Validation study to compare effects of processing protocols on measured N (ε)-(carboxymethyl)lysine and N (ε)-(carboxyethyl)lysine in blood.

Epidemiological studies show that elevated plasma levels of advanced glycation end products (AGEs) are associated with diabetes, kidney disease, and heart disease. Thus AGEs have been used as disease progression markers. However, the effects of variations in biological sample processing procedures on the level of AGEs in plasma/serum samples have not been investigated. The objective of this investigation was to assess the effect of variations in blood sample collection on measured N (ε)-(carboxymethyl)lysine (CML), the best characterised AGE, and its homolog, N (ε)-(carboxyethyl)lysine (CEL). The investigation examined the effect on CML and CEL of different blood collection tubes, inclusion of a stabilising cocktail, effect of freeze thaw cycles, different storage times and temperatures, and effects of delaying centrifugation on a pooled sample from healthy volunteers. CML and CEL were measured in extracted samples by ultra-performance liquid chromatography-tandem mass spectrometry. Median CML and CEL ranged from 0.132 to 0.140 mM/M lys and from 0.053 to 0.060 mM/M lys, respectively. No significant difference was shown CML or CEL in plasma/serum samples. Therefore samples collected as part of epidemiological studies that do not undergo specific sample treatment at collection are suitable for measuring CML and CEL.

Introduction

Advanced glycation end products (AGEs) are a heterogeneous group of bioactive compounds, which can be formed from the non-enzymatic glycation of proteins, with reducing sugars, such as glucose.( Glycation occurs through the formation of a Schiff base intermediate followed by an Amadori rearrangement to give the ketoamine adduct. When glucose is the reducing sugar, the Amadori rearrangement product (ARP) is known as fructoselysine (FL). Further rearrangement, fragmentation and oxidation reactions of FL lead to the formation of AGEs, such as Nε-(carboxymethyl)lysine (CML). AGEs can be formed endogenously in the body under normal physiological conditions,( and also exogenously during food processing.( CML is one of the best characterised AGEs, and has been frequently used as a marker of the Maillard reaction in human tissue.( CML formation can result from two main pathways involving glucose as a precursor.( Nε-(carboxyethyl)lysine (CEL), the homolog of CML, is formed during the reaction of methylglyoxal.(

Epidemiological studies have shown that elevated plasma or serum advanced glycation end-products levels are associated with disease conditions such as diabetes, kidney disease, and heart disease.( There has been some debate and speculation as to the requirement for a stabilisation step at the time of blood collection to prevent artefactual CML and/or CEL formation. This has led some investigators to propose including a protocol to prevent artefactual formation of CML and CEL (as well as AGEs requiring oxidation steps for their formation) at the time of blood collection (Professor SR Thorpe, University of South Carolina, Columbia, SC, personal communication) in a similar way to the stabilisation of ascorbic acid in serum and plasma samples using metaphosphoric acid at time of sample collection. However, it has not been shown whether this additional step in the collection protocol is necessary to prevent artefactual CML and/or CEL formation. Whether additional treatments are required at the time of blood collection is important as including additional steps in handling of blood samples at collection would inevitably increase the costs associated with CML and/or CEL analysis, and reduce the likelihood of such samples being collected in large epidemiological studies. The need for these additional treatments also inevitably limits the use of blood samples that have already been collected and have not undergone pre-treatment at time of collection; thus excluding many cohorts of blood samples from the opportunity to analyze AGEs.

The aim of this validation study was therefore to determine the effect of variations in blood processing conditions, common across epidemiological studies, on the level of CML and CEL in blood samples quantified using a validated ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) method adapted from.( The rationale for the various treatments tested in this validation study are summarised in Tables 1 and 2.

