Article pubs.acs.org/ac
Assessment of Local Tobacco Consumption by Liquid Chromatography−Tandem Mass Spectrometry Sewage Analysis of Nicotine and Its Metabolites, Cotinine and trans-3′-Hydroxycotinine, after Enzymatic Deconjugation Tania Rodríguez-Á lvarez, Rosario Rodil, María Rico, Rafael Cela, and José Benito Quintana* Department of Analytical Chemistry, Nutrition and Food Science, IIAAInstitute for Food Analysis and Research, University of Santiago de Compostela, 15782−Santiago de Compostela, Galicia, Spain S Supporting Information *
ABSTRACT: Cotinine (COT), trans-3′-hydroxycotinine (OH-COT), cotinine-N-β-glucuronide (COT-GLUC), and trans-3′-hydroxycotinineO-β-glucuronide (OH-COT-GLUC) are excreted in urine following the intake of nicotine (NIC), and, as such, they have been detected in sewage. Thus, they also constitute convenient biomarkers for NIC tracing through the sewage epidemiology approach at the local scale. Such estimation requires granting a good stability of the target biomarkers in sewage. However, it was found that glucuronides are not stable, particularly in the case of OH-COT-GLUC, which could render variable concentrations of COT, OH-COT, and their glucuronides, depending on sampling and storage time or temperature. Thus, an enzymatic deconjugation with β-glucuronidase was optimized. With the optimized method, after enzymatic deglucuronization, the limits of quantification obtained were in the range of 0.2−1 μg L−1, relative standard deviations were 10 μL (data not shown). Stability of NIC and Its Metabolites in Wastewater. In the work of Castiglioni et al., COT and OH-COT are mentioned to be stable and glucuronides are assumed to be rapidly deconjugated in sewage.22 This contrasts with earlier studies where the stability of NIC, COT, and OH-COT in river water and wastewater with activated sludge had been previously investigated, showing in both cases to be biodegradable: within a few hours with activated sludge and some days in sediments.11,24 However, such experiments do not necessarily represent the stability in sewage, as they may contain bacterial activity levels that are too high or too low, compared to real sewage. Thus, the stability of NIC and their metabolites, including glucuronides, in wastewater samples was evaluated at three different temperatures: room temperature (20 ± 2 °C), 4 ± 2 °C, and −20 ± 2 °C. These are typical storage conditions. However, it is necessary to also take into consideration that a composite sampling is performed during 24 h, so that degradation may also already take place during sampling or previously in the sewage network. In these experiments, the glucuronides and the rest of compounds (NIC, COT, and OHCOT) were studied in separate batches. Nonfiltered samples (spiked with 50 μg L−1 of analytes) were used to better represent environment conditions, and subsequently stored and analyzed at different times, as detailed in the Materials and Methods section. As presented in Figure 1, at room temperature, COT and OH-COT showed no degradation, indicating that they are stable in wastewater for at least 1 week (also if refrigerated; see Figure S4 in the Supporting Information) also guarantying the representativeness of composite samples. However, NIC should not be stored longer than 2 days unless frozen (see Figure 1, as well as Figure S4 in the Supporting Information). In the case of OH-COT-GLUC, its concentration dropped over 35% in 8 h, even if refrigerated, and the other glucuronide, COT-GLUC was more stable and degradation began only at the seventh day (see Figure 1 and Figure S4 in the Supporting Information). Although freezing can stop degradation (Figure S4c in the Supporting Information), the concentration ratio of OH-COTGLUC and OH-COT would be variable in time, since both glucuronides were simply transformed to their nonconjugated species (data not shown) and, as mentioned, sampling plus transport to the laboratory was already longer than 24 h. In the case of NIC, its decrease in concentration did not yield COT or OH-COT, so it would not interfere with the analysis of its metabolites. Given the above findings, it was decided to perform an enzymatic deconjugation of glucuronides in order to avoid variable concentrations with time of glucuronides and their deconjugated species, so that total COT and OH−COT concentrations could be measured.
