Introduction

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous environmental pollutants linked to a variety of adverse health effects [1–3]. Humans may be exposed to PAHs through occupation, such as in work involving diesel fuels and coal tars [4, 5], as well as through diet and other lifestyle activities (e.g., smoking) [6–8]. The urinary concentrations of PAH metabolites, specifically mono-hydroxylated PAHs (OH-PAHs), have been used as biomarkers of human exposure to PAHs [7, 9, 10].

The pioneering quantification of OH-PAHs in urine was conducted by high performance liquid chromatography (HPLC) coupled with fluorescence detection [4]. Thanks in part to technology advances of the last few decades, isotope dilution gas chromatography-mass spectrometry (GC-MS) became widely used for the determination of urinary OH-PAHs [11–16]. GC-MS showed improved accuracy, sensitivity, and precision, mainly because of the high specificity of mass spectrometry detection. However, sample preparation involved derivatization and solvent evaporation steps, and was, therefore, labor intensive and time consuming. More recently, HPLC-tandem MS (HPLC-MS/MS), introduced to measure OH-PAHs [17–26], eliminated the derivatization step and yet maintained the high specificity of mass spectrometry. Still, a relatively large volume of urine, e.g., 2–5 mL, was required in HPLC-MS/MS methods, and automated sample preparation was not fully applied [20, 21]. These conditions have limited HPLC-MS from application in large-scale biomonitoring studies when matrix volume is often limited.

For the present study, we developed a fully automatic on-line solid phase extraction coupled with isotope dilution-high performance liquid chromatography-tandem mass spectrometry (on-line SPE-HPLC-MS/MS) method for the accurate and reliable measurement of nine OH-PAHs in human urine.

Experimental section

Materials and methods

All solvents were HPLC grade, and chemicals were reagent grade. We purchased acetonitrile, ethanol, 0.1 % formic acid in water, methanol, water, and ammonium fluoride from Thermo Fisher Scientific (Waltham, MA, USA); ascorbic acid, sodium acetate, and Helix pomatia β-glucuronidase type H-1 (β-glucuronidase ≥300,000 units/g, sulfatase ≥10,000 units/g) from Sigma-Aldrich (St. Louis, MO, USA). We obtained 1-hydroxynaphthalene (1-OH-NAP), 2-hydroxynaphthalene (2-OH-NAP), 2-hydroxyfluorene (2-OH-FLU), 3-hydroxyfluorene (3-OH-FLU), 1-hydroxyphenanthrene (1-OH-PHE), 2-hydroxyphenanthrene (2-OH-PHE), 3-hydroxyphenanthrene (3-OH-PHE), 4-hydroxyphenanthrene (4-OH-PHE), 1-hydroxypyrene (1-OH-PYR), and their corresponding 13C-labeled internal standards (IS, listed in Table 1) from Cambridge Isotope Laboratories (Andover, MA, USA).

Table 1 Optimized MS parameters for nine hydroxylated PAH metabolites
Full size table

We purchased smokers’ urine samples from BioreclamationIVT (Westbury, NY, USA). We also collected urine anonymously in 2015 from non-smoker adult volunteers with no documented occupational exposure to PAHs in Atlanta, GA. We obtained two Standard Reference Materials® (SRMs), SRM 3672 (smoker urine) and SRM 3673 (non-smoker urine), from the US National Institute of Standards and Technology (NIST) (Gaithersburg, MD, USA). All urine specimens were stored upon collection or arrival at −70 °C until use.

Appropriate safety control measures (including engineering, administrative, and personal protective equipment) were used for all procedures based on a site-specific risk assessment that identified physical, health, and procedural hazards.

Preparation of standard stock solutions and quality control materials

We prepared the stock solutions of individual analytical standard in ethanol. Standards with all nine OH-PAHs were generated by serial dilution of the individual stock in 40 % ethanol/60 % water. The final concentrations of the mixed stock standards ranged from 0.08 to 200 ng/mL (1-OH-NAP and 2-OH-NAP) and 0.005 to 50 ng/mL (all other analytes). The standard solutions were aliquoted into 2 mL silanized amber glass vials and stored at 4 or -70 °C until use. The internal standard solution with 13C-labeled analytes was prepared in water with 0.2 % acetonitrile so that a 50-μL spike would result in approximate concentrations of 32 ng/mL (13C-1-OH-NAP and 13C-2-OH-NAP) or 8 ng/mL (other 13C-labeled analytes). Internal standards were aliquoted into 15 mL amber glass vials and stored at −70 °C.

