Abstract
Background
Current gold standard markers for myocardial damage are troponins I and T, which are both sensitive and specific for the detection of myocardial infarction, but require up to 6 h to become reliably elevated in serum. Investigation into markers with potential to identify patients with early ischaemic changes is therefore intense. Choline is reported to be prognostic in patients presenting with acute coronary syndromes via its release from ischaemic cell membranes.
Methods
Liquid chromatography tandem mass spectrometry was used to develop a method to quantitate choline in plasma and blood. The method involves addition of a deuterated internal standard to an aliquot of plasma or blood followed by organic solvent addition, which precipitates the proteins in the sample. Preparation was carried out directly into a 96-deep-well plate. Chromatography of choline used a strong cation exchange column and separation used a Waters Atlantis dC18 analytical column positioned directly before the mass spectrometer source, allowing on-line preanalytical clean up of the sample.
Results
The lower limit of quantitation was 0.38 μmol/L, linearity was observed up to 754 μmol/L, with a working concentration range of 0.38–224 μmol/L, inter- and intra-assay coefficients of variation were <6% and <4%, respectively. Samples were stable throughout five freeze–thaw cycles and recovery was between 94% and 114%.
Conclusions
The assay was successfully validated in accordance with FDA guidelines and is suitable for quantitation of choline in research and clinical settings.
Introduction
The redefinition of myocardial infarction (MI) by the European Society of Cardiology/American College of Cardiology 1 in 2003 placed a greater emphasis on the use of biochemical markers troponin T and troponin I in the diagnosis of MI in patients presenting with symptoms consistent with acute coronary syndromes. However, troponins are only reliably raised after around 6 h, 2 with many hospitals in the UK choosing one measurement at 12 h as their protocol for investigating possible MI because although serial measurements can be helpful in diagnosis, the troponin assay is relatively expensive. Therefore, interest in diagnostic and prognostic markers, which could allow more rapid and effective intervention in patients with MI and earlier discharge of those without, is intense. 3
A number of different markers have been suggested for the investigation of differing aetiologies of the vascular disease leading to MI, which include plaque destabilization, plaque rupture, ischaemia and myocardial stretch 4 and one of these is choline. Choline is implicated 4–6 in myocardial ischaemia, along with unbound free fatty acids and ischaemia-modified albumin. Choline should reflect pathology earlier in the disease process than troponins T or I, which only become elevated in the blood after myocyte necrosis has occurred. Whole blood and plasma choline have been reported to reflect different pathological processes, 7 therefore the method developed needed to be able to quantitate both.
There are a number of methods published for the measurement of choline, either alone or in tandem with other related metabolites. Some of these methods, such as enzymatic–radioisotopic assays 8,9 or those involving chromatographic separation of choline either by paper or liquid prior to quantitation, 10,11 are largely of historical interest as they have predominantly been replaced by more modern methods such as tandem mass spectrometry. 12,13 Other methods involve complex extraction or derivatization processes 8–11,14 or suffer from a loss of analytical sensitivity within the time frame needed to analyse a batch, a problem noted in papers published by groups using electrochemical detection. 15,16
A rapid method for the analysis of choline in plasma and whole blood was required, using the liquid chromatography tandem mass spectrometry instrumentation available in our laboratory. Here we present a novel method of quantitating choline in both whole blood and plasma, involving on-line clean up of a protein precipitation preparation. Following full validation according to industry guidelines, 17 as reported below, this method was used to quantitate the choline in the recently published paper, 18 which further discusses the findings and predictive capacity of the choline results in this particular study and implications for future research in the area.
Materials and methods
Internal standards, standards, quality control and patient samples
Deuterated (d9) choline (CDN isotopes, Thaxted, UK) was used as internal standard. Working internal standard concentration was 1.1 mg/L (in water). Standards were prepared by spiking choline (Sigma, Poole, UK) into phosphate buffer solution with bovine serum albumin (0.1% [w/v]) at pH 7.4 from a concentrated stock of choline (10.4 mg/mL in water stored at −30°C) to give concentrations covering the range 0.38–224 μmol/L. Quality control (QC) solutions were prepared in a similar way but using a separate highly concentrated stock. QC was prepared at three concentrations: 1.5, 7.5 and 37.3 μmol/L to cover the expected range of results.
