Assessment of urinary 6-hydroxy-2,4-cyclohexadienyl mercapturic acid as a novel biomarker of benzene exposure

Abstract

Assessing benzene exposure is a public health priority due to its deleterious health effects and ubiquitous industrial and environmental sources of exposure. Phenyl mercapturic acid (PhMA) is a commonly used urinary biomarker to assess benzene exposure. However, recent work has identified significant interlaboratory variation in urinary PhMA concentrations related to methodological differences. In this study, we present urinary 6-hydroxy-2,4-cyclohexadienyl mercapturic acid (pre-PhMA), a metabolite that undergoes acid-catalyzed dehydration to form PhMA, as a novel and specific urinary biomarker for assessing benzene exposure. We developed and validated the first quantitative liquid chromatography-tandem mass spectrometry assay for measuring urinary concentrations of pre-PhMA. The pH effect on the method of ruggedness testing determined that pre-PhMA is stable across the normal human urine pH range and that neutral conditions must be maintained throughout quantification for robust and accurate measurement of urinary pre-PhMA concentrations. The method exhibited below 2 ng/mL sensitivity for pre-PhMA, linearity over three orders of magnitude, and precision and accuracy within 10%. Urinary pre-PhMA concentrations were assessed in 369 human urine samples. Smoking individuals exhibited elevated levels of pre-PhMA compared to non-smoking individuals. Furthermore, the relationship between benzene exposure and urinary pre-PhMA levels was explored by examining the correlation of pre-PhMA with 2-cyanoethyl mercapturic acid, a smoke exposure biomarker. The urinary biomarkers exhibited a positive correlation (r = 0.720), indicating that pre-PhMA levels increased with benzene exposure. The results of this study demonstrate that urinary pre-PhMA is a rugged and effective novel biomarker of benzene exposure that can be widely implemented for future biomonitoring studies.

Introduction

Benzene is a widespread volatile organic compound (VOC) that is classified as a Group 1 carcinogen by the International Agency for Research on Cancer (1). As the simplest aromatic hydrocarbon, benzene ranks in the top 20 most produced chemicals in the USA, serving as the chemical building block for other industrial chemicals and as a starting material for a variety of synthetic resins and plastics (1, 2). Benzene is also typically found in gasoline ~1% by volume (2, 3). Historically, occupational exposure to benzene has been significantly associated with increased incidences of aplastic anemia and acute myeloid leukemia in exposed workers (2, 4, 5). The realization of these adverse health outcomes has led to the replacement of benzene with chemical substitutes and more stringent standards for its occupational exposure (5, 6). However, outside of occupational settings, environmental exposure to benzene occurs from sources such as tobacco smoke, gasoline, automobile exhaust, attached garages, wildfires and industrial emissions (2, 7, 8).

Due to the numerous exposure sources and serious health concerns, benzene exposure has been extensively assessed in occupational and population-based studies, including many studies measuring benzene exposure biomarkers. Elevated levels of benzene exposure have been documented in firefighters (9, 10) and workers exposed to petrochemical products such as gasoline (11, 12). In non-occupationally exposed people, greater benzene exposure has been detected in urban residents (13, 14), residents adjacent to major roadways or gas stations (15, 16) and, most significantly, among smoking individuals (7, 17). In the general population, benzene exposure has been examined in large studies such as the US National Health and Nutrition Examination Survey (18), the Canadian Health Measures Survey (19), the German Environmental Survey on Children and Adolescents (20) and the Population Assessment of Tobacco and Health study (21).

