Epidemiologic studies suggest a link between the intake of refined sugars and an increased risk for colorectal,1, 2 breast,3, 4 pancreatic5 and endometrial6 cancer. However, several studies have failed to show any association7–12 and findings on the relationship between sugars intake and obesity – a further risk factor of several cancers – have been largely inconclusive.13 The reason for these inconclusive findings may be the ambiguity of dietary assessment methods employed14 due to the limitation of dietary assessment – in general by self-reporting – in epidemiological studies;15, 16 there is for example a substantial underreporting of foods which are considered unhealthy.16 In contrast to self-reporting, biomarkers provide an unbiased tool to assess dietary intake. Sucrose has been established as a biomarker for intake of sugars,14 as sucrose is excreted in small amounts into the urine.17 Using sucrose and fructose as biomarker of sugars intake, we could show the discrepancy between self-reported and actual sugar intake in obese participants of the EPIC Norfolk cohort.18
For biomarkers to be used successfully in large-scale epidemiological studies, analytical methods must be available which allow the rapid and reliable analysis of large numbers of samples. These methods should also be easily transferable to different laboratories to ensure comparable results in multi-centre studies. Previously, sucrose has been determined using either high-performance liquid chromatography (HPLC) with electrochemical detection19 or an enzymatic assay,20 which are labour-intensive, time consuming and often lack sensitivity.21 Gas chromatography/mass spectrometry (GC/MS) and liquid chromatography (LC)/MS provide fast and sensitive alternatives which allow the automated analysis of large numbers of samples. Although these methods require expensive equipment and expert users, GC/MS and LC/MS are established in many bioanalytical laboratories and are commonly used for bioanalytical analyses. In this study, we present and compare two analytical methods for the determination of biomarkers for sugar consumption using MS detection with either GC or LC separation.
Sucrose was purchased from Sigma (Poole, Dorset, UK) and 13C12-sucrose from CK Gas (Hook, Hampshire, UK). Water, methanol, acetonitrile, dry hexane and pyridine were purchased from Sigma (Poole, Dorset, UK) and Fisher Scientific (Loughborough, Leicestershire, UK). To inhibit losses of target compounds by adsorption to glassware, only silanised glassware was used. As sucrose is not stable at low concentrations in aqueous solutions, standards were prepared freshly on the day from a concentrated (10 mM in water) stock solution. The stock solution was stable for at least 3 months.
Urine from seven healthy normal weight (body mass index (BMI): 20.6–24.6) volunteers following diets of different sucrose content (4 days, 1.1% and 35% of total energy, urine collection on days 3 and 4) was collected as described previously;14 1.5 g/L boric acid was added as preservative, the urine samples aliquoted and stored at −20°C. The completeness of 24 h urine collection was assessed by p-aminobenzoic acid.22 The study was approved by the LERC (Local Ethic Research Committee), study number 06/Q0108/220.
To 200 µL urine, 50 µL internal standard solution (13C12-sucrose, 100 µM in water) and 50 µL urease (29 Units, recombinant from E. coli; USB, Staufen, Germany) were added, vortex-mixed and incubated for 1 h at 37°C. Following incubation, 800 µL cold methanol were added to each sample, the samples were vortex-mixed and centrifuged for 10 min at 14 000 g. Then 500 µL of the supernatant were transferred into a glass vial and dried under reduced pressure. The samples were then derivatised and analysed as described previously.23 The precision of this method is 8.2%.24 Briefly, the samples were reconstituted in 30 µL methoxyamine hydrochloride (20 mg/mL in dry pyridine, Sigma). After 16 h at room temperature, samples were trimethylsilylated for 1 h at room temperature with 30 µL N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA, Sigma); the derivatised sample was diluted with 540 µL hexane prior to injection. The samples were analysed with a Trace GC Ultra and a Trace DSQ quadrupole mass spectrometer (ThermoElectron, Hemel Hempstead, UK). The derivatised sample was injected with a 1:10 split onto a 30 m × 0.25 mm i.d. 5% phenylpolysilphenylene-siloxane column with a chemically bonded 25 µm TR-5MS stationary phase (Thermo Electron). The oven temperature was kept at 60°C for 2 min and then increased by 5°C/min to 310°C. The carrier gas was helium (flow rate 1.2 ml/min). The mass spectrometer (transfer line temperature: 250°C; ion source temperature: 220°C; electron beam: 70 eV) was operated in full scan mode (50–650 m/z; 3 scans/s) and compounds were identified by their retention time (23.97 min) and characteristic fragments including a NIST library search. For quantification, the following ions were used (m/z): sucrose 361, 13C12-sucrose: 366. Full scan mode was used to allow the identification of additional metabolites and the application of metabolomics methods.