Table 1

Experimental design

		Exp 1	Exp 2	Exp 3	Exp 4	Exp 5.1	Exp 5.2	Exp 5.3
Collection tube	Serum	√
	EDTA	√	X	X	X	X	X	X
	Lithium heparin	√
Stabilisation cocktail included	Yes	√	X			√	X	X
	No	√		X	X	√
Time stabilisation cocktail prepared	Fresh		√
	1 day in advance		√
	1 week in advance		√
Time until stabilisation	Immediate						√
	Delayed						√
Temperature samples maintained during processing	Room temp					√	√	√
	Refrigerated					√	√	√
Time until centrifugation	Immediately					√	X
	Delayed					√
Combined stabilisation and centrifugation	Immediately							√
	Delayed							√
Frozen storage temperature after processing	−20°C				√
	−80°C				√
Freeze-thaw cycles	0			√
	1			√
	2			√
Delay until UPLC-MS/MS analysis	Immediately				√
	Delayed				√

√, conditions varied in experiment; X, conditions fixed in experiment; EDTA, ethylenediamine tetraacetic acid; UPLC-MS/MS, ultra-performance liquid chromatography-tandem mass spectrometry.

Table 2

Effect of blood collection tube type and inclusion of stabilisation cocktail

	CML (mM/M lys)			CEL (mM/M lys)
	Stabilisation cocktail inclusion			Stabilisation cocktail inclusion
	Without	With	p1 value	Without	With	p1 value
EDTA	0.135 (0.005)	0.135 (0.005)	1	0.056 (0.003)	0.060 (0.002)	0.5
Serum	0.140 (0.004)	0.137 (0.005)	0.83	0.056 (0.003)	0.057 (0.003)	0.51
Lithium heparin	0.139 (0.004)	0.136 (0.005)	0.66	0.057 (0.004)	0.057 (0.002)	0.49
p2 value	1	0.66		0.66	0.66

Values are median (interquartile range). p1 values indicate differences between columns (comparison of effect ± stabilistion cocktail; Mann-Whitney U test). p2 values indicate differences between rows (comparison of blood collection tubes; Kruskall Wallis test). CML, Nε-(carboxymethyl)lysine; CEL, Nε-(carboxyethyl)lysine; EDTA, ethylenediamine tetraacetic acid.

Materials and Methods

Materials

Sodium hydroxide 98%, indomethacin, butylated hydroxytoluene (BHT), DTPA, ethanol 99%, phosphate buffered saline (PBS), sodium bicarbonate, boric acid 99.5%, sodium borohydride, trifluoroacetic acid (TFA), trichloroacetic acid (TCA), nonafluoropentanoic acid (NFPA) 97%, acetonitrile (HPLC grade), L-Lysine, and C18 (SupelcleanTM LC-18) SPE tubes were purchased from Sigma (Gillingham, UK). Other chemicals and their suppliers are as follows: EDTA, serum and lithium heparin blood collection tubes (Vacuette Greiner, Frickenhausen, Germany). Methanol (HPLC grade) (VWR International, Lutterworth, UK), hydrochloric acid 37%, (J.T. Baker, Devender, The Netherlands), chloroform (Bios Europe, Skelmersdale, UK), CML, d2-CML, CEL, and d4-CEL (NeoPMS, Strasbourg, France), d4-Lysine Cambridge Isotopes (Andover, MA).

Subjects

The study was conducted according to the guidelines laid down in the declaration of Helsinki and all procedures involving human subjects were approved by School of Medicine Dentistry & Biomedical Sciences Research Ethics Committee in Queen’s University Belfast. Each participant signed before donating blood to indicate informed consent. Blood samples (~40 ml) were collected from healthy volunteers (n = 10) who had been fasting for 12 h.

Pooling of blood samples

The serum supernatants were collected and combined to create a pooled sample (n = 10) for subsequent treatments as described below. A similar method was employed for pooling plasma.

Preparation of stability cocktail

Indomethacin (20 mM; prepared in 5% NaHCO3), was diluted to 0.2 mM indomethacin by addition of PBS (1×). BHT (5 mM) was prepared in ethanol. DTPA (10 mM, pH 7.4) was prepared in PBS (1×), and then the pH was adjusted firstly with 6N NaOH for coarse adjustment and then with 1N NaOH for fine adjustment. The stabilisation cocktail was comprised of the following per 1 ml of blood: 75 µl of indomethacin, 4 µl of BHT and 10 µl of DTPA. It was added to the collection tubes before blood collection. Similar additions were made for serum and plasma where applicable.

Stabilization cocktail treatment

Samples treated with stabilisation cocktail were kept at 4°C for 30 min and then centrifuged (2400 × g) for 20 min at 4°C. Unless otherwise stated, untreated EDTA and lithium heparin samples were centrifuged immediately after blood collection. Untreated serum samples were allowed to clot before centrifugation. In all experiments, unless otherwise stated, the stability cocktail was prepared on the day of experiment and stored at 4°C.

