There has been an explosion in the use of liquid chromatography/tandem mass spectrometry as a universal analytical endpoint for drug metabolism and pharmacokinetics (DMPK) applications in recent years, e.g. pharmacokinetic (PK) analysis, intrinsic clearance, CYP inhibition/induction, PAMPA and Caco 2 permeability, metabolite identification, plasma stability studies, etc. Although the technique is extremely sensitive and robust there is potential for ion suppression,1–5 especially using electrospray ionisation (ESI), which can give rise to incorrect data interpretation.6, 7 This effect can be enhanced by the use of ‘ultra-fast’ gradients, which allow many analytes to co-elute. In some cases, it may not be obvious that differing ion suppression may be occurring during an analytical run, further complicating the quantitative results. This is potentially exacerbated due to the variety of matrices used during analysis, which may give differing ion suppression depending on the application. This is further complicated when used in a discovery environment, where potentially hundreds of molecules with differing physicochemical properties (pKa, logD7.4) are analysed. These compounds may be differentially ion suppressed depending on their elution on a typical LC gradient, compared with the ion-suppressing agent, and their ability to compete with charge from the suppressing agent.
The studies described here concentrate on a particular known ion-suppressing agent (polyethylene glycol – PEG 400). This agent is often present in large quantities in DMPK samples, either from deliberate addition as an excipient in a dosing solution or as a contaminant from plastic-ware used in sample preparation. Ion suppression by PEG is well documented and strategies to minimise the effects have been published.8, 9 Identification of ion suppression due to PEG and effective removal of ion suppression are discussed along with the impact of PEG effects on typical PK parameters.
Chemicals and reagents
Chemicals were purchased from Sigma Chemical (Poole, Dorset, UK). All new chemical entities (NCEs) were synthesised at AstraZeneca R&D Charnwood. Sprague-Dawley rat plasma was bought from Harlan UK (Leicester, UK) or prepared in-house.
Blood collection tubes were sourced from Sarstedt (Leicester, UK – EDTA microvette CB300 K2E, EDTA S-monovette K3E 2.7 mL) and Teklab (Durham, UK – EDTA K10V9PP 0.5 mL).
Mobile phase – Solvent A (0.1% formic acid in water), solvent B (0.1% formic acid in MeOH) or solvent C (0.1% ammonium hydroxide in water), solvent D (0.1% ammonium hydroxide in MeOH). A volume of 5–20 µL of sample was injected onto a Waters Xterra C18 column (2.5–3.5 µm particle size; 3.9–4.6 i.d.; 2.5 or 5 cm). Various gradients were used depending on the particular application. A typical gradient is given below. 5% B/D (0–0.5 min), 5%–100% B/D (0.5–1.5 min), 100% B/D (1.9 min), 5% B/D (2 min). The stop time was 2.5 min, the flow rate was 1.5 mL/min, and the column temperature was 50°C.
High-performance liquid chromatography (HPLC) was performed with a Waters Alliance 2790 coupled to a triple quadrupole Quattro Ultima (Waters) operating in positive ESI mode, with Masslynx 3.5 running in multiple reaction monitoring (MRM) mode (MRM – dwell time = 0.2 s). The source parameters were as follows: capillary 3.5 kV; cone – variable depending on analyte; source temperature 120°C; desolvation temperature 350°C; desolvation gas flow 750 L/h; cone gas flow 50 L/h. The HPLC flow was split 4:1 (1.2 mL/min to waste, 0.3 mL/min into the ESI source).
Preparation of plasma samples – PEG 400 investigations
1Identification of the PEG retention time on a typical analytical gradient. Plasma samples (50 µL) from a rat intravenous (IV) pharmacokinetic (PK) study (compound dosed at 1 mL/kg in PEG 400/DMA/water 2:2:1) were protein precipitated with methanol (150 µL). The samples were left at −20°C for 2 h, shaken, centrifuged at 3500 rpm for 15 min, and transferred into vials.The samples were injected onto an Xterra C18 cartridge (3.9 µm × 20 mm) with a rapid 2 min gradient (95% aq-0% aq). The Quattro Ultima was run in full-scan mode and the relevant PEG ion chromatograms were extracted and the peaks integrated. The disappearance of PEG with time and the retention times (RTs) of the different oligomers of PEG were determined relative to the analyte.
2Blank rat plasma (Sprague-Dawley, Harlan) was spiked with typical dosing vehicles used in a PK study, e.g. saline, 6% glucose, DMA/propylene glycol/water (2:2:1) and DMA/PEG 400/water (2:2:1), all at 1% v/v final. This composition (1% dosing vehicle in plasma) was chosen as an approximate maximal plasma composition of dosing vehicle that could be theoretically achieved immediately following an IV dose of 1 mL/kg, assuming limited distribution at t = 0.Ten basic project compounds were spiked into the above aliquots of plasma (50 µL) and precipitated with 150 µL MeOH, to give a final concentration of 2000 ng/mL. The samples were prepared and analysed as described above. The peak areas from the different excipient experiments were expressed as percentage of the saline control.
