Commercially available hydrazine (HZ) sulfate salt (P/N 16046), ¹⁵N₂-HZ sulfate (P/N 489735), p-anisaldehyde (P/N 97063), and formic acid were purchased from Sigma Aldrich (St. Louis, MO, USA). HPLC grade solvents acetonitrile, methanol and water were obtained from Fisher Scientific (Pittsburgh, PA, USA). Deionized (18 MΩ) water was used when specified.

HZ is a known mutagen and a probable human carcinogen [1–2,9]. HZ should always be handled in a chemical fume hood.

HZ sulfate and ¹⁵N₂-HZ sulfate were weighed at milligram quantities using a Mettler Toledo XP6 microbalance (Columbus, OH, USA). Hydrazine concentrations were calculated based on dissociation of hydrazine sulfate in water as previously reported. [15] ¹⁵N₂-HZ was prepared at 123 ng/mL in water, dispensed in one-time use aliquots and stored at −70 °C. HZ sulfate was dissolved in water, dispensed in one-time-use aliquots and stored at working stock solutions of 1.23–307 ng/mL HZ (or 38.4 nM-9.58 μM HZ) at −70 °C. QC high, mid and low range samples were prepared at concentrations of 2.96, 0.887, and 0.296 ng/mL HZ (or 92.3, 27.7, and 9.24 nM HZ) in human urine and stored in one-time-use aliquots at −70 °C. Pooled human urine was purchased from SPEX CertiPrep (Metuchen, NJ, USA). The urine products used were acquired from commercial sources, and the work did not meet the definition of human subjects as specified in 45-CFR 46.102 (f).

For all calibrators, 100 μL of water/acetonitrile (4/1, v/v) was added to a MultiScreen 96-well Ultracel-10 filter plate (10 kDa, MAUF1010, EMD Millipore Billerica, MA) followed by 4 μL of the appropriate working stock of HZ, 4 μL of internal calibrator stock solution of ¹⁵N₂-HZ, and 100 μL aliquot of sample matrix (pooled human urine). Using a shallow 96-well plate as a receiving plate, the filter plate was vortexed at 500 rpm for 10 seconds and then centrifuged at 3,000 × g for 30 minutes at 10 °C. Clinical samples and QC samples were processed similarly but without the addition of HZ calibrators.

To the filtrate, 200 μL of 20.5 mM p-anisaldehyde in methanol was added to each sample and allowed to react under a stream of N₂ gas in a Porvair UltraVap Blowdown Evaporator (Porvair Sciences, Wrexham Wales, UK) at 60 °C for 30 minutes. Once dry, samples were dissolved in 100 μL mobile phase acetonitrile/water (60/40, v/v), 0.1% formic acid and transferred to a 96-well PCR plate (Eppendorf, Hamburg, Germany or equivalent).

HZ concentrations in human urine were determined on an AB Sciex 4000 triple quadrupole instrument (AB Sciex, Framingham, MA, USA). The instrument was tuned and calibrated monthly over a mass range of m/z 118–2720 using an Agilent ESI tuning mixture (P/N G2421A). Conventional HPLC elution was performed using an Agilent 1200 HPLC system (Agilent, Santa Clara, CA, USA). Samples (5.0 μL) were injected onto an Agilent Zorbax SB-C18 Rapid Resolution HT column (2.1 × 50 mm, 1.8 μm) equipped with an Agilent low-dispersion in-line filter (2 μm frit). Column and autosampler temperatures were 60 °C and 10 °C, respectively. An isocratic separation of acetonitrile/water (60/40), 0.1% formic acid was used at a constant 500 μL/min flow rate and average back pressure of 200 bar. The following optimized instrument parameters were applied for the detection of the analyte and internal standard: collision gas 7 psig; curtain gas 10 psig; ion source gas 1 60 psig; ion source gas 2 60 psig; ion spray voltage 4800 V; temperature 500 °C; collision exit potential 8 V; declustering potential 50 V; entrance potential 10 V; dwell time 32.5 ms; and a ‘unit’ resolution of 0.7 amu at full width half max. Quantitation was determined by MRM (HZ quant. ion m/z 269.1 → 134.1, collision energy 28.6 V; HZ confirmation ion m/z 269.1 → 136.1, collision energy 26.9 V; ¹⁵N₂-HZ m/z 271.1 → 135.1, collision energy 28.6 V) in ESI positive ion mode.