In these biomonitoring studies, phenyl mercapturic acid (PhMA), a urinary biomarker of benzene, is routinely used to assess human exposure and is considered one of the most specific urinary biomarkers for benzene exposure assessment (22). Numerous analytical methods for the quantification of PhMA are reported in the literature (8, 23). However, accurately and precisely measuring PhMA is challenging. Recent work from us and other collaborators’ labs has identified substantial quantitative disagreement in urinary PhMA results when measured across five analytical methods (24). The quantitative disagreement was attributed to the presence of N-acetyl-S-(6-hydroxy-2,4-cyclohexadien-1-yl)-l-cysteine (pre-PhMA), a precursor metabolite that undergoes acid-catalyzed dehydration to form PhMA (25). Among the five analytical methods with different sample treatment pH levels (pH 6.8, 4.5, 2.9, 1.4 and 0.5), only the methods that use pH 2.9 or below had a strong correlation between results (r ≥ 0.978). However, the actual concentration was higher for methods that use lower pH treatment conditions, with the highest concentration reported for pH 0.5. This indicates that the lowest possible pH is required for the maximum conversion of urinary pre-PhMA to PhMA with no clear understanding of the optimal pH required for complete conversion (26, 27).

The dependence of urinary PhMA on the acidity of the analytical method creates challenges for its use in monitoring human exposure to benzene. A review of the literature on analytical methods for detecting PhMA reveals considerable variation in the pH of sample treatment conditions, ranging from neutral to strongly acidic conditions (26, 28–33). This variation in treatment conditions leads to incongruous urinary PhMA concentrations between analytical methods that complicate interlaboratory comparisons. From a biomonitoring perspective, a lack of consistency in the interlaboratory data reduces the impact of each exposure assessment and creates a patchwork of isolated studies on benzene exposure. Furthermore, in targeted exposure studies, the analytical method dependence of PhMA challenges the ability to contextualize analytical results by comparison to urinary concentrations observed in the general population (9, 10, 34–36) or to establish a health-based metric such as a Biological Exposure Index (9, 10). Benzene exposure assessment would benefit from a more robust exposure biomarker.

In this study, we present pre-PhMA as a novel and specific biomarker of benzene exposure that circumvents the analytical challenges presented by PhMA in exposure assessments. We developed and validated the first quantitative liquid chromatography-tandem mass spectrometry (LC-MS-MS) method for the measurement of urinary pre-PhMA. Furthermore, we determined the pH range at which the urinary pre-PhMA can be measured for the accurate and reproducible assessment of benzene exposure.

Materials and methods

Materials

OPTIMA LC-MS-grade acetic acid, acetonitrile, ammonium hydroxide, methanol and water were purchased from Fisher Scientific (USA). LiChropur LC-MS-grade ammonium acetate and ammonium bicarbonate were purchased from Sigma-Aldrich (USA). N-Acetyl-S-(6-hydroxy-2,4-cyclohexadien-1-yl)-l-cysteine (pre-PhMA; catalog no. 79–5544) and its internal standard N-acetyl-S-(6-hydroxy-2,4-cyclohexadien-1-yl)-[¹³C₃-¹⁵N] (pre-PhMA-[¹³C₃-¹⁵N]; catalog no. 79–5760) were custom synthesized by ASI Chemicals (Cheyney, PA, USA) and are now commercially available.

LC-MS-MS analytical method

The pre-PhMA biomarker was incorporated into a previously reported LC-MS-MS analytical method that utilizes a simple dilute-and-shoot sample preparation to quantify mercapturic acid biomarkers of VOC exposure in human urine (28). Briefly, 50 μL calibrators, quality control (QC) materials and urine specimens were diluted to 500 μL with 25 μL working internal standard and 425 μL 15 mM aqueous ammonium acetate. Diluted solutions were homogenized, and 2 μL sample aliquots were injected onto the LC-MS system The chromatographic separation was performed with an Acquity I-Class LC system (Waters Corporation, Milford, MA, USA) equipped with Acquity HSS T3 (15×2.1 mm, 1.8 μm) LC column (Waters Corporation, Milford, MA, USA). The column operating temperature was 40°C. The weak wash was LC-MS-grade water. The strong wash was a solution of equal parts by volume of LC-MS-grade water, methanol, acetonitrile and isopropanol. Gradient elution was performed with 15 mM ammonium acetate (pH 6.8) in water as mobile phase A and acetonitrile as mobile phase B. See Supplementary Table SI for additional gradient details. The LC system was coupled to a 5,500 triple quadrupole mass spectrometer equipped with an electrospray ionization source (Sciex, Framingham, MA, USA). Pre-PhMA was detected in negative ion mode with the transitions and mass spectral parameters described in Table I.