3HPLC gradient modifications were investigated to minimise co-elution of PEG with analytes. Blank rat plasma (Sprague-Dawley, Harlan) was spiked with 0.3% PEG 400 and a typical project compound. The gradient was modified and the PEG ([M+H]+ = 415, 459, 503, 591, etc.) and compound ([M+H]+ = 331) masses were extracted from the total ion chromatogram (TIC) to check for co-elution.
4Ten AstraZeneca research compounds, comprising of acids, bases and neutrals with a wide range of lipophilicities (clogP 1–5), were spiked at 100 ng/mL into 50 µL of (a) water/MeOH 1:3; (b) plasma (Harlan)/MeOH 1:3; (c) plasma prepared from fresh rat blood in Sarstedt EDTA monovette tubes/MeOH 1:3; (d) plasma prepared from fresh rat blood in Sarstedt EDTA microvette tubes/MeOH 1:3. The sample preparation was the same as described above.
Preparation of plasma samples from PK studies
MeOH with internal standard (150 µL) was added to rat plasma (50 µL) from pre-clinical rat PK studies. The samples were vortexed thoroughly, frozen at −20°C for 2 h, centrifuged at 3500 rpm for 15 min, and the supernatants transferred to microtitre plates. Standards and quality control samples were prepared by spiking 10 µL of a suitable methanolic standard into 50 µL of rat plasma with 140 µL of MeOH/internal standard. The samples were analysed as described above.
Preparation of plasma samples – PEG investigations
From the plasma samples from a rat IV study, in which PEG 400 was part of the dosing vehicle, PEG 400 was found to elute over a large retention time (0.4 min wide) of the LC gradient. Presumably, this poor chromatography was due to overloading of the stationary phase by PEG. The RTs of the different oligomers of PEG, e.g. [414+H]+, [458+H]+, etc., increased with molecular weight, thus enhancing the overall spread of PEG on the LC gradient (data not shown). In fact, the PEG was eluted over a range of 40% of the total of the run time (Fig. 1). In this case there is a high chance that the analyte may elute with PEG and be subjected to ion suppression.
The IV and oral (PO) concentration/time profile of PEG (oligomer [M+H]+ = 503) is given in Fig. 2. The oral bioavailability of PEG 400 was measured as 42% with a rapid terminal half-life of approximately 2 h, which is consistent with results reported by He et al.10
Ion suppression was seen for the basic project compounds spiked into plasma containing 1% DMA/PEG 400/water (Fig. 3). All other dosing vehicles gave no appreciable suppression (the LC/MS/MS response is inherently variable by ±3–5% in our hands for protein-precipitated samples). However, one analyte gave ion enhancement by ∼15% in the presence of 1% glucose. The reason for this apparent single enhancement is unclear, since glucose is known to elute near the solvent front and well separated from analyte.
The degree of suppression was variable, with one compound showing 88% suppression. The degree of suppression did not correlate with the physicochemical properties of the basic compounds studied, e.g. pKa, log P (data not shown).
The gradient was altered to retard elution of PEG relative to the analyte peak. A good compromise was to load in 25% B/75% A. This was held for 0.5 min, before a linear gradient from 0.5 min to 1.5 min (25% B to 100% B) was used. This was held for 0.5 min, before re-equilibrating back to starting conditions over 0.5 min. Under these modified conditions, the majority of the lower molecular weight PEG oligomers were eluted before the analyte and only oligomers with mass >750 co-eluted with the analyte (Fig. 4). For this example, the ion suppression was reduced by approximately 90% relative to the standard gradient (data not shown). Note: using our standard LC method (5% B loading), all PEG peaks co-eluted with analyte.
Potential ion suppression due to effects from blood collection tubes was investigated as follows. The MS responses from ten AstraZeneca compounds covering acid, basic and neutral structures were compared in a variety of blood collection tubes using water/MeOH as a control. Control plasma obtained from Harlan and plasma prepared from fresh rat blood in 2.7 mL EDTA monovette tubes gave similar responses to control in all cases (data not shown). However, using plasma prepared from fresh rat blood in 1 mL EDTA microvette tubes, the ion suppression was significant and variable for most analytes, with one analyte giving a higher response compared with control (Fig. 5).