(Domnick Hunter, Durham, U.K.) and used as a nebulizing and drying gas. Argon (99.999%, Carburos Metálicos, A Coruña, Spain) was used as collision gas. Electrospray parameters were as follows: nebulizing gas: 55 psi, 50 °C; drying gas: 18 psi, 200 °C; ion spray voltage, 4500 V. NIC and its metabolites were determined in the electrospray positive mode and multiple-reaction monitoring (MRM) mode of acquisition (Table 1). According to the 2002/657/EC decision, two different MS/MS transitions were recorded, the most intense being used for quantification and the second one for confirmation. Table 1. LC-MS/MS Experimental Parameters Employed for the Determination of Analytes and Internal Standards transitions ratiob
tr (min)
MRM transitionsa
NIC
10.1
163 (48) → 84 (14) 163 (48) → 106 (10.5)
1.2 ± 0.2
NIC-d4
10.1
167 (48) → 84 (14) 167 (48) → 110 (10.5)
1.2 ± 0.2
COT
8.7
177 (60) → 80 (19) 177 (60) → 98 (15.5)
3.4 ± 0.9
COT-d3
8.7
180 (60) → 80 (19) 180 (60) → 101 (15.5)
3.4 ± 0.9
OH−COT
8.2
193 (56) → 80 (21.5) 193 (56) → 134 (13.5)
1.4 ± 0.3
OH−COT-d3
8.2
196 (56) → 80 (21.5) 196 (56) → 137 (13.5)
1.4 ± 0.3
COT-GLUC
1.7
353 (48) → 177 (10) 353 (48) → 146 (27)
8.7 ± 2.6
OH−COT-GLUC
7.2
369 (36) → 193 (9) 369 (36) → 175 (22.5)
5.4 ± 1.6
compound
m/z precursor ion (capillary voltage/V) → m/z product ion (collision energy/V). bTolerances established according to 2002/ 657/EC. a
■
RESULTS AND DISCUSSION Chromatography. Two different chromatographic separation modes were considered. Given the highly polar character of these compounds, a hydrophilic interaction chromatography (HILIC) column (Luna 3 μm HILIC column, 100 mm × 2 mm, Phenomenex) was tested with water−acetonitrile eluents, adjusted with 5 mM ammonium formate buffers at two different pH values (pH 3.5 and pH 6.5). Although a good retention was obtained for both glucuronides, their peak shape was quite bad, and, moreover, the remaining compounds were poorly retained (see Figure S2 in the Supporting Information). Thus, a reversed-phase (RP) column was tested: a Synergi Fusion-RP, which is a polar-end-capped C-18 column that permits the use of 100% aqueous mobile phases and provides improved retention of polar compounds. In this case, three different modifiers in water/MeOH eluents were considered: 0.1% formic acid (which favors ionization in positive mode) and 5 mM ammonium acetate, either adjusted to pH 7 or pH 10276
dx.doi.org/10.1021/ac503330c | Anal. Chem. 2014, 86, 10274−10281
Analytical Chemistry
Article
Figure 2. Percentage of COT-GLUC remaining after the enzymatic deconjugation at different reaction times and enzyme concentrations.
incubating samples during 5 h, or with 600 or more units already with 3 h. Given the price of the enzyme it was decided to sacrifice time and reduce the amount of enzyme. Thus, incubation with 300 units of β-glucuronidase for 5 h was finally selected. Under these optimal conditions, the experiment was repeated (n = 4), the amount of free COT and OH-COT measured and the conversion yields calculated from the original COT-GLUC and OH-COT-GLUC concentrations. The conversion yields obtained were 95% ± 7% and 97% ± 6% for COT and OH-COT glucuronides, respectively. A final test was performed in order to assess the stability of the other compounds (NIC, COT, and OH-COT) in this medium. Thus, wastewater samples were spiked with 500 μg L−1 of NIC, COT, and OH-COT and, incubated for 5 h with 300 units of β-glucuronidase at pH 5. Under these conditions (see Figure S5 in the Supporting Information), no degradation was detected (Student’s t-test, 95% confidence level). Therefore, since glucuronides are quantitatively transformed to COT and OH-COT and these two metabolites are stable during deconjugation, a satisfactory back-calculation of NIC consumption can be obtained. Moreover, given these results (glucuronides will no longer be measured as themselves), the injection volume was increased to 100 μL as to decrease the limit of detection (LOD) and the LOQ (see the Chromatography section). With this method, two samples (taken on two consecutive days during the last week of August) were analyzed with and without the enzymatic deconjugation being carried out (n = 5 each). As presented in Figure 3, there was no statistical significant difference for OH−COT, but an statistically 35%−40% higher concentration of COT was measured after deconjugation (Student’s t-test, 95% confidence interval). Therefore, it is clear that the enzymatic deconjugation is required, at least in the case of COT. LC-MS/MS Performance. The matrix effects, LODs, LOQs, linearity, repeatability, and trueness of method for NIC and its metabolites determination in wastewater were studied and the results are compiled in Table 2. (Matrix effects are presented in Figure S6 in the Supporting Information.) The matrix effects were studied in the LC-MS/MS system with filtered wastewater submitted to enzymatic deconjugation and spiked at two different concentrations levels (10 μg L−1 and 50 μg L−1, taking into account nonspiked samples), and expressed as %ME as detailed elsewhere,26,27 such that 100% means no matrix effects, while lower values indicate signal suppression. As displayed in Figure S6 in the Supporting Information, COT and OH-COT showed a strong signal
Figure 1. Stability of NIC and its metabolites in sewage stored at room temperature (20 °C): (a) NIC, COT, and OH-COT and (b) COTGLUC and OH-COT-GLUC. Response normalized to time t = 0.
