Detection Times for Urinary EtG and EtS 57 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtG (mg/L) 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtG/Crea (mg/g) A B C D 20 40 60 80 100 120 140 Time until EtG 0.5 mg/L (h) 30 40 50 60 70 80 90 100 Time until EtG/Crea 0.5 mg/g (h) 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtS (mg/L) 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 EtS/Crea (mg/g) Fig. 2. Urinary excretion time profiles for EtG and EtS concentrations (A and C) and the corresponding values after normalizing the results to urinary creatinine (B and D) in 32 alcohol-dependent patients during detoxification (5–8 samples/patient, mean 7.3). Inset: Box-and-whisker plots for the times until urinary EtG and EtG/creatinine had returned to below the cut-off limits ( 0.5 mg/L and 0.5 mg/g, respectively). weight of 41.5–124 kg (mean 77.0, median 75.8) fulfilled the inclusion criteria of this study, by showing an ethanol concentration of at least 1 g/L on admission to the hospital and a markedly lower or negative second test result carried out on average 17.6 h (median 16.8, range 5.5–41.5) later. Breath alcohol measurements demonstrated initial ethanol concentra- tions ranging from 1.0 g/L to 3.4 g/L with a mean value of 2.0 g/L (median 1.9). Measurement of urinary EtG and EtS was carried out by LC-MS with all positive results confirmed by LC-MS/MS. For comparison, EtG was also quantified using the DRI-EtG en- zyme immunoassay. An overall good agreement of the EtG results obtained with the DRI-EtG and LC-MS/MS was seen over the entire concentration range (0–2440 mg/L) (Fig. 1). In the calculations, however, only the LC-MS/MS data were employed. Individual time course graphs for urinary EtG and EtS are shown in Fig. 2. For all 32 patients, EtG and EtS remained pos- itive for considerably longer time than urinary ethanol (ethanol data not shown). All urine samples collected on admission to the hospital were positive for ethanol (range 1.0–4.4 g/L, mean 3.0, median 3.1) but in only four (12.5%) patients, ethanol was also detectable in the second urine sample collected the next morning. For EtG, the time from admission (i.e. first testing) until return to below the applied clinical cut-off limit ( 0.5 mg/L) ranged from ∼40 h to ∼130 h (Fig. 2A) with a mean of 77 h (median 78, 25th–75th percentile 64–88). A sim- ilar time course was seen for urinary EtS with detection times of ∼55–110 h when using a quantification limit of 0.1 mg/L (Fig. 2C). The EtG and EtS values showed an overall good correlation with an EtG/EtS mean molar ratio of 2.5 (Fig. 3). In at least two of the patients, increased EtG and EtS values were seen after the return to low or negative values, indicating some kind of ethanol exposure (either drinking on purpose or due to unintentional ingestion from ethanol-containing prod- ucts) (Fig. 2). However, no positive breath tests were recorded during the study period. Because the EtG and EtS concentrations are known to be in- fluenced by urine dilution (Dahl et al., 2002 Bergstr¨ om et al., 2003 Helander and Beck, 2005), the effect of normalizing values to urinary creatinine was evaluated. This practice re- sulted in a lower inter-individual variability for both metabolites by guest on February 18, 2011 alcalc.oxfordjournals.org Downloaded from

58 Helander et al. 0 500 1000 1500 2000 2500 3000 3500 12000 10000 8000 6000 4000 2000 0 EtS (μmol/L) EtG (μmol/L) 0 1 2 3 4 5 6 7 8 35 30 25 20 15 10 5 0 EtG/EtS molar ratio Frequency Fig. 3. Correlation between urinary EtG and EtS concentrations determined by LC-MS. Regression equation: y = 5.974 + 2.786 ×, n = 297 (all values included), r = 0.951, P 0.0001. Inset: the corresponding EtG/EtS molar ratio (mean 2.5, median 2.25, range 0.21–7.22). (Fig. 2B and D). For EtG/creatinine (Fig. 2B), the time until return to below 0.5 mg/g ranged from ∼40 h to ∼90 h with a mean of 67 h (median 65, 25th–75th percentile 62–72). A weak positive correlation was seen between the initial breath ethanol concentrations and the urinary detection times for EtG (r = 0.434, P 0.013) but did not reach statistical sig- nificance for EtS (r = 0.189). To compensate for the variable initial alcohol concentrations and infrequent urine sampling times, the approximate time for a zero ethanol concentration was estimated for each patient, based on their ethanol concen- tration on admission and applying an average ethanol elimina- tion rate of 0.18 g/L/h (Jones et al., 1997). When expressed in this way, the excretion curves for EtG and EtS showed even better inter-individual agreement (Fig. 4A and C) (time range until EtG 0.5 mg/L ∼30–110 h, mean 66, median 66, 25th– 75th percentile 54–80 P = 0.013 compared with the original standard deviation). After also correcting for variations in urine dilution by calculation of ratios to urinary creatinine (Fig. 4B and D), the time range until EtG/creatinine was 0.5 mg/g was ∼30–70 h (mean 56, median 56, 25th–75th percentile -20 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 EtS/Crea (mg/g) -20 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtS (mg/L) -20 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtG/Crea (mg/g) -20 0 20 40 60 80 100 120 140 160 Time (h) 1 10 100 1 000 EtG (mg/L) A B C D 20 30 40 50 60 70 80 90 Time until EtG/Crea 0.5 mg/g (h) 20 40 60 80 100 120 Time until EtG 0.5 mg/L (h) Fig. 4. Urinary excretion time profiles for EtG and EtS concentrations in relation to the estimated time for a zero ethanol concentration (A and C), and the corresponding values after normalizing the results to urinary creatinine (B and D) in 32 alcohol-dependent patients during detoxification (5–8 samples/patient, mean 7.3). Inset: Box-and-whisker plots for the times after the estimated zero ethanol concentration until urinary EtG and EtG/creatinine had returned to below the cut-off limits ( 0.5 mg/L and 0.5 mg/g, respectively). by guest on February 18, 2011 alcalc.oxfordjournals.org Downloaded from