Experimental design

The series of 5 experiments are described briefly below and summarised in Table 1. Unless otherwise stated in an experimental manipulation, samples were stored at −80°C following treatment and measurement of CML and CEL were conducted immediately after blood collection. Samples were prepared in triplicate in all experiments.

Experiment 1: To test the combined effects of blood collection tube type and inclusion of stabilization cocktail.

Blood was collected in 3 different blood tube types (EDTA, serum and lithium heparin) either with or without stabilisation cocktail (6 treatments in total). Corresponding serum and plasma sample aliquots were collected, extracted, and analyzed for CML and CEL immediately.

Experiment 2: To test the effect of using fresh vs pre-prepared stabilisation cocktail.

Experiment 1 showed that there was no difference between blood collection tube type, so all subsequent experiments were conducted on EDTA tubes only. The stability cocktail was prepared on the day, the day before, and 1 week before blood collection, and stored in the cold room at 4°C prior to the addition in the appropriate collection tubes. Stabilisation cocktail (that had been (i) prepared fresh or (ii) prepared 24 h in advance or (iii) prepared 7 days, in advance) was added to EDTA blood collection tubes. Samples were then kept at 4°C for 30 min, and centrifuged (2400 × g) for 20 min at 4°C.

Experiment 3: To test the effect of multiple freeze-thaw cycles.

Blood samples were collected in EDTA collection tubes and then centrifuged (2400 × g) for 20 min at 4°C. From the pooled sample 3 × 1 ml aliquots were transferred to different tubes. One sample was analyzed immediately for AGE levels, and the remaining 2 treatment samples were put through the freeze thaw cycle (−80°C) either 1 or 2 times, before CML and CEL analyzes.

Experiment 4: To test the effect of frozen storage temperatures and immediate vs delayed measurement of AGE.

Blood samples were collected in EDTA collection tubes and then centrifuged (2400 × g) for 20 min at 4°C. The samples were then stored at either −20°C or −80°C for varying storage periods (1, 2, 4 and >6 months).

Experiment 5: To test the effects of immediate vs delayed centrifugation and/or addition of stabilising cocktail solution.

Experiment 5.1: Effect of immediate vs delayed centrifugation. Blood samples were collected in EDTA tubes (with or without stabilisation cocktail) and stored at either 4°C or 21°C for varying periods of time (0, 1, 2, 4, 8 and 24 h). Samples were removed from storage at the corresponding time, kept at 4°C for 30 min (stabilisation cocktail-treated samples only) and all samples were then centrifuged (2400 × g) for 20 min at 4°C.

Experiment 5.2: To test the effect of delay in stabilisation. Blood samples were collected in EDTA tubes and kept at 4°C for 30 min followed by centrifugation (2400 × g) for 20 min at 4°C. Samples were then stored at either 4°C or 21°C for varying periods of storage time (0, 1, 2, 4, 8 and 24 h), following which stability cocktail was added to samples at the corresponding delay time.

Experiment 5.3: Effect of conducting immediate vs delaying both stabilisation and centrifugation. Blood samples were collected in EDTA tubes and stored at either 4°C or 21°C for varying periods of time (0, 1, 2, 4, 8 and 24 h). Samples were removed from storage at each time point and the stability cocktail was added. Samples were kept at 4°C for 30 min followed by centrifugation (2400 × g) for 20 min at 4°C.

CML and CEL extraction

The samples were extracted based on the method previously reported.( Briefly, a 10 µl portion of the supernatant was reduced overnight at 4°C in sodium borate buffer (0.5 M, pH 9.2, 400 µl) and 2 M sodium borohydride in 0.1 M NaOH. Proteins were then precipitated by addition of 200 µl of 60% TCA and pelleted by centrifugation (2000 × g for 10 min). The supernatant was carefully removed by aspiration with a Pasteur pipette. The protein pellet was washed once by resuspension in 1 ml of 20% TCA and centrifuged (2000 × g for 10 min). The internal standards (CML, d2-CML, CEL, and d4-CEL) were added and samples were hydrolyzed in 6 N HCl (1 ml) at 110°C for 24 h. The acid was removed under vacuum (Genevac evaporator (EZ-2), Genevac, Ipswich, UK), and the dried hydrolysates were reconstituted in 5 mM NFPA, and solid phase extracted using a C18 cartridge.