Two levels of quality control (QC) materials, QC low (QCL) and QC high (QCH), were prepared by pooling urines from smokers and non-smokers. The QC concentrations were fortified, as needed, with native target compounds to encompass the ranges described for the US general population [27]. All QC materials were stored in 4 mL amber glass vials at −70 °C until used. The stability of spiked material stored at −70 °C has been previously evaluated for up to 1 year (data not shown), and no obvious degradation of OH-PAHs was observed. The QC materials stored at −70 °C for more than 1 year will be re-evaluated for their stability.

Sample preparation

Urine samples were thawed and mixed at room temperature. QC samples, reagent blanks, and standards were processed the same way as urine samples, going through all of the sample preparation steps. Sample preparation was automatically conducted on a Perkin-Elmer Staccato® System (controlled by the Perkin-Elmer iLink and Maestro software) (Waltham, MA, USA). The robotic system included six main components: Sciclone G3/G3T, Fluidx CESD-24PRO decapper, Hettich Rotanta 460 centrifuge, ThermoScienfic ALPS 3000 sealer, IVD Inheco Incubator shaker DWP, and Mitsubishi robotic arm. We programmed this system to aliquot urine samples, standards, QCs, and reagent blanks (100 μL) into a 96-well plate (Corning, NY, USA), and subsequently add ascorbic acid solution (20 μL, ∼12.5 mg/mL), internal standards solution (50 μL), and sodium acetate buffer (50 μL, ∼1 mol/L, pH 5.5) containing ∼10 mg/mL β-glucuronidase/arylsulfatase. The accuracy and precision of automatic aliquating by Sciclone robotic system were previously evaluated, and good aliquoting accuracy (recovery rate from 95 to 105 %) and precision (CV <3 %) were achieved from current procedures. Details regarding enzymatic de-conjugation were previously described [16]. The robotic system then sealed and transferred the plate for overnight incubation at 37 ± 2 °C. After enzymatic hydrolysis, the robotic system automatically added methanol (175 μL) to all sample wells, mixed the solution, resealed the plate, and centrifuged for ∼15 min at 5000 rpm (5900 rcf). Finally, the robotic system transferred 200 μL of the supernatant in each well to a new 96-well plate, and added 350 μL of water to each well before on-line SPE-HPLC-MS/MS analysis. Non-spiked and spiked synthetic urine [28] used for determining the limit of detection (LOD) was prepared as study samples.

On-line SPE-HPLC-MS/MS

The on-line SPE-HPLC-MS/MS system consisted of a Sciex 5500 or 6500 triple quadrupole mass spectrometer (Foster City, CA, USA) equipped with an electrospray ion source and controlled by AB Sciex AnalystTM software, and one Agilent 1260 pump and one degasser (Santa Clara, CA, USA), and an on-line SPE Spark Holland system (Glassboro, NJ, USA) controlled by the Sparklink® software (iChrom Symbiosis system).

After injection (300 μL), the sample was loaded onto an Oasis WAX on-line SPE cartridge with 0.1 % formic acid in water (1.5 mL), and the cartridge was washed with acetonitrile/methanol/water (∼0.4 mL, 1/1/2, v/v/v). The target analytes were eluted with methanol (350 μL) and focused on a Chromolith HighResolution RP-18 endcapped guard column (5 × 4.6 mm, Merck KGaA, Darmstadt, Germany) with the initial HPLC gradient.

We separated the target analytes on a pair of Chromolith HighResolution RP-18 endcapped HPLC columns (100 × 4.6 mm, Merck KGaA) by a programmed HPLC gradient (Table 2). The mobile phases were water with 0.1 mM ammonium fluoride (A) and methanol with 0.1 mM ammonium fluoride (B).

Table 2 Optimized LC gradient
Full size table

The ionspray voltage and source temperature were −3.0 kV and 500 °C, respectively. Curtain gas, ion source gas 1, ion source gas 2, and collision gas were 35, 50, 70, and 9 psi, respectively. The representative decluster, entrance and exit potentials were −120, −3, and −12 V, respectively. We quantified OH-PAHs by selected reaction monitoring in the negative ion mode by the ion transitions listed in Table 1 and optimized the collision energies for all ion transitions (Table 1).

Data analysis

We used Analyst (version 1.6.2, Sciex, Foster City, CA, USA) and MultiQuant (version 3.0, Sciex, Foster City, CA, USA) for data processing. We defined quality control limits and evaluated analytical runs using SAS (version 9.3, SAS Institute Inc.; Cary, NC, USA) with a multi-rule quality control approach [29].