Table I.

Compound-Specific Mass Spectrometric Parameters

		Ion transition (m/z)
Analyte	Retention time (min)	Quantitation	Confirmation	DP	EP	CE	CXP
pre-PhMA	4.40	256.0→109.0		−50	−10	−18	−9
		238.1→109.1		−50	−10	−22	−9
pre-PhMA-[13C3-15N]	4.40	260.0→109.0		−50	−10	−18	−9

DP = declustering potential, EP = entrance potential, CE = collision energy, CXP = cell exit potential. All units are in V.

Standard solutions

Separate master stocks of pre-PhMA were prepared by dissolving neat standard material in 10 mM ammonium bicarbonate in water at a concentration of 114 μg/mL. Ammonium bicarbonate was used for its buffering capacity at neutral pH conditions. Master stock was diluted with neutral pH water to make working calibrator stocks. Working calibrator concentrations ranged from 0.8 to 800 ng/mL with spacing between calibrator levels formulated to be a factor of sqrt(10) lower than the adjacent higher calibrator level. Master stock and working calibrator stock were stored at −70°C in a freezer prior to use. Working calibrator stocks were freshly diluted 10 times with 15 mM ammonium acetate in water (pH 6.80) to make the prepared calibrators to be assayed with the described analytical method. A set of seven calibrators was analyzed with each set of unknowns. A weighted 1/x (where x is the standard concentration) least-square model was fit to the calibration. Calibration curves were linear with r² > 0.99. No carryover was observed for pre-PhMA at the highest calibrator levels, indicating that the weak and strong washes of the established assay were sufficient. The use of water-based calibrators in place of matrix-matched calibrators was validated by confirming that calibrators prepared in ammonium acetate buffer had a similar slope (within 5% difference) as calibrators prepared in pooled human urine. This testing was performed across three separate days.

Method accuracy and precision

The method accuracy was assessed by examining the percent recovery of pre-PhMA spiked in two separate pools of anonymous human urine across two separate days. Urine pools were spiked at 8.0, 25.3 and 80.0 ng/mL, and the unspiked urine pools and spiked urine pools were prepared in triplicate and analyzed by the assay. Results were calculated for the three concentrations after subtracting unspiked background concentrations. The precision of the method was assessed by examining the repeated measurement of low concentration (QC Low) and high concentration (QC High) QC materials. QC material was prepared with urine fortified with known amounts of pre-PhMA. QC Low and QC High were prepared at 10 and 100 ng/mL pre-PhMA, respectively. Measurements spanned 20 independent runs over a more than 2-month period. Two QC Low and two QC High were analyzed in each analytical run. Within-run and between-run precisions were calculated.

Limit of quantitation

The analytical method limit of quantitation (LOQ), the lowest quantifiable concentration, was determined using Taylor’s method by running freshly prepared low-concentration calibrators across 3 days on three separate LC-ESI-MS-MS instruments (37). The LOQ was calculated as 10S₀, where S₀ is the extrapolated standard deviation at zero concentration from the standard deviation versus concentration plot.

Pre-PhMA stability testing

The analyte stability in spiked urine pools at 10 and 100ng/mL and prepared in triplicate was examined to determine optimal handling and storage conditions. Bench-top stability at room temperature was assessed for 8 h to represent typical sample handling during sample processing. Processed sample stability was evaluated on prepared samples stored in the autosampler at room temperature for 24 h. Freeze-thaw stability was assessed for five freeze-thaw cycles. Samples were stored at −70°C for at least 24 h between each cycle and then thawed for sample preparation.