Enzymatic Deconjugation. Urinary metabolites such as COT-GLUC and OH-COT-GLUC are particularly susceptible to enzymatic deconjugation by β-glucuronidase.23,25 In this way, COT-GLUC and OH-COT-GLUC may be transformed to COT and OH-COT, respectively, and the total COT and OH-COT may be measured (i.e., resulting from both conjugated and nonconjugated forms). In order to obtain full deconjugation of the glucuronides, the β-glucuronidase enzyme concentration (100−1000 units) and time of incubation (2−5 h) at 37 °C were optimized. The experiments were done with sewage spiked with the two glucuronides as detailed in the Materials and Methods section. Experiments were performed twice, and a third replicate was performed in case the difference between the two replicates was higher than 20%. OH-COT-GLUC was readily deconjugated, and traces of the glucuronide (ca. 7% of the initial concentration) were only detected in the experiments carried out with 100 units of enzyme during 1 h (data not shown). COT-GLUC was however more resistant to deconjugation with the βglucuronidase enzyme. As presented in Figure 2, a complete deglucuronization of COT-GLUC (>99.9%, given the limits of quantification (LOQs) for this compound) could only be achieved at 3 h with 1000 units or at 5 h with 600 or more enzyme units However, a 99% deconjugation was considered as sufficient and it could be achieved with 300 units when 10277
dx.doi.org/10.1021/ac503330c | Anal. Chem. 2014, 86, 10274−10281
Analytical Chemistry
Article
Figure 3. Concentrations of COT and OH-COT measured in two sewage samples that were measured without enzymatic deconjugation and after enzymatic deconjugation (n = 5 replicates per sample).
Table 2. Analytical Performance Figures of the Analytical Method Relative Standard Deviation, RSD (%)b Spiked Milli-Q NIC COT OH-COT
Relative Recovery (%)c
Spiked Sewage
Spiked Sewage
R2a
5 μg L−1
100 μg L−1
10 μg L−1
50 μg L−1
10 μg L−1
50 μg L−1
LODd (μg L−1)
LOQe (μg L−1)
0.9994 0.9990 0.9991
6 3 3
3 3 4
9 7 7
7 6 7
106 95 107
96 98 112
0.3 0.05 0.1
1 0.2 0.4
From LOQ to 200 μg L−1 (n = 3 replicates). bn = 6 replicates. cRelative recoveries obtained from the internal standard calibration method, n = 6 replicates. dCalculated from sewage water for a signal-to-noise ratio of S/N = 3. eCalculated from sewage water for a signal-to-noise ratio of S/N = 10. a
Finally, the trueness was assessed with influent wastewater samples, which were spiked with analytes (10 and 50 μg L−1, n = 6), internal standards (20 μg L−1) and submitted to the deconjugation procedure. After calculating the concentrations in the spiked and subtracting the concentrations in the nonspiked samples, satisfactory trueness values in the 95%− 112% range were obtained (see Table 2) by the internal standard calibration method. Concentrations in Sewage. The occurrence of NIC, COT, and OH-COT in 21 24-h-composite influent samples, all of them collected from the same STP during 1 week in three different years, was evaluated using the method described above. As an example, Figure 4 presents a chromatogram of one of the samples (Tuesday from 2012). Mean concentrations values encountered of the three compounds are presented in Table 3. As shown, total NIC, total COT, and total OH-COT (total, because they include the corresponding glucuronides) were detected in all wastewater influent samples at concentrations in the ranges of 0.7−9.4 μg L−1, 0.3−1.9 μg L−1, and 1.0−3.3 μg L−1, respectively. As expected, OH-COT levels were higher than those of COT (Table 3), in concordance with the known metabolic fate of NIC in humans.2 Actually, their values were highly correlated (R2 = 0.926), with the OH-COT concentration being ∼1.6 times greater than that of COT (see Figure S7 in the Supporting Information) in agreement with the NIC metabolism data (see next section). However, the NIC concentrations were much higher than those of its metabolites, despite its lower metabolic excretion ratio, likely because ashes and butts are directly disposed into the sink. Actually, its concentrations did not correlate well with any of the metabolites (R2 ≈ 0.6). The concentrations reported in this work are similar than those reported in Switzerland (averages: for COT, 1.3 μg L−1;
