Analytical aspects of measuring 2H/1H and 18O/16O ratios in urine from doubly labelled water studies by high- temperature conversion elemental analyser-isotope-ratio mass spectrometry

International audienceRATIONALE: Total Body Water (TBW) and Total Energy Expenditure (TEE) are routinely measured in free-livingconditions by the 2H218O method. Isotope eliminations can be measured from spot urine samples by HTC-EA IRMS,but only after cumbersome cryogenic distillation to extract water. Distillation may, however, be replaced by charcoaltreatment and filtration. This study tested (1) the effect of sample treatments (filtration versus distillation) on the isotoperatios, (2) the effect of different ways of normalization that respect or not the principle of identical treatment of thesample and references, and (3) the impact on the biological outcomes.METHODS: Two filters (PES membrane; 10 kDa) accepting volumes of urine samples (V500: 0.5 mL versus V6: 3.0 mL)were tested. In-house water standards and in-house urine standards were prepared and normalized against theinternational scale to calibrate the urine samples. The δ2H and δ18O values from water in the urine were measured byHTC-EA IRMS.RESULTS: Filtered urine normalized with water standards showed a bias in the δ2H values that was corrected whencalibration was performed with urine standards. At a δ2H value of 1101.4 ‰, the accuracy increased from –11.9 to –0.2δ‰ (V500) and from –3.8 to 0.4 δ‰ (V6). The TBW errors were greatest with V500 and water calibration (1.20 %) andlowest with V6 and urine calibration (0.34 %; preparation-by-calibration interaction p = 0.027). For the δ18O values theaccuracy of enrichments and TBW were not affected whatever preparations and normalization were used. The averageTEE was not affected but the variability increased from 0.6 to 2.7 % versus cryogenic distillation.CONCLUSIONS: Cryogenic distillation remains the gold standard for small sample size experiments where smallchanges in TEE are to be detected. Filtration offers an alternative for large-scale experiments. When the body compositionis derived from 2H2O dilution, it is strongly recommended that urine standards should be used to eliminate the effect offiltration. Copyright © 2015 John Wiley & Sons, Ltd