Measuring pH-dependent stability of pre-PhMA

The pH stability of pre-PhMA was measured to examine its suitability as a biomarker of benzene exposure in human urine. Buffered urine solutions were prepared at 12 pH levels ranging from pH 3.2 to pH 10.9 by diluting 100 mM aqueous ammonium acetate solutions 10× with pooled non-smoking urine (i.e., 1:9 100 mM ammonium acetate:pooled non-smoking urine). Desired pH levels for the buffered urines were achieved by adjusting pH with LC-MS-grade acetic acid or ammonium hydroxide. The pH of each buffered urine was confirmed using a calibrated pH meter (Mettler-Toledo, Columbus, OH, USA). After this preparation, the buffered urine solutions were spiked with an equal amount of pre-PhMA standard and were allowed to react at room temperature for 4 h. A neutral pH water matrix sample was also spiked with equivalent pre-PhMA and stored with the spiked buffered urine samples to act as a control. After 4 h, each buffered urine was quenched by adjusting the pH level to be at least pH 6 or greater with LC-MS-grade ammonium hydroxide. Samples were analyzed as unknowns in triplicate using the previously described analytical method.

Application to human specimens

A set of 369 anonymized spot human urine samples were purchased from Tennessee Blood Services (Memphis, TN, USA) and analyzed for pre-PhMA and 2-cyanoethyl mercapturic acid (2CyEMA) using the described analytical method. All urine samples were stored at −70°C prior to analysis. Urinary concentrations of PhMA were also measured in a subset of 100 samples using a separate analytical method (30). Urinary results below the LOQ for these analytes were imputed with their corresponding LOQ/sqrt(2). LOQ was used to impute data since it is the lowest reportable value of the assay for pre-PhMA. Urinary results were logarithmically transformed for the Student’s t-test at 95% confidence interval (CI) and the Pearson correlation to meet the statistical expectation of normalized data.

Alternative selectivity LC-MS method confirmation

Two alternative selectivity chromatographic separations were developed and performed on an ACE Excel CN-ES column (150×2.1 mm, 2 μm) and on an ACE Excel C18-PFP (150 × 2.1 mm, 2 μm). Both methods engaged a gradient with 15 mM ammonium acetate (mobile phase A) and methanol (mobile phase B) at a column temperature of 40°C. On the C18-PFP column, the gradient proceeded as follows: 0.250 mL/min; initial, 97% A; 2.0 min 95% A; 3.0 min 50% A; 5.0 min 30% A; 6.5 min 20% A and 7.0–9.0 min 97% A, held for re-equilibration. The gradient for the CN-ES column matched the gradient used with the developed assay for pre-PhMA (Supplementary Table SI). The sample manager temperatures were set to 25°C. Similarly, MS detection of pre-PhMA followed the validated assay (Table I). No true method validation was performed on these alterative selectivity LC-MS assays due to the simple confirmatory nature of their use. The rapidly developed assays quantified pre-PhMA in a subset of 96 human urine samples previously analyzed for pre-PhMA. Analytical results generated by the alternative assays were compared to results from the validated assay using conventional analytical method comparison tools of correlation-regression analysis and Bland-Altman analysis (38).

Data analysis

All LC-MS-MS data were generated in Analyst 1.7 (Sciex, Framingham, MA, USA) and processed in MultiQuant 3.0.3 (Sciex, Framingham, MA, USA). Statistical analysis was performed in JMP (SAS Institute, Cary, NC, USA). Graphical presentations were created in GraphPad Prism 9.0 (San Diego, CA, USA).

Results and discussion

Analytical method and method validation

LC-MS-MS analysis

We assessed the feasibility of urinary pre-PhMA (Supplementary Figure S1) as a biomarker of exposure to benzene and validated the first analytical method for quantifying this novel biomarker in human urine. The analytical method employs a simple dilute-and-shoot sample preparation that involves a 10-fold dilution of urine samples in 15 mM aqueous ammonium acetate (pH 6.8) prior to the reversed phase LC-ESI-MS-MS analysis.

For MS-MS detection, we optimized the ion transitions of pre-PhMA to quantify it selectively and accurately in human urine. The observed ion transitions were m/z 256→109, 238→109 and 256→238 (Table I). Out of the three observed ion transitions, m/z 256→109 and 238→109 were the predominant transitions with near equivalent intensity. The ion transition m/z 256→109 was chosen as the quantitative transition, and m/z 238→109 was selected as the qualitative transition. The ion transition m/z 256→238 represents the loss of a water molecule from pre-PhMA and therefore was not used in the analytical method. The fragmentation pattern of pre-PhMA and a representative chromatogram of a smoking individual urine sample are presented in Figure 1.