To track down the source of the ion suppression, a full-scan acquisition of the plasma samples from the microvette tubes revealed large peaks due to PEG. These peaks were absent in the water/MeOH controls, Harlan plasma and in-house rat plasma from 2.7 mL monovette EDTA tubes. Following the addition of 200 µL of blank plasma to the microvette tubes and LC/MS/MS analysis using PEG 400 in blank plasma as a standard curve, the PEG concentrations were estimated at ∼200 µg/mL.
A test research compound with a basic centre was run on our original generic gradient for PK analysis. Following acquisition by full scan, the analyte co-eluted with PEG (leached from blood collection microvette tubes). However, changing the aqueous and organic mobile phase additive from formic acid to ammonium hydroxide led to an increased retention time for the analyte whilst not affecting the neutral PEG peaks (Fig. 6).
Re-running the same samples using the basic mobile phase gave different PK parameters, resulting in a lower plasma clearance, smaller volume of distribution and, as anticipated, identical terminal half-life and bioavailability (Fig. 7).
Ion suppression effects in LC/MS/(MS) have been the subject of numerous publications in recent years.1–9 This has partly been due to the expanding use of this technique in the analysis of pharmaceutical, trace product and environmental samples. Ion suppression is known to have the effect of reducing overall sensitivity as well as giving rise to flawed analysis due to potentially differential suppression depending on matrix. The mechanism of ion suppression has been proposed to be due to the following:
1Competition for charge between analyte and ion-suppressing agent leading to an overall reduced ionisation of analyte.
2Large concentrations of ion-suppressing agent leading to increases in surface tension and viscosity of the droplets formed in atmospheric pressure chemical ionisation (APCI), ultimately leading to decreasing evaporation efficiency. This could certainly be a possibility for PEG, since it is often present in large concentrations and may well affect nebulisation and droplet formation due to changes in surface tension and conductivity of the liquid phase.
3Gas-phase reactions between analyte ions and other sample components leading to an overall loss in charge from the analyte ions.
However, the latter reason has been disputed5 as the authors argue that gas-phase species should be present in APCI and ESI and the latter often shows more pronounced ion suppression. The authors suggested that changes to the solution properties due to high concentration of non-volatile solutes probably controlled ion suppression. The underlying reason why non-volatile species can give rise to ion suppression has not been clearly demonstrated to date, but possibilities include precipitation of analyte with the non-volatile species, which could limit transfer from the liquid to gas phase,2 or that the presence of non-volatile species changes the spray droplet solution properties.1, 5
The present study focused on the ion suppression caused by polyethylene glycol (PEG 400) in pharmacokinetic (PK) analysis of pre-clinical studies, which was either deliberately included in the dose formulation or extracted as part of sample workup from plastic-ware.
With regard to the use of PEG as part of a dosing vehicle, it was important to assess the concentration present in plasma samples from rats dosed either IV or PO in a typical protocol and whether the PEG interfered with the analytes of interest during LC/MS/MS analysis. The PEG PK profile was monitored following IV and PO administration of a typical NCE dosed in a PEG 400 formulation. For the compound studied, the initial PEG plasma concentrations following an IV dose were very high, but rapidly declined with a terminal half-life of 2 h. Following oral administration the Tmax was at ∼30 min with a rapid terminal half-life of 2 h. As initially the PEG concentrations are vastly in excess of analyte concentrations, one may assume that ion suppression could occur. In fact, the PEG concentrations are so high (100–1000 µg/mL) that overloading of the column stationary phase occurs, leading to the PEG eluting over a wide proportion of the chromatogram. In addition the retention time of the PEG oligomers increases with molecular weight, spreading out the elution even further. It was shown that, for the majority of analytes studied, co-elution occurred between PEG and analytes.
In order to investigate whether this co-elution could cause ion suppression, an equivalent in vivo PEG concentration was spiked into blank rat plasma along with ten project compounds. Ion suppression was observed for most analytes, which were subsequently shown to co-elute with PEG. In addition, the degree of ion suppression was variable depending on analyte (Fig. 5).
This unpredictable variability coupled with the ever-changing plasma concentrations of PEG (due to excretion) and subsequent reduction of ion suppression with time could potentially lead to incorrectly determined PK parameters including in vitro/in vivo scaling. This has indeed been shown to be the case with the effect of PEG in dosing vehicles.9 In this report, the authors had noted that clearance and volume of distribution could be overestimated by as much as six- and four-fold, respectively, due to ion suppression by PEG.
Attempts were made to modify the generic rapid LC gradient to shift the elution of PEG away from analytes. Partial success was achieved (Fig. 4), in which the bulk of the PEG oligomers eluted before 1 min, with the analytes generally eluting after 1.5 min. An alternative approach to eliminate PEG from plasma samples could be the use of strong anion/cation solid-phase extraction (SPE) for acid/basic chemistry prior to analysis.
