deposition prior to sampling, and reliable sampling of the biological matrix, including overcoming the haematocrit error for whole blood sampling, when the sample is subsequently collected.
Thus, even in its current "proof of principle" state SSI has demonstrated clear promise. First of all, the volumetric aspect of the VAMS is already a big advantage over paper spray. Furthermore, already at this stage its precision is comparable to PSI. Thus, after careful optimization of the methodology it has a potential to become the method of choice for biological sample collection and following direct quantitative SSI-MS analysis.
References & Acknowledgements:
We greatly appreciate Dr. Lewis Couchman’s input and help. This work was supported by the institutional research grant of Ministry of Education and Research of Estonia IUT20-14 (TLOKT14014I) and IUT2-25.
(1) Wang, H.; Liu, J.; Cooks, G. R.; Ouyang, Z. Angewandte Chemie 2010, 49, 877–880.
(2) Manicke, N. E.; Bills, B. J.; Zhang, C. Bioanalysis 2016, 8, 589-606.
(3) Shi, R. Z.; Gierari, T. M. E.; Manicke, N. E.; Faix, J. D. Clin Chim Acta 2015, 441, 99-104.
(4) Espy, R. D.; Manicke, N. E.; Ouyang, Z.; Cooks, G. R. Analyst 2012, 137, 2344–2349
(5) Zhang, Z.; Xu, W.; Manicke, N. E.; Cooks, G.; Ouyang, Z. Analytical Chemistry 2012, 84, 931-938.
(6) Manicke, N. E.; Abu-Rabie, P.; Spooner, N.; Ouyang, Z.; Cooks, R. G. Journal of The American Society for Mass Spectrometry 2011, 22, 1501–1507.
(7) Vega, C.; Spence, C.; Zhang, C.; Bills, B. J.; Manicke, N. E. Journal of The American Society for Mass Spectrometry 2016, 27, 726-734.
(8) Evard, H.; Kruve, A.; Lõhmus, R.; Leito, I. Journal of Food Composition and Analysis 2015, 41, 221–225.
(9) Liu, J.; Wang, H.; Manicke, N. E.; Lin, J.-M.; Cooks, R. G.; Ouyang, Z. Analytical Chemistry 2010, 82, 2463–2471.
(10) Pitt, J. J. Clin Biochem Rev 2010, 31, 57-68.
(11) Lehmann, S.; Delaby, C.; Vialaret, J.; Ducos, J.; Hirtz, C. Clin Chem Lab Med 2013, 51, 1897-1909.
(12) Kesel, P. M. M. D.; Sadones, N.; Capiau, S.; Lambert, W. E.; Stove, C. P. Bioanalysis 2013, 5, 2023-2041.
(13) Spooner, N.; Denniff, P.; Michielsen, L.; Vries, R. D.; Ji, Q. C.; Arnold, M. E.; Woods, K.; Woolf, E. J.; Xu, Y.; Boutet, V.; Zane, P.; Kushon, S.; Rudge, J. B. Bioanalysis 2015, 7, 653-659.
(14) Denniff, P.; Spooner, N. Analytical Chemistry 2014, 86, 8489-8495.
(15) Kesel, P. M. D.; Lambert, W. E.; Stove, C. P. Analytica Chimica Acta 2015, 881, 65–73.
(16) Roberts, J.; Abdul-Aziz, M.; Lipman, J.; Mouton, J.; Vinks, A.; Felton, T.; al, e. The Lancet Inf Dis 2014, 14, 498–509.
(17) Spyridis, N.; Sharland, M. Arch Dis Child 2008, 93, 909-910.
(18) (EARS-Net), A. R. o. t. E. A. R. S. N., Antimicrobial resistance surveillance in Europe 2014; European Centre for Disease Prevention and Control2014.
(19) Roberts, J.; Ulldemolins, M.; Roberts, M.; McWhinney, B.; Ungerer, J.; Paterson, D.; Lipman, J. Int J Antimicrob Agents 2010, 36, 332-339.
(20) Patel, B.; Paratz, J.; See, N.; Muller, M.; Rudd, M.; Paterson, D.; Briscoe, S.; Ungerer, J.; McWhinney, B.; Lipman, J.; Roberts, J. Ther Drug Monit 2012, 34, 160-164.
(21) Gill, V. J.; Manning, C. B.; Ingalls, C. M. Journal of Clinical Microbiology 1981, 14, 437-440.
(23) Vazquez, G.; Alvarez, E.; Navaza, J. M. J. Chem. Eng. Data 1995, 40, 611-614.
(24) El-Faramawy, A.; Siu, K. W. M.; Thomson, B. A. Journal of the American Society for Mass Spectrometry 2005, 16, 1702-1707.
(25) EMA. 2011.
(26) Narayanan, R.; Sarkar, D.; Cooks, R. G.; Pradeep, T. Angewandte Chemie 2014, 53, 5936–5940
IP Royalty: no
|Planning to mention or discuss specific products or technology of the company(ies) listed above:||