Sample preparation
Standard, QC or sample (20 μL) and internal standard (20 μL) were pipetted directly into a 96-deep-well plate. Sample types were ethylenediaminetetraacetic acid (EDTA) plasma or EDTA anticoagulated whole blood. Plasma required no pretreatment, but an aliquot (200 μL) of whole blood was frozen until solid at −80°C (40 min) before thawing completely. This lysed the red blood cells. Acetonitrile (100 μL) was added to precipitate the proteins. The plate was thermo-sealed, vortex mixed for 30 s and then centrifuged for 5 min at 800
High-performance liquid chromatography
Liquid chromatography was performed using a Waters Alliance 2795 HT Liquid Chromatography instrument (Waters, Manchester, UK). Supernatant (3 μL) was loaded onto a 4.0 × 3.0 mm security guard strong cation exchange column (Phenomenex, Macclesfield, UK). The flow from the guard column was then diverted via a switching valve onto the analytical column. The analytical column, an Atlantis dC18 5 μm 2.1 × 20 mm (Waters, Manchester, UK), was positioned directly before the ion source. The first 0.7 min of eluant was diverted to waste, allowing on-line clean up of the sample.
The mobile phases used were A (deionized water containing 2 mmol/L ammonium acetate with 0.1% [v/v] formic acid), B (methanol containing 2 mmol/L ammonium acetate with 0.1% [v/v] formic acid) and C (deionized water containing 100 mmol/L ammonium acetate and 0.1% [v/v] formic acid). The starting conditions were 85% A, 5% B and 10% C at a flow rate of 0.6 mL/min. At 0.1 min this stepped to 5% A, 5% B and 90% C with a flow rate of 0.8 mL/min, continuing until 0.7 min. Until this point, all eluant was diverted to waste, thereafter eluant was directed into the analytical column. At 0.7 min, the gradient stepped to 85% A, 5% B and 10% C with a flow rate of 0.6 mL/min. This continued until 0.8 min when a linear gradient commenced. The final composition of this gradient was reached at 1.9 min with 50% A, 40% B and 10% C, flow rate 0.6 mL/min. This composition continued until 2.0 min when it stepped back to the starting conditions of 85% A, 5% B and 10% C, maintaining a flow rate of 0.6 mL/min. This gradient is summarized in Table 1.
Mobile phase gradient used during the analysis of choline
Mass spectrometry
A Waters Quattro™ Micro Tandem Mass Spectrometer (Waters, Manchester, UK) with a Z spray ion source was coupled to the liquid chromatography system. The mass spectrometer was operated in electrospray positive mode; the optimum settings were found to be: capillary voltage of 1.00 kV, cone energy of 28 V, collision energy of 16 eV, extractor cone voltage of 3.00 V, RF lens voltage of 0.4 V, source temperature of 140°C, desolvation temperature of 350°C and desolvation gas flow of 600 L/h.
Validation
Validation was carried out according to published guidelines. 17
Ion suppression
Investigation of ion suppression, 23 a matrix-associated phenomenon that can cause inefficient droplet formation and competition for ionization within the source leading to loss of sensitivity, was investigated using a postcolumn infusion of choline at 10 μmol/L (in water), which represents a normal physiological concentration. This provided a constant signal in the choline MRM channel. When a prepared sample was loaded onto the column via the LC system, ion suppression would be seen as a reduction in the specific signal for choline.
Accuracy and imprecision
Internal QC samples were prepared at concentrations of 1.52, 7.46 and 37.3 μmol/L. These were used to determine intra- and inter-assay imprecision. For intra-assay imprecision, 15 samples at each concentration were extracted and measured during one analytical run. A coefficient of variation (CV) of less than 15% and a bias of less than 15% from the theoretical concentration were required for intra-assay imprecision to be acceptable.
Inter-assay imprecision was assessed by analysing the QC samples over a two-week period in 15 separate analytical runs. The requirements for acceptable CV and bias were the same as for intra-assay imprecision.
The lower limit of quantification (LLoQ) was defined by measuring 10 replicates of choline at concentrations of 0.76, 0.38, 0.19, 0.095 and 0.076 μmol/L. The LLoQ was deemed to be the lowest value at which the result obtained had a CV of 20% or less and a bias of 20% or less.
Linearity
Linearity of choline in its matrix was assessed by serial double dilutions of neat patient plasma or blood with water. A standard curve over the range 0.37–224 μmol/L was constructed. The calibration curve was also extended to 754 μmol/L on one occasion to determine the highest concentration at which linearity was maintained and that patient samples would fall on a linear portion of the curve.