Figure 1.

Extracted ion chromatograms of pre-PhMA in a urine specimen from a representative smoking individual with (A) quantitative, (B) qualitative and (C) internal standard ion transitions extracted. The mass spectral fragmentation (D) for each transition.

Method validation

Typical method validation experiments were performed to validate the quantitative analysis of urinary pre-PhMA with the existing LC-MS-MS assay. The method accuracy was assessed by spiking known amounts of pre-PhMA into human urine specimens at three concentrations spanning the calibration range. The percent error across concentrations and urine matrices was within 8%. The method precision was assessed throughout method validation by repeatedly measuring the pre-PhMA result for low and high QC material twice in each analytical run. The within-run coefficients of variation (CVs%) were calculated to be 3.4% and 2.3% for QC Low and QC High, respectively. The between-run CVs% were calculated to be 9.1% and 7.4% for QC Low and QC High, respectively. The sensitivity of the analytical method was determined for pre-PhMA using Taylor’s method to determine LOQ (37). The analytical LOQ was determined to be 1.91 ng/mL in human urine.

The influence of matrix effects on the quantitative measurement of pre-PhMA was examined through the validation of non-matrix-matched calibrators and spike accuracy in matrix. Validation of the non-matrix-matched calibrators was performed by comparison of the slope of calibrators prepared in 15 mM ammonium acetate against the slope of calibrators prepared in pooled urine matrix. Across three separate run days, the average difference between calibrator slopes was 3.1%, indicating that there was no matrix-related modification of ionization for pre-PhMA (39). Furthermore, the accuracy testing via spiking known amounts of pre-PhMA into human urine matrices was another tool to examine the influence of matrix effects on the quantitative capability of the assay. The results, reported earlier, demonstrate that no significant bias was observed with the urine matrix. The insignificant influence of matrix effects on the assay is largely expected due to the use of a ¹³C-/¹⁵N-labeled internal standard that can properly account for matrix-to-matrix differences (40).

The stability of pre-PhMA under various typical sampling handling and storage conditions was assessed in spiked human urine pools at two different concentrations. All temporal stability experiments reported here were performed under normal urine pH conditions at which pre-PhMA exhibits stability (refer to the section “pH effect on pre-PhMA stability”). Bench-top stability at room temperature for 8 h was assessed to confirm that pre-PhMA exhibited sufficient stability to be measured in a typical laboratory setting. The calculated percent error for the bench-top stability was within 8.25% for both concentrations with reference to the control (i.e., t = 0 h). The processed sample stability was then tested to confirm that pre-PhMA exhibited sufficient stability in the processed sample state for 24 h at room temperature in the LC autosampler. The percentage errors for the processed sample stability were within 2% across concentrations. The percentage errors for five freeze-thaw cycles were 5.7% and 11.1% for the low and high concentration ranges, respectively. The stability testing results indicate that pre-PhMA exhibits sufficient stability to be a suitable biomarker to be included in this existing assay.

pH effect on pre-PhMA stability

A critical step in the analytical method validation for pre-PhMA was to assess its pH stability. It has been known that pre-PhMA converts to PhMA under acidic conditions, but no research to our knowledge has yet systematically explored the stability of pre-PhMA across the pH spectrum. This information is particularly important for two reasons: (i) pre-PhMA is a urinary metabolite and the pH of normal human urine typically ranges from pH 4.5 to pH 7.8 (41) and (II) analytical methods quantifying pre-PhMA must avoid converting pre-PhMA to PhMA. Thus, as part of the ruggedness testing in the analytical method validation, the stability of pre-PhMA in urine was examined from pH levels roughly spanning pH 3 to pH 11, including the entire physiological urine pH range.

