Background D/H before ingestion of labeled water in the 36 subjects ranged between 147 and 154 ppm and was not significantly different in urine, saliva, plasma or blood cells (149.0 ± 1.0, 149.1 ± 0.8, 148.8 ± 0.9 and 149.1 ± 1.4 ppm, one-way ANOVA for repeated measurements, P = 0.30). Following ingestion of labeled water, plasma and blood cell D/H quickly increased over basal values (Fig. ). Both plasma and blood cell D/H were significantly above baseline values (see above) beginning at min 5: 192.8 ± 30.6 and 192.2 ± 30.8 ppm, respectively (Student t test for paired data, P < 0.0001); 95% confidence interval around the mean = 148–149 and 182–202 ppm for both plasma and blood cells at baseline and min 5, respectively. Individual D/H values in blood cells were remarkably similar to those observed in plasma (data not shown: mean difference ± SD = 0.8 ± 1.6 ppm, n = 252). Accordingly, the average value and dispersion of blood cell D/H were similar to the corresponding plasma values at all time points during the 60-min observation period [Fig. ; two-way ANOVA for repeated measures (time × medium): no significant difference between media, P = 0.114]. Average peak plasma and blood cell D/H were reached at min 20, although individual time to peak ranged between min 15 and 60 (Fig. ).
Fig. 1 Plasma and blood cell deuterium to protium ratio (D/H) over the 60-min period following ingestion of labeled water (mean ± SD, n = 36). Blood samples were taken at min 0, 5, 10, 15 20, 30 and 60 but for the sake (more ...)
Fig. 2 Changes in plasma deuterium to protium ratio (D/H) following ingestion of labeled water and urine D/H over the nine following days in subjects from the first and second group (panela and b, n = 15 and 21, respectively. Individual values (more ...)
No significant difference was present at any time point between the six groups of subjects for plasma and blood cell D/H during the 60-min period following water ingestion and for the corresponding kinetic parameters computed with the three pharmacokinetic models, respectively (Kruskal–Wallis one-way analysis of variance for independent samples; P > 0.05). The data obtained with the six waters were thus pooled for subsequent analysis.
Visual inspection of individual changes in plasma D/H following ingestion of labeled waters as well as examination of the goodness of fit with the compartmental models allowed segregating the 36 subjects into two groups (Fig. ). In group 1 (n = 15, 42% of the subjects), plasma D/H peaked between min 15 and 30 (mean value = 326 ± 32 ppm observed at 20.0 ± 6.5 min) before decaying bi-exponentially and the data were well fitted by the 2-CM. In group 2 (n = 21, 58% of the subjects), plasma D/H peaked later, then decayed mono-exponentially and the data were well fitted by the 1-CM. In both groups, plasma D/H value was also well fitted by the non-compartmental model (N-CM). The goodness of fit of actual D/H values with predicted values yielded by the three pharmacokinetic models was assessed by the coefficient of variation, weighted corrected sum of squared observations, sum of weighted squared residuals, and weighted correlation coefficient between predicted and observed values (r > 0.93).
The decrease in urine D/H over the 9 days following ingestion of labeled water was adequately described by the slope-intercept method in each of the 36 subjects as shown by the correlation coefficients between predicted and observed values (r > 0.985, P < 0.0001) and no significant difference was observed between the two groups for any of the parameters (Table ; one-way ANOVA for independent measurements, P > 0.15).
Kinetic parameters of water absorption and distribution, and volume and turnover of the BWP computed with the different models (mean ± SD)
As depicted in Table , when compared with the kinetic parameters calculated with the 1-CM (group 2), the values of Cmax and Tmax calculated with the 2-CM (group 1) were higher and shorter, respectively (P < 0.001). The Cmax value computed using a compartmental model and the N-CM in the two groups was very similar. In group 1, the Tmax values predicted with the 2-CM and the N-CM were not different and close to that observed by visual inspection of the plasma D/H curve (~20 min). The Tmax value predicted by the N-CM in group 2 was much longer than that estimated in group 1 and much longer than that calculated with the 1-CM, and with a large inter-individual variability.
Based on the average values for ta½ estimated with the two-compartmental pharmacokinetic models (~11 and ~13 min with the 2-CM and 1-CM, respectively), water absorption was almost complete within ~75 to ~90 min after ingestion (i.e., 7 × ta½ with the 2-CM and 1-CM, respectively: >99% absorbed).
When the volume and kinetic parameters of the BWP estimated using the pharmacokinetic models were compared with those computed with the reference slope-intercept method in group 1, the 2-CM slightly but not significantly overestimated the volume of the BWP (e.g., ~3% for the absolute value and ~7% for the value corrected for LBM; not significant: one-way ANOVA for repeated measures, P = 0.40 and P = 0.19, respectively). The turnover of the BWP was, thus, slightly but significantly underestimated (e.g., ~8% for Clw; P = 0.041). In contrast, in group 2, no significant difference was observed between the volume and the kinetic parameters of the BWP obtained with the three models (one-way ANOVA for repeated measures; P > 0.52).
As shown in Fig. , for group 1, the volume of the central compartment computed using the 2-CM [18.5 ± 4.3 l, corresponding to 37 ± 10% of the total BWP and including the blood (~6 l in 76.9-kg subjects) but also a portion of extra-vascular space] was smaller than that of the peripheral compartment (31.6 ± 6.4 l or 63 ± 10% of the BWP). The microconstants depicted in Fig. show that the rate constant of absorption was much greater than the inter-compartmental transfer constants and the elimination rate constant. The half-life of water distribution between the central and peripheral compartment (td½) was estimated to be 12.5 ± 4.3 min, indicating that ~90 min after ingestion, labeled water was almost evenly distributed within the entire BWP (>99% of the pool).
Fig. 3 Mean kinetic parameters within a two compartment model and volume of the central (V1) and peripheral compartment (V2) (mean ± SD). ka rate constant of water absorption in the central compartment; k12 and k21 rate constant of water (more ...)