suppression with %ME values in the range of 33%−40% and 29%−30%, respectively. In the case of NIC, the suppression was not so strong (%ME: 80%−88%). Despite these effects, still good LODs and LOQs were obtained while trueness was also satisfactory as matrix effects were compensated through the use of the deuterated internal standards (see below). Hence, LODs and LOQs were calculated for a signal-to-noise ratios (S/N) of 3 and 10, respectively, from real sewage samples, which contained 4.3 μg L−1 of NIC, 1.3 μg L−1 of COT, and 2.3 μg L−1 of OH-COT. In this way, the LODs for NIC and its metabolites were in the 0.05−0.3 μg L−1 range and LOQs were 1 μg L−1 for NIC, 0.2 μg L−1 for COT, and 0.4 μg L−1 for OH-COT (see Table 2). These values are similar to those reported by UPLC-MS/MS (0.5−0.8 μg L−1)12,13 and lower than the obtained by conventional LC-MS/MS (16−50 μg L−1),11,15 thus, permitting avoiding SPE or other enrichment procedures. Regarding the method proposed by Castiglioni et al., it also requires an SPE step, and neither LOD nor LOQ values were reported.22 Linearity was investigated by injection of standard solutions (in Milli-Q water) at eight different concentrations up to 200 μg L−1 (internal standard: 20 μg L−1 level), each level by triplicate. A good linearity was obtained with determination coefficients (R2) ranging from 0.9990 to 0.9994 (see Table 2). The precision of the method, expressed as relative standard deviation (RSD), was evaluated in different water samples at two concentration levels: in Milli-Q water spiked at 5 and 100 μg L−1 level (n = 6 for each level) and, in wastewater spiked at 10 and 50 μg L−1 level (n = 6 for each level). The real samples were deconjugated (see the Materials and Methods section) and, deuterated internal standards were added (20 μg L−1) as surrogates in all cases to compensate matrix effects. The results, in terms of RSD, varied between 3% and 9% (see Table 2). 10278
dx.doi.org/10.1021/ac503330c | Anal. Chem. 2014, 86, 10274−10281
Analytical Chemistry
Article
Figure 4. LC-MS/MS chromatogram showing the two MRM transitions of the sewage sample from Tuesday 2012, containing 2.2 μg L−1 NIC, 1.1 μg L−1 COT, and 1.8 μg L−1 OH-COT.
for OH-COT, 2.8 μg L−1)11 and much lower than in northeastern Spain (mean/median values: for NIC, 13−33 μg L−1; for COT, 2.7−5.0 μg L−1),12,13 despite the fact that glucuronides were not taken into account in those works. These differences in the concentrations may be explained by the different year of sampling (and corresponding decrease in tobacco prevalence), populations, flows arriving to the STPs, the impact of rainwater into sewage, and sampling technique. Back-Calculation of NIC and Tobacco Consumption. To estimate the amount of NIC consumed locally, detected concentrations were combined with wastewater flow rates arriving to the STP and excretion rates to calculate the mass of NIC consumed per day (g day−1) (see Table 3). From smoked NIC, 10%−15% is excreted as COT and 12%−17% is excreted as COT-GLUC;2 thus, after deconjugation, the concentration of COT would represent 22%−32%. Similarly, for OH−COT, 33%−40% of smoked NIC is excreted as free OH−COT and 7%−9% of smoked NIC is excreted as conjugated OH−COT.2 Hence, the OH−COT measured would represent 40−49%. So, the average values (i.e., 27% for COT and 44.5% for OH−COT), together with the molecular weight ratio between NIC and the deconjugated metabolite and the flow data (ca. 60 000 m3 day−1) were used to calculate the daily loads, as summarized in the following equations:
daily load of NIC consumed (g day −1) ⎛ MWNIC ⎞ ⎛ 100 ⎞ ⎟ × flow = [COT]⎜ ⎟×⎜ ⎝ MWCOT ⎠ ⎝ 27 ⎠
(1)
daily load of NIC consumed (g day −1) ⎛ MWNIC ⎞ ⎛ 100 ⎞ ⎟ × flow = [OH‐COT]⎜ ⎟×⎜ ⎝ MWOH‐COT ⎠ ⎝ 44.5 ⎠
(2)