Recovery
Recovery of choline from matrix does not need to be exactly 100% but must be consistent, precise and reproducible. To assess recovery, three different concentrations of choline were spiked into both blood and plasma at concentrations expected to correspond to the low, mid and high levels of choline in patient samples, giving (in addition to endogenous choline) concentrations of 10.6, 21.4 and 42.7 μmol/L. Recovery from blood or plasma was compared with that from phosphate-buffered saline–bovine serum albumin once all samples had been blank corrected.
Stability
Freeze–thaw stability
Aliquots (200 μL) of blood and plasma (6 plasma, 6 blood) were subjected to between one and five freeze–thaw cycles. Samples were frozen at −80°C until solid (40 min), defrosted, vortex mixed and refrozen. All samples were defrosted and analysed during the same analytical run. Stability was defined as a deviation of less than 10% from the original value.
Extracted sample stability
To determine the stability of extracted samples, patient samples (15 plasma, 15 blood) were extracted and analysed in duplicate. The plate was immediately resealed, stored at room temperature (20°C) and then reanalysed 24 h later.
Sample stability
The stability of EDTA samples was investigated with reference to stability during storage at either 20 or 4°C and storing plasma separated or on cells. Samples were held under appropriate conditions for between 24 and 192 h. This study was then extended and stability was investigated using lithium heparin and EDTA anticoagulants. For this extended study, both whole blood and plasma choline concentrations were measured on samples that had been stored for between 0 and 144 h at either 4 or 20°C. Acceptable stability was defined as a change of 10% or less. Samples were aliquoted and stored under relevant experimental conditions before being frozen. All samples were defrosted and analysed in a single batch for each of the two stability experiments.
Stability of detector response
Extracted sample was injected repeatedly during an overnight analytical run to determine sample stability and source saturation. The run time for plasma was 13.5 and 10.8 h for whole blood.
Results
The transitions for choline and d9 choline were m/z103.9 > 59.7 and m/z113.0 > 68.8, respectively. Choline and the deuterated internal standard co-eluted at 1.5 min. The total run time, injection to injection, was 4.5 min. Ion suppression studies using patient blood and plasma samples (6 plasma, 6 blood) showed no significant ion suppression in the region of 1.5 min where choline elutes. The lower trace in Figure 1 shows the steady response obtained by injecting a 10 μmol/L choline solution directly into the ion source. The upper trace in Figure 1 shows a typical choline chromatogram obtained during the infusion of the 10 μmol/L solution via the guard and analytical columns, showing the time at which choline elutes. If ion suppression were present, there would be a large dip in the lower trace, significant if it occurred at the time during which choline is eluting from the column. It can be seen on the lower trace that there is no significant reduction in signal at any time point.
Chromatogram showing results of post column infusion experiment and lack of ion suppression with d9 choline acting as a retention time marker
Linearity of patient samples upon dilution was calculated with an R 2 value of 0.9998, supporting lack of matrix effect. Linearity of the standard curve was seen up to 754 μmol/L (R 2 = 0.997). The LLoQ was 0.38 μmol/L. At this concentration, the CV was 12.9% and the bias 3.6%.
Overall imprecision results for intra-assay imprecision were CV <4% and bias <4% with inter-assay CV <6% and bias <10% at all three concentrations, representing acceptable imprecision. Inter- and intra-assay imprecision and bias results for each concentration of quality control material can be seen in Table 2.
Inter- and intra-assay imprecision and bias at different QC concentrations
QC, quality control; CV, coefficient of variation
The CV for the analyte peak height over the period of repeated injections was 3.0% for EDTA plasma and 4.3% for EDTA whole blood. The mean recovery from EDTA plasma at each of the three spiked concentrations was 101–103% (range 94–107%) and 107–108% (range 98–114%) from EDTA blood.
There was no significant change in extracted samples, either EDTA plasma or EDTA blood between the initial analysis and analysis 24 h later, with CVs of 1.3% and 1.7%, respectively. Freeze–thaw studies showed that samples were stable for up to five freeze–thaw cycles, as shown in Figure 2.