UPLC-MS/MS analysis

Lysine, CML, and CEL concentrations of hydrolysates were determined by a Waters Acquity Ultra Performance LC system coupled to a Waters Micromass Quattro Premier XE tandem quadruple mass spectrometer (Manchester, UK). Briefly, protein hydrolysates (0.25 µl plasma/serum, 5 µl injection) were injected onto a Waters BEH C18 UPLC column, 1.7 µm, 2.1 mm × 50 mm, housed in a column oven at 50°C. The flow rate was 0.3 ml/min. Solvent A was aqueous nonafluoropentanoic acid (NFPA, 5 mM) and solvent B was acetonitrile. Gradient elution was employed starting at 10% solvent B which was maintained for 0.4 min, followed by a linear gradient from 10% to 80% solvent B in 3.8 min, with a hold at 80% solvent B for 1.5 min, and then returned to 10% solvent B for 2 min. The run time was 7.5 min. The analysis was performed using a Waters Acquity UPLC (Manchester, UK) coupled to a triple quadruple MS operating in multiple reaction monitoring (MRM) mode. The MS was operated in electrospray ionization (ESI) positive mode using MRM. Lysine, CML, and CEL were quantified with the use of isotopically labelled internal standards and by reference to an external standard calibration curve. Data were reported as median ± interquartile range of triplicates.

Statistical analysis

Statistical analyzes were performed with SPSS 12.0 for Windows. Significance of differences between 2 and 3 groups was tested by the nonparametric Mann-Whitney U test and Kruskall Wallis tests, respectively, for independent samples. A two-tailed probability <0.05 was considered significant.

Results

Samples were extracted in triplicate and inter-sample variability was <7%. The inter-day and intra-day variability of the UPLC system was calculated by analysing a pooled plasma sample (n = 10), over five consecutive days (inter-day), and four times on the same day (intra-day). The coefficients of variation obtained for the reproducibility tests described above, were <10%. The LOD (0.01 µM/L) and LOQ (0.02 µM/L) were calculated at a signal-to-noise ratio of 3 and 10, respectively, for both CML and CEL.

Calibration curves, obtained by linear regression of a plot of the analyte/internal standard peak-area ratio versus analyte concentration, were linear over the range of concentrations tested with correlation coefficients >0.98. Selected results of experiments 1, 2, 3, 4, 5.1, 5.2 and 5.3 are shown in Tables 2, 3 and 4. There was no statistically significant effect of either blood tube type or inclusion of stabilisation cocktail on CML or CEL levels (p>0.48 in all cases; Table 2). When the cocktail was prepared in advance, there was similarly no effect on CML or CEL levels (data not shown). Varying frozen storage temperature (Table 3), delay after blood collection until AGE analysis (Table 3), and combined delay in both stabilisation and centrifugation (Table 4) all similarly had no effect on measured AGE levels. Repeat (up to 2) freeze-thaw cycles, delaying of centrifugation only, delayed addition of stabilisation cocktail only (data not shown) all similarly had no effect on measured AGE levels.

Table 3

Effect of frozen storage temperature and delay until analysis (EDTA tubes)

	CML (mM/M lys)			CEL (mM/M lys)
	Stored at −20°C	Stored at −80°C	p1 value	Stored at −20°C	Stored at −80°C	p1 value
Delay between collection and analysis
No delay	0.135 (0.007)	0.133 (0.004)	1.0	0.054 (0.004)	0.055 (0.003)	0.6
1 month	0.136 (0.005)	0.137 (0.005)	0.8	0.057 (0.003)	0.056 (0.004)	0.8
2 months	0.135 (0.006)	0.136 (0.004)	1.0	0.054 (0.003)	0.055 (0.004)	0.5
4 months	0.137 (0.006)	0.137 (0.004)	1.0	0.055 (0.004)	0.057 (0.003)	1.0
>6 months	0.133 (0.005)	0.137 (0.006)	0.4	0.057 (0.003)	0.057 (0.004)	1.0
p2 value	0.7	0.6		0.5	1.0

Values are median (interquartile range). Room temperature and cold room experiments were conducted separately. p1 values indicate differences between columns (comparison of effect ± stabilistion cocktail; Mann-Whitney U test). p2 values indicate differences between rows (comparison of blood collection tubes; Kruskall Wallis test). EDTA, ethylenediamine tetraacetic acid; CML, Nε-(carboxymethyl)lysine; CEL, Nε-(carboxyethyl)lysine.