where [COT] and [OH-COT] are, respectively, the concentrations of COT and OH-COT (in g L−1); MWNIC, MWCOT, and MWOH−COT are the molecular weights of NIC, COT, and OH-COT, respectively; and flows are expressed in terms of L day−1. As presented into Table 3, the total daily NIC consumption derived from the metabolites was very similar, with the average ranging between 222 and 263 g day−1, whereas the values obtained from the concentration of NIC itself were 5−10 higher (data not given), in agreement with Castiglioni et al.,22 since ashes and cigarettes can be directly disposed into the sewage system, as already mentioned. Finally, per capita figures were obtained from the number of inhabitants served by the STP. In this case, the number of house connections were multiplied by 2.5 to obtain an estimation of inhabitants (ca. 130 000), as advised by the 10279
dx.doi.org/10.1021/ac503330c | Anal. Chem. 2014, 86, 10274−10281
Analytical Chemistry
Article
in the Supporting Information (Table S1) and results in an average of 2.8, 2.2, and 2.4 cigarettes day−1 inhabitant−1 sold during the above-mentioned months of 2012−2014, respectively. Assuming an average of 0.8 mg of NIC being adsorbed per smoked cigarette, according to cigarette manufacturers, this translates to values of 2.3, 1.8, and 1.9 mg NIC day−1 inhabitant−1 smoked (2012, 2013, and 2014, respectively). Hence, the values obtained from sewage analysis in Santiago de Compostela are 26% lower, 11% higher, and 6% lower than those derived from cigarette sales statistics in 2012, 2013, and 2014, respectively. Generally, the data from the two last years agree quite well with retail statistics, while the lower results on 2012 may be attributed to the different smoking profile into the local area studied (Santiago de Compostela, ca. 130 000 people), as compared to the broader region where statistics are available (ca. 2.7 million people in Galicia) or differences on a short-term basis. Further investigations considering a larger set of STP and different months would be necessary in order to better compare the results of sewage analysis and sales. With regard to the week consumption profile, in contrast to other substances of abuse, such as cocaine, ecstasy, or alcohol,16−18,20,21 no higher consumption is detected during the weekend respective to the remaining week days. This can be attributed to the fact that tobacco is consumed on a daily basis and is not clearly related to recreational activities, as are illicit drugs, for instance; moreover, its urinary clearance is quite slow.30,31
Table 3. Concentrations of NIC and Its Metabolites and Derived Loads and Consumption Figuresa Concentration (μg L−1)
NIC Loadsb (g day−1)
Per Capita NIC Consumptionc (mg day‑1 person−1)
week day
NIC
COT
OHCOT
COT
OHCOT
COT
OHCOT
mean
April 2012 Tuesday Wednesday Thursday Friday Saturday Sunday Monday average 2012
2.2 0.7 1.6 1.8 2.0 2.2 3.7 2.0
1.1 0.3 0.9 0.9 1.0 1.1 1.4 1.0
1.8 1.0 1.8 1.8 1.7 1.9 2.9 1.8
252 92 204 222 218 243 326 222
220 140 238 229 203 241 373 235
1.8 0.7 1.5 1.6 1.6 1.8 2.4 1.6
1.6 1.0 1.7 1.7 1.5 1.8 2.7 1.7
1.7 0.9 1.6 1.7 1.5 1.8 2.6 1.7
March 2013 Tuesday Wednesday Thursday Friday Saturday Sunday Monday average 2013
9.4 7.0 4.8 6.3 6.8 3.8 3.0 5.9
1.9 1.5 1.1 1.5 1.4 0.8 0.6 1.3
3.3 2.5 2.0 2.6 2.8 1.4 1.1 2.3
373 310 243 309 285 183 140 263
371 280 234 309 303 180 148 261
2.7 2.3 1.8 2.3 2.1 1.3 1.0 1.9
2.7 2.0 1.7 2.3 2.2 1.3 1.1 1.9
2.7 2.2 1.7 2.3 2.2 1.3 1.1 1.9
March 2014 Tuesday Wednesday Thursday Friday Saturday Sunday Monday average 2014
3.4 4.7 5.6 5.8 6.7 6.2 6.3 5.5
1.3 1.6 1.7 1.6 1.5 1.4 1.8 1.6
2.5 3.0 2.8 2.9 2.8 2.5 2.8 2.8
219 278 259 254 243 220 287 251
232 288 242 255 248 218 247 247
1.6 2.0 1.9 1.9 1.8 1.6 2.1 1.8
1.7 2.1 1.8 1.9 1.8 1.6 1.8 1.8
1.3 1.6 1.4 1.4 1.4 1.2 1.5 1.8
■
CONCLUSIONS A new method for the determination of nicotine (NIC) and its main metabolites in sewage for further assessment of NIC consumption is presented. The use of liquid chromatography− tandem mass spectrometry (LC-MS/MS) permits affording the required limits of quantification (LOQs) (