Stability of choline concentrations in six ethylenediaminetetraacetic acid whole blood samples over five freeze–thaw cycles
Six EDTA plasma samples had CVs of between 1.1% and 3.7% and six EDTA blood samples had CVs ranging from 0.9% to 1.5%. Since these CVs were within the intra-assay imprecision limits, the samples were deemed to be stable. However, sample stability studies revealed that significant (>10%) increases in sample choline concentrations occurred after 5 h if stored at room temperature (20°C) or 23 h if stored at 4°C, with smaller changes occurring at earlier time points. This is shown in Table 3. A graphical representation of this change is given in Figure 3 using choline concentrations rather than % changes in choline concentration.
Stability of choline measured in ethylenediaminetetraacetic acid and lithium heparin blood and plasma stored at either 4 or 20°C over time
Percentage changes in measured choline concentration in plasma and blood anticoagulated with lithium heparin (LiHep) and ethylenediaminetetraacetic acid (EDTA) when stored at either 4 or 20°C at various time points up to 144.5 h
Figure 4 shows the stability of choline in EDTA plasma stored either as separated EDTA plasma, or together with the cells (the latter samples were subsequently separated from the cells before being frozen).
Change in choline concentration of ethylenediaminetetraacetic acid anticoagulated plasma samples stored either as whole blood (i.e. on the cells) or plasma, and at 4 or 20°C
Method discussion
Validation of the method according to the published criteria was generally straightforward, encountering few difficulties. Sample preparation was simple and rapid, requiring no derivatization or extraction steps. This is facilitated by the novel arrangement of guard and analytical columns allowing the sample to be trapped onto a guard cartridge and washed before being eluted onto the analytical column. Not only does this save on sample preparation time, but extensive washing of the guard column protects and prolongs the life of the analytical column. Sample volume requirement is small (20 μL plasma or blood) and only 3 μL of supernatant is injected. The materials used to formulate standards and QC are cheap and readily available from commercial chemical companies, making the assay cheap to run in a laboratory that already possesses a mass spectrometer. Co-elution of the d9 internal standard with choline makes it an ideal standard, successfully correcting for any potential losses in sample preparation or machine sampling. Extending the standard curve to 754 μmol/L revealed that although linearity was maintained, saturation of the source was occurring due to the abundance of choline in the sample. Although the internal standard was affected in an identical manner, a cut-off of 224 μmol/L was chosen as the top standard as little saturation was seen at this point and patient results were unlikely to be higher than this. Analysis of the data for the extended analytical run for each of plasma and blood showed that the overall CV (3.0% for plasma and 4.3% for blood) during these experiments was less than the inter-assay imprecision, the response was stable (CV of 3.0% for plasma and 1.8% for blood), therefore an observation of a slight tendency to lose height over the course of the run was not seen as significant.
Stability of the sample in vitro was shown to be an important issue, see Table 3 and Figure 3, and has been recognized by other groups. 12,19 Samples appeared to be stable for 2.5 h, whether at room temperature or refrigerated but then deteriorated, and as such it would seem appropriate to recommend samples collected for choline should be processed urgently. From the results of this validation the recommendation would be to freeze either whole blood or separated plasma as soon as possible, ideally within 2 h or if this is not possible, refrigeration and freezing as soon as possible thereafter.
The investigation of lithium heparin and EDTA as anticoagulants showed that there appears to be a more rapid initial change in choline concentration in EDTA than in lithium heparin samples; thereafter, choline is much more stable in EDTA samples. This experimental finding is difficult to explain and would require further investigation before EDTA samples were used if a delay in processing could be possible, as for routine assays. These results are shown in Figure 3, in agreement with those of Yue et al. 19 Some studies have used lithium heparin anticoagulated samples, and reference ranges may therefore be slightly different according to sample collection and storage procedure. 6,19,20 All study samples were collected into EDTA anticoagulated bottles.
The stability of extracted samples confirms the report by Holm et al. 12 of complete stability of choline mixed with three volumes of acetonitrile at both 0 and 25°C for at least 72 h.
Choline has been measured in a number of different situations; whole blood and plasma for their prognostic value in acute coronary syndromes, 6,21 looking at the differential information given by choline raised only in blood and not plasma, 7 as a marker of phospholipase D activation 5 and as a serial marker to monitor treatment with glycoprotein IIIb/IIa inhibitors. 22 The method presented here would be ideal for the ongoing work in these areas.
Conclusion
In conclusion, a rapid, simple and robust method for the quantitation of choline in either whole blood or plasma using an SCX guard cartridge for online sample clean up has been developed. This method has been validated according to industry guidelines and would be useful in clinical research.
DECLARATIONS