Results and discussion

OH-PAHs were enriched and extracted from the urine matrix by on-line SPE. We separated the target analytes on a pair of monolithic RP-18 column by using a gradient of water and methanol, including ammonium fluoride (0.1 mM) in both mobile phases. We used ammonium fluoride to improve method sensitivity in the negative ion mode [30]. An example of LC-MS/MS selected ion chromatogram is shown in Fig. 1. Within 27 min, we were able to separate several pair of isomers: 2-OH-NAP, 1-OH-NAP, 3-OH-FLU, 2-OH-FLU, 1-OH-PHE, 4-OH-PHE, and 1-OH-PYR, but 2-OH-PHE and 3-OH-PHE were eluted together so we had to measure these two analytes as a sum (∑2,3-OH-PHE). 9-Hydroxyphenanthrene and 3-hydroxyfluoranthene could also be separated from the other target metabolites (data not shown).

Fig. 1
figure 1

Example chromatogram. 2-OH-NAP and 1-OH-NAP at 10 ng/mL, ∑2,3-OH-PHE at 5 ng/mL, and other analytes at 2.5 ng/mL

Full size image

Method accuracy was assessed by repeated analyses (n = 7) of synthetic urine spiked with the target analytes at three spiking concentrations. Accuracy, expressed as a percentage of recovery, was 105–113 % (level 1), 94–100 % (level 2), and 98–102 % (level 3), depending on the analyte (Table 3). Furthermore, accuracy was evaluated by analyzing two NIST SRMs: SRM 3672 and SRM 3673. The calculated OH-PAHs concentrations were in good agreement with the certified concentrations [31], and accuracy ranged from 83 to 102 %, depending on the analyte (Table 3).

Table 3 Accuracy for the measurement of hydroxylated PAH metabolites
Full size table

We determined the method precision from repeated measurements of low and high QC pools by following the CLSI protocol EP5-A2 [32] on 51 different days (two results from each of two daily runs) over a period of 8 months that involved multiple analysts. The relative standard deviations (RSDs), which reflect the within- and between-run variability, ranged from 3.2 to 12.1 % (within-run) and 4.8 to 13.0 % (within and between runs) for all analytes (Table 4).

Table 4 Precision, limits of detection (LODs), and upper linearity limits
Full size table

The LOD was determined according to procedures previously described [33] from 60 repeated measurements of non-spiked and spiked synthetic urine analyzed by multiple operators and using four different mass spectrometers (Sciex 5500 or 6500). The LODs ranged from 0.007 to 0.09 ng/mL for all analytes (Table 4), indicating the good sensitivity of the method, especially considering the relatively low volume of urine used (100 μL). The method also provided wide dynamic ranges, with upper linearity of the method at 200 ng/mL (2-OH-NAP, 1-OH-NAP), 25 ng/mL (2-OH-FLU), 20 ng/mL (∑2,3-OH-PHE), and 10 ng/mL (3-OH-FLU, 1-OH-PHE, 4-OH-PHE, 1-OH-PYR).

SPE recovery was calculated as previously described [34]. The mean recoveries of three repeated measurements were 67 ± 4 % (2-OH-NAP), 67 ± 5 % (1-OH-NAP), 81 ± 4 % (3-OH-FLU), 64 ± 2 % (2-OH-FLU), 63 ± 2 % (∑2,3-OH-PHE), 72 ± 5 % (1-OH-PHE), 55 ± 2 % (4-OH-PHE), and 49 ± 6 % (1-OH-PYR). These recoveries, which are comparable with those reported before using a GC-MS method [16], are adequate for quantitative measurement of OH-PAHs in urine from both smokers and non-smokers. We also evaluated matrix effects using a matrix factor, defined as the ratio of IS peak area in the presence of urine matrix to the IS peak area in the absence of urine matrix [34]. The matrix factor, calculated from two different QC concentrations of three repeated measurements varied from 63 to 101 %, depending on the analyte.

To validate the method, we measured OH-PAHs in 36 non-smokers and 36 self-identified smokers’ urine samples (Table 5). Among non-smokers, detection frequencies were 56 % for 1-OH-PYR, 14 % for 4-OH-PHE, 94 % for 3-OH-FLU, and 100 % for the other analytes. For smokers, detection frequencies were 97 % for 1-OH-PYR, 53 % for 4-OH-PHE, and 100 % for the rest of analytes. The geometric mean concentrations of OH-PAHs were 2.5–12.4 (average 6.6) times higher in smokers than non-smokers (Table 5). 2- and 3-OH-FLU and 1- and 2-OH-NAP showed the largest concentration differences between smokers and non-smokers.

Table 5 Geometric mean (GM) and select percentile concentrations (ng/mL) of OH-PAHs in 36 non-smokers and 36 smokers
Full size table

Compared with our previous GC-MS method [13, 16], the current approach eliminated the derivatization and solvent evaporation steps, simplified the sample processing procedure, and greatly improved throughput, while using 10 times less urine volume (0.1 vs 1.0 mL). This robust, sensitive, and highly automated on-line SPE-HPLC-MS/MS method is suitable for the quick analysis of human samples in national surveys and other large epidemiological studies to evaluate human exposure to PAHs.