Table 4

Effect of immediate and delayed stabilisation and centrifugation (EDTA tube plus cocktail)

	CML (mM/M lys)			CEL (mM/M lys)
	4°C (Room temp)	21°C (Cold room)	p1 value	4°C (Room temp)	21°C (Cold room)	p1 value
Delay in both stabilisation and centrifugation
No delay	0.136 (0.004)	0.137 (0.005)	1.0	0.059 (0.003)	0.058 (0.004)	0.7
1 h	0.137 (0.005)	0.138 (0.005)	1.0	0.054 (0.004)	0.056 (0.004)	1.0
2 h	0.136 (0.006)	0.138 (0.005)	0.7	0.055 (0.004)	0.055 (0.004)	0.8
4 h	0.136 (0.005)	0.137 (0.005)	0.5	0.057 (0.003)	0.059 (0.003)	1.0
8 h	0.138 (0.004)	0.139 (0.006)	0.7	0.055 (0.003)	0.053 (0.003)	0.4
24 h	0.136 (0.005)	0.139 (0.005)	1.0	0.057 (0.004)	0.058 (0.002)	1.0
p2 value	1.0	0.95		0.89	0.92

Values are median (interquartile range). p1 values indicate differences between columns (comparison of effect ± stabilistion cocktail; Mann-Whitney U test). p2 values indicate differences between rows (comparison of blood collection tubes; Kruskall Wallis test) EDTA, ethylenediamine tetraacetic acid; CML, Nε-(carboxymethyl)lysine; CEL, Nε-(carboxyethyl)lysine.

Discussion

The study was designed to mimic the commonly used procedures and storage conditions that occur in an epidemiological study. The results obtained showed no significant difference between any of the blood processing treatments or storage conditions investigated, on either the level of CML or CEL in plasma/serum samples. The study showed that use of the stability cocktail had no effect on the level of CML and CEL, and thus is an unnecessary step in their analysis. The results from the current investigation demonstrate that the levels of CML and CEL in plasma/serum samples are not significantly affected by the normal procedures and conditions employed in an epidemiological study, and thus are suitable for measurement in large scale epidemiological studies. Samples collected across different population studies use collection procedures which can vary by many factors including: (i) delays to centrifugation, (ii) blood collection tube, (iii) blood collection temperature pre-centrifugation, (iv) frozen storage temperature, (v) delay between collection and analysis and (vi) exposure to multiple freeze-thaw cycles. In the case of AGE analysis specifically, variations which may occur include: (i) inclusion of stabilisation cocktail, (ii) advance or fresh preparation of stabilisation cocktail, or (iii) immediate and delayed addition of stabilisation cocktail.

An exponential increase in the number of publications relating to AGEs has occurred in the last 30 years. A PubMed search shows that in the 1980s, 1990s, and 2000s, the number of articles on AGEs was 21, 978, and >3700, respectively. Epidemiological studies have suggested that individuals with higher elevated circulating CML, are at greater risk of arterial stiffness,( chronic kidney disease,( anemia,( cardiovascular and all-cause mortality.( The data provided here support the use of stored samples, regardless of collection and storage method, from other epidemiological studies for the analysis of CML and CEL to further examine the association between AGEs and disease risk. These results are based on analysis of samples collected from apparently healthy volunteers whose AGE blood levels are typically low. A comparison of the effect of pre-treatment of samples would need to be conducted on blood samples with higher AGE levels (such as amongst diabetics) to demonstrate if the non-effect of pre-treatment is maintained.

Furthermore, it supports the conclusion that future studies planning to measure CML and CEL do not need to employ additional processing steps at blood collection.

In conclusion additional processing of blood at collection does not elevate measured levels of CML and CEL. This indicates that specific collection protocols are not required for measuring CML and CEL, thus indicating that many samples currently stored for epidemiological studies could be used for CML and CEL analyzes.