Data acquisition and quantitative spectral analysis were carried out utilizing AB Sciex Analyst v.1.4.2 build 1346. Percent relative error was reported as %RE=[(C_e-C_t)/C_t]×100 where C_e is the experimental concentration determined from the calibration curve slope, and C_t is the theoretical concentration. Percent relative standard deviation %RSD=(SD/C_avg)×100 was calculated as a measure of assay precision, where C_avg is the average concentration calculated, and SD is the standard deviation. Peak area ratios of HZ/¹⁵N₂-HZ were plotted versus concentration to construct calibration curves of a series of six HZ urine standards. Each standard was injected (n=21) and validated over the concentration range of 0.0493–12.3 ng/mL (or 1.5–384 nM). Human urine QCs were prepared at 0.296, 0.887 and 2.96 ng/mL (or 9.2, 27.7, and 92.3 nM) and injected alongside calibrators. QC characterization was completed over the course of six weeks, with three analysts participating and no more than two curves run per day [33]. The ratio of the response for a urine sample to the response for a water sample was used to determine the method recovery in urine as compared to water.

Commercially available urine was purchased from Tennessee Blood Services (Memphis, TN, USA) in order to evaluate baseline responses. The acquired urine was from 100 random, anonymous individuals representative of a normal population. As such, the work did not meet the definition of human subjects research as specified in 45-CFR 46.102 (f).

HZ is a low molecular-weight compound, and derivatization was needed for quantitation with HPLC-MS/MS. Derivatization was carried out with p-anisaldehyde which reacts 2:1 with HZ (Figure 1a). The predominant product ions from fragmentation of derivatized HZ (m/z 269.1) were m/z 134.1 and 136.1 (Figure 1b). Quantitation of HZ was based on the transition m/z 269.1 → 134.1, and confirmation by m/z 269.1 → 136.1. The isotopically-labeled standard HZ* (¹⁵N₂-HZ) also reacts with p-anisaldehyde and was analyzed by the transition m/z 271.1 → 135.1. An isocratic separation was used for the reversed-phase HPLC separation, with a mobile phase composition of 40/60 (v/v) water/acetonitrile with 0.1% formic acid. Under these conditions, derivatized HZ was retained chromatographically for 0.98 min, with a k′ of 3.0. Calibrators were processed in pooled human urine obtained commercially. The peak signal intensity of the lowest calibrator was at least 2-fold higher than the matrix blank. Extracted ion chromatograms collected for blank urine and urine spiked with 0.0493 and 12.3 ng/mL HZ are provided in Figure 2. The pooled human urine used to prepare calibrators and QC materials should be screened prior to use to ensure that the signal intensity of the matrix blank does not exceed one-half of the signal intensity observed for the lowest calibrator of the method.

A small amount of HZ was detected in blank pooled urine, but was not observed above the lowest calibrator for the method. Table S1 includes the average peak areas (n=4) measured for blank urine, double blank urine (no internal calibrator added), water (deionized, 18 MΩ), and unfiltered water (deionized, 18 MΩ). The average peak area for blank urine was (1.31 ± 0.15) × 10⁴ whereas the average peak area for double blank urine was (1.24 ± 0.14) × 10⁴ confirming that the peak was not due to incomplete labeling of the internal standard. The average peak area for deionized water was (4.10 ± 0.78) × 10³ as compared to unfiltered deionized water with an average peak area of (4.89 ± 1.03) × 10³ confirming that the presence of hydrazine in the blank pooled urine is not from the filtration of the samples. The peak areas observed in deionized water are below the lowest reportable limit for this method. The small peak observed may be due to low amounts of hydrazine in the water, a possible source since hydrazine is used in wastewater treatment and is a byproduct of chloramine disinfection [23–27].