The results shown in Figure 2 indicate that pre-PhMA is largely stable within the physiological urine pH range with only insignificant loss at the acidic extremity of the range. At the most acidic physiological pH (pH 4.5), the pre-PhMA recovery remained relatively high (94.5%). At pH ≥5, no measurable loss of pre-PhMA was detected. However, substantial instability was observed for pre-PhMA at pH <4.5. The recovery of spiked pre-PhMA progressively decreased as urine conditions became more acidic with the lowest recovery of 7.8% at the overall most acidic test condition (pH 3.2). The observed stability of pre-PhMA across the physiological urine pH range up to at least pH 11 indicates that neutral pH conditions are required for any robust analytical method to accurately and reproducibly quantify pre-PhMA for benzene exposure assessment.

Figure 2.

The stability of pre-PhMA was measured as a function of urine pH from pH 3 to pH 11. Vertical bars indicate the lower (pH 4.5) and upper (pH 7.8) ends of the physiological urine pH range. The data indicate that pre-PhMA is stable under physiological urine pH conditions. Under pH 4.5, a significant decline in pre-PhMA stability is observed.

Application to human specimens

Pre-PhMA as a biomarker of benzene exposure

The performance of pre-PhMA as a biomarker of benzene was assessed by measuring its concentration in 369 human urine samples and examining how the results correlate with expected benzene exposure. Table II presents the summary statistics for the urinary pre-PhMA results, and Supplementary Table SII provides individual urinary results. In the general population, tobacco smoke is one of the primary sources of benzene exposure; smoking a pack of cigarettes a day leads to inhalation of low milligram quantities of benzene (42). Elevated benzene exposure in smoking individuals (SM) compared to non-smoking individuals (NS) is routinely observed with blood benzene and other urinary biomarkers (17, 36, 43). Thus, we examined pre-PhMA concentrations in the urine of SM and NS. The smoking status was classified based on the urinary concentration of smoke exposure biomarker 2CyEMA (42, 43). Overall, SM had significantly higher urinary pre-PhMA concentrations compared with NS (P <0.0001) when values below LOQ were imputed by LOQ/sqrt(2). Out of 40 NS specimens, only two exhibited detectable levels of pre-PhMA, whereas the majority of SM specimens contained quantifiable levels of pre-PhMA (Table II). The lower detection frequency in NS may likely be attributed to the small sample size of individuals who may not have been exposed to benzene through other environmental sources. Furthermore, this outcome is as expected given the biomarker’s high specificity for benzene exposure and the absence of other known endogenous or exogenous sources of pre-PhMA. However, improving the method LOQ by implementing sample preconcentration in future studies may enhance the ability to examine trace levels of pre-PhMA in human urine.

Table II.

Summary Statistics of Pre-PhMA (ng/ml) in 369 Urine Samples Classified by 2CyEMA (≥7.32ng/mL) Determined Smoking Status

Sample type	N	Geometric mean	Upper range	Detection frequency (%)
Non-smoking	40	NRa	2.23	5.0
Smoking	329	8.95	378	89.7

^aNot reported: the proportion of results below the limit of detection was too high to provide a valid result.

Furthermore, we examined the relationship between urinary 2CyEMA and pre-PhMA concentrations. A correlation analysis between log₁₀-transformed 2CyEMA and pre-PhMA concentrations (Figure 3) resulted in a Pearson correlation coefficient (r) of 0.720, indicating that the two biomarkers are positively correlated. Since urinary 2CyEMA is strongly correlated with tobacco smoke exposure, this positive correlation demonstrates that pre-PhMA levels increase with greater benzene exposure from tobacco smoke (44–47). Notably, this exposure analysis has some limitations. The analysis does not account for other sources of benzene exposure such as gasoline or industrial emissions. Furthermore, recent cannabis smoke exposure has been shown to increase urinary 2CyEMA levels, but additional research is needed to better understand the extent of benzene exposure from cannabis smoke (48, 49). Additionally, demographic variables were not included due to the relatively small sample size. Future measurement of pre-PhMA in population-based exposure studies such as the National Health and Nutrition Examination Survey and targeted cannabis exposure studies should provide a more thorough understanding.

Figure 3.

Correlation analysis (N = 369) between 2CyEMA and pre-PhMA resulted in a Pearson correlation coefficient of r = 0.720. 2CyEMA is a selective smoke biomarker that has been used to delineate smoking from non-smoking individuals at a cutoff of 7.32 ng/mL. The dashed line is the log₁₀[2CyEMA, ng/mL] cutoff at 0.865.