Following the discovery of PEG contamination of blood from the blood collection tubes used for pre-clinical sampling, the impact of this potential ion suppression was investigated. A similar effect was seen to the investigations described earlier with PEG 400 in the dosing vehicle. Analyte response was suppressed to varying degrees up to a maximum of 88%. Clearly, with this example, incorrectly determined PK parameters could be obtained, which would be enhanced further if the standards and quality controls were prepared using plasma from an alternative source which may not have been contaminated with PEG to a similar extent.
To test this theory in practice, a test research compound that had been dosed to rats IV and PO and the blood collected in the PEG-contaminated blood tubes was investigated. The resultant PK parameters were rather unexpected and, at first, puzzling. Unusually, the compound had a high in vivo clearance that equated to a typical rat liver blood flow (70 mL/min/kg), which would not have been predicted from the in vitro rat hepatocyte screen. The volume of distribution was also higher than anticipated and, strangely, the bioavailability was good at 67%, which was not in keeping with a compound that should have undergone a near complete ‘first-pass effect’ by the liver following oral dosing.
Following some investigational LC/MS/MS work it was found that the analyte directly co-eluted with the PEG leached from the blood tubes. As the analyte was a basic compound (pKa ∼ 9), the mobile phase was changed from acidic to basic (pH 10.5) which had the effect of dramatically increasing the retention time of the analyte. The PEG peaks were unaffected by the mobile phase change due to their neutral nature. The result was chromatographic separation between analyte and PEG.
Following reanalysis of the samples by the modified HPLC method, the analyte plasma concentrations were determined to be much higher than the original analysis. This resulted in certain PK parameters changing from the original analysis.
Clearance was reduced from 70 to 14 mL/min/kg, which was in line with the value expected from the in vitro analysis. The volume of distribution was reduced from 23 to 10 L/kg, with the terminal half-life unchanged as expected. The bioavailability was also similar to the previous analysis as expected (similar effects on both IV and PO samples) and now was understandable due to the reduced clearance and lower first-pass effect. This effect of ion suppression from PEG in blood collection tubes culminated in similarly distorted PK parameters to those described previously with ion suppression due to PEG in the dosing vehicle.7, 9 Following this discovery, alternative blood collection tubes have been sourced which are found not to contain PEG and these are now used exclusively for rodent PK studies. In order to minimise ion suppression further, we avoid PEG 400 in dosing vehicles wherever possible, relying on alternative excipients, e.g. methyl cellulose, cyclodextran, etc. However, to ensure that these excipients do not cause ion suppression, two test solutions (±excipient) at nominal concentration are prepared in rat plasma, and following sample workup the two samples are injected onto the analytical HPLC gradient. The MS/MS peak responses are compared to check for ion suppression. If appreciable, then HPLC method development is applied until the analytes are chromatographically separated from the ion-suppressing agent.
The above rat pharmacokinetic example clearly demonstrates that the power of chromatographic resolution of analytes from PEG (or other ion-suppressing agents) should not be underestimated. Understanding the physicochemical properties of any particular analyte versus the analytical conditions used is essential to decide the best course of action for resolving analytes/ion-suppressing agent. Clearly, it is simpler to resolve the two via simple pH changes to the mobile phase when the analyte is either basic or acidic assuming the ion-suppressing agent is a neutral molecule. Ion suppression issues with neutral analytes can also be achieved relatively simply with changes in column chemistry, mobile phase changes or gradient modifications.
Ion suppression is an important consideration in DMPK analysis and should not be over-looked. It can lead to a reduction in analyte response and hence a higher limit of quatitation (LOQ), or more worryingly incorrect assignment and interpretation of PK parameters. We have shown that the source and degree of ion suppression due to PEG can be variable depending on the NCE analysed and the dosing vehicle/apparatus used during sample workup at the time. Although there have been accounts of PEG removal from biological samples,7, 9 at present there is no universal generic method of removing PEG from biological samples in a discovery environment, due to the range of NCE chemistries and their associated charge and polarities encountered on a daily basis.
However, the source and identification of ion suppression is crucial in order to address the phenomenon in each case. Relatively simple chromatographic modifications including changing the pH of the mobile phase or changes to overall best practise (e.g. alternative dosing vehicles, changes in blood collection tubes) can be applied to overcome this issue. It should be stressed that understanding the chemistry and chromatographic properties of the analytes is imperative, in order that appropriate chromatographic changes can be implemented to resolve them from PEG. This approach should be applied on a case-by-case basis, as it is unlikely that a generic solution can be found for the range of chemistries encountered in a typical research programme.
The authors would like to thank Iain Beattie, Andy Wright, Peter Littlewood, David Ryan, Charles O'Donnell and Jas Singh for their contribution to the work.