To 200 µL urine, 50 µL internal standard solution (13C12-sucrose, 100 µM in water) were added and vortex mixed. Thereafter, 500 µL cold acetonitrile were added to each sample, the samples were vortex-mixed and centrifuged for 10 min at 14 000 g. The supernatant was transferred into clear glass vials and analysed using a Quattro Ultima mass spectrometer (Waters, Manchester, UK) combined with two PU1585 micro-HPLC pumps (Jasco, Great Dunmow, UK) and a HTC PAL autoinjector (CTC Analytics, Zwingen, Switzerland). Samples were separated on a NH2-column (Varian Pursuit, 3 µ, 150 × 2.0 mm; Varian, Oxford, Oxfordshire, UK), kept at 15°C (CO-1560 column oven, Jasco) using the following gradient (mobile phase A: water, B: acetonitrile; flow rate 300 µL/min): 0 min: 100% B, 2 min: 100% B, 2.1 min: 82% B, 7 min: 21% B, 7.5 min: 21% B, 8 min: 100% B. Total analysis time (including equilibration) was 12 min. Samples were analysed using electrospray ionisation in the negative ion mode with a spray voltage of −2.6 kV. (Cone voltage: −70 V; source temperature: 120°C; desolvation temperature: 350°C; cone gas flow: 36 L/h; desolvation gas flow: 800 L/h.) Compounds were identified using multiple reaction monitoring (MRM, sucrose: 341 → 179; 13C12-sucrose 353 → 185) and their retention time (6.8 min).
Samples were quantified using the area ratio between analyte and internal standard. A calibration line was prepared by spiking water with an aqueous stock solution of sucrose to obtain the following concentrations (in µM): 1, 5, 10, 50, 100, 200 and 500. These calibration samples were prepared as described above. To quantify samples, a linear calibration term with 1/x weighting25 was used.
Quality control and validation
Urine collected as described above following a sucrose-free diet was used to prepare quality control (QC) samples. Samples were spiked with sucrose to achieve the following concentrations (in µM): 2, 20, 70 and 450, frozen and analysed as described above.
Statistical analysis was performed using Prism (version 4 for Mac OS X; Graphpad, San Diego, CA, USA) and GPower26 on values transformed to their decadic logarithm. The power of the analysis (paired t-test, α = 0.05, two sided) is 0.99.
Method validation and comparison
The developed methods were used to determine sucrose concentrations in urine between 1 and 500 µM. Within this range, the calibration line remained linear (r2 > 0.999). Table 1 shows precision and accuracy data for the LC/MS method. The precision (intra- and inter-batch) for the method is between 2% and 11% which is well within the limits suggested by the FDA for bioanalytical methods.27 As the urine used to prepare QC samples contained residual sucrose, the accuracy was determined from the recovery of spiked sucrose, i.e. the determined sucrose concentration corrected by the sucrose concentration in unspiked urine. The accuracy is also well within the recommended range of 15%. Figure 1 shows typical chromatograms of a sample and 10 µM calibration standard analysed with GC/MS and LC/MS.
Table 1. Validation data for LC/MS sucrose determination. Concentrations are given in µmol/L ± standard error. Accuracy is calculated as the closeness of the result to the theoretical result, precision is given as the coefficient of variation. ‘Corrected concentration’ is the sucrose concentration corrected by the residual sucrose determined as Q 0. Accuracy and intra-batch precision were determined from three samples, inter-batch precision from three different samples analysed on three consecutive days
Determined concentration [µM]
6.4 ± 0.1
8.1 ± 0.2
25 ± 2
77 ± 1
447 ± 22
Intra-batch precision (%CV)
Corrected concentration [µM]
Samples (n = 51) were analysed with both, LC/MS and GC/MS, and a comparison of the data shows a highly significant (p < 0.0001) correlation (Pearson's ρ = 0.99). Using the Bland-Altman28 method to compare the methods shows a good agreement (Fig. 2); the difference between both methods can be estimated as 1.3 µM (±1.6 µM).
Urinary sucrose as biomarker
Figure 3 shows the daily urinary sucrose excretion determined by LC/MS following a high- and low-sugar diet. There is a significant (p < 0.005) difference between both diets, confirming previous results.14
Mass spectrometry – either with LC or GC separation – is a suitable alternative to an enzymatic assay for the determination of urinary sucrose as biomarker for sugar intake. The LC/MS method developed here shows good precision and accuracy and involves only few sample preparation steps which can be performed by a liquid handling robot. The results are comparable with results obtained by GC/MS using a method which has been developed previously for the analysis of carbohydrates. Although the GC/MS method has the advantage of being able to identify more compounds and potentially allows the determination of several biomarkers in one run, it requires a much longer sample preparation and has a longer run time, reducing the throughput. In addition to the determination of precision and accuracy, this method was also used to determine urinary excretion in normal-weight volunteers following a low- and high-sucrose diet. Our results confirm previous findings about sucrose as biomarker for sugars intake14 and show the suitability of this method.