Comparative study of urinary pre-PhMA to PhMA

Since pre-PhMA is a precursor of the widely used benzene exposure biomarker PhMA, we compared pre-PhMA results with PhMA concentrations measured with a separate LC-MS-MS method that uses an acidic (pH 2.90) sample diluent (30). A 100-sample subset of the samples was assayed for urinary PhMA. A comparison of the molar concentrations of pre-PhMA and PhMA indicates quantitative disagreement between the magnitudes of analytical results for the biomarkers. A linear regression analysis (Figure 4) of pre-PhMA vs. PhMA resulted in a slope of 5.34 (95% CI: 5.06–5.63). The slope of the regression line being >1 indicates that, of the same urine samples, the urinary pre-PhMA concentrations were higher compared to the corresponding PhMA concentrations. The greatly elevated urinary pre-PhMA concentrations compared to PhMA concentrations are likely explained by pH factors associated with the analytical measurements of each biomarker. According to the pH stability profile (Figure 2), the urinary pre-PhMA measurement of this study represents a total assessment of its urinary concentrations. In contrast, the PhMA measurement likely represents a partial assessment of urinary PhMA concentrations since only weakly acidic conditions (pH 2.90) imparted by the analytical method resulted in the partial conversion of pre-PhMA to PhMA. Previous research has determined that dehydration of pre-PhMA with strong acids results in higher PhMA concentrations, but there appears to be significant uncertainty about the dehydration conditions for total and reproducible measurement of urinary PhMA (25–27). Ongoing research in our laboratory with the pre-PhMA analytical standard seeks to understand the proper reaction conditions to form a total and reproducible PhMA measurement. Nonetheless, we believe that pre-PhMA reduces analytical variation between laboratories that is experienced by PhMA and therefore may be the preferred biomarker for future biomonitoring studies.

Figure 4.

Regression analysis of urinary pre-PhMA (nmol/mL) vs. PhMA (nmol/mL) resulted in a slope (5.34) that indicates that urinary pre-PhMA concentrations are significantly higher than PhMA concentrations.

Confirmation of urinary pre-PhMA concentration

Although the urinary pre-PhMA results were strongly correlated with the PhMA results, the significant quantitative disagreement warranted additional investigation. Two additional LC-MS-MS methods with alternative LC column selectivity were rapidly developed to assess the novel pre-PhMA urinary results. The alternative chromatographic separations were performed on cyano-substituted and pentafluorophenyl-substituted reversed-phase LC columns that offer various physicochemical interactions in addition to general hydrophobic interactions, such as dipole-dipole, hydrogen bonding, shape selectivity and π-π (50). This added selectivity was in contrast to the original pre-PhMA measurements that were made using an established traditional C18 reversed-phase chromatographic separation with no optimization or adjustments for pre-PhMA. After the selection of the LC columns, the mobile phase B was changed to methanol from acetonitrile, and the chromatographic separations were assessed to ensure sufficient analyte retention and peak shape. Representative chromatograms from the three assays of pre-PhMA in the urine of a representative smoking individual are shown in Supplementary Figure S2. Notably, the cyano-substituted separation partially resolved the chiral isomers of the pre-PhMA analytical standard with one of the isomers being predominant in urine samples. Once developed, the adjusted pre-PhMA assays were applied to a 96-sample subset of previously analyzed human specimens (Supplementary Table SIII), and the analytical results were compared to the pre-PhMA results generated by the validated assay (Supplementary Figure S3). The analytical method comparisons found that both alternative selectivity assays generated statistically equivalent urinary pre-PhMA results compared to the validated pre-PhMA assay. The regression line analyses exhibited slopes and y-intercepts that included 1.0 and 0.0, respectively, in the 95% CIs. Furthermore, the Bland-Altman biases were not significantly different from 0%. The measurement of urinary pre-PhMA with alternative chromatography adds confidence to the novel urinary pre-PhMA results that were significantly elevated relative to PhMA.

Data availability

The data underlying this article are available in the article and in its online supplementary material.