The peak area ratios of derivatized HZ to the derivative of isotopically-labeled HZ were linearly proportional to the ratio of the expected concentration of HZ in urine over the range of 0.0493–12.3 ng/mL. Over this linear range, the average (n=21) coefficient of determination, R² was 0.9985 ± 0.0013 with a line equation of y = (0.147 ± 0.010) x (−0.001 ± 0.004). The lowest reportable limit for the method is the lowest calibrator, 0.0493 ng/mL. This corresponds to an on-column mass of 0.25 pg based on a 5.0 μL injection volume. The intraday and interday accuracy and precision values for HZ in Table 1 were determined by calculating the % error and % RSD for the standards run as unknowns in one day (n=4) and the standards run as calibrators during the 6 week characterization (n=21). For the 6-week method characterization, the accuracies across the range fell within 15% relative error. For the intraday accuracy, the lowest calibrator (0.0493 ng/mL) had a % error of 30% at 0.0643 ng/mL. This elevated error at the low end of the curve is representative of the characterized method in which calculated concentrations of S1 must fall between 0.0390 and 0.0736 ng/mL. The acceptable range of values for the calibrator was based on the full characterization, where a value was considered ‘passing’ if it was within 2 standard deviations of the average observed [33]. Three analysts participated in a 6 week QC characterization, analyzing no more than 2 calibration curves and corresponding QCs per day. The calibrators demonstrated ≤ 14% error with a corresponding %RSD ≤ 15%. All three levels of quality control materials were evaluated with ≤ 11% RSD for the 6 week validation. The method’s accuracy and precision follow the guidelines in the FDA’s guidance for bioanalytical method validation and thus show applicability for the analysis of clinical samples [34].

To assess the recovery of HZ during sample filtration, calibrators were added to the pooled urine matrix either before filtration (“Processed”) or immediately prior to derivatization (“Unprocessed”). The “Processed” samples were representative of the sample preparation technique used for this analysis. The average peak areas (n = 4) from processed and unprocessed samples were used to calculate recovery. As shown in Table 2, recovery for both analytes was measured at low, mid, and high calibrator concentrations. The mean percentage recovery of HZ was 52 ± 9%, 45 ± 6%, and 52 ± 6%, at 0.123, 1.23 and 12.3 ng/mL, respectively. Although losses occur during the filtration recovery step, the loss is precise and normalized by isotope dilution across the linear range and therefore compliant with the FDA’s guidance for bioanalytical method validation [34]. Matrix effects were evaluated by making an injection of urine while infusing derivatized HZ [35]. No decrease was observed in the signal for the quantitation ion during the expected elution time of derivatized HZ, and no matrix effects were found to affect quantitation (Figure 3).

The stability of HZ in urine was evaluated by allowing the HZ QC materials to stand for 4, 8 and 24 hours at −20, 4 and 37 °C prior to the addition of HZ*. The average response ratios (n=3) were within 10% of the initial value for up to 24 hours at −20 °C and 8 hours at 4 °C. At 37 °C, the average response ratio dropped by greater than 15% at 4 hours. These data indicate that HZ stock solutions should be stored at −20 °C or less. Stability of HZ at the mid-level QC is provided in Figure 4, while low- and high-level QC stability is included in Figure S1. These data also indicate that HZ is not stable at human body temperature, and urine samples should be collected and frozen as soon as possible after suspected exposure. Storage effects were assessed by measuring the peak area ratios after three freeze-thaw cycles from −70 to 25 °C and were found to be stable within ± 10% of the theoretical value. Because the materials are stored in one-time use aliquots, stability up to three freeze-thaw cycles is adequate.

The ruggedness of the method was tested by assessing the changes in the following parameters: LC column temperature, collision energy, injection volume, LC flow rate, multiple column lots (n=2), and multiple MWCO filter lots (n=3). The ruggedness of the method was evaluated by comparing the calculated quality control concentration at the adjusted parameter to the calculated concentration obtained using the final method parameters. Each parameter was evaluated at a higher level and a lower level than the final method. For example, the flow rate, 500 μL/min, was also evaluated at 450 and 550 μL/min. The QC values for all ruggedness testing parameters fell within the acceptable 2σ range from the mean established during QC characterization. The collision energy was evaluated at 27.0 and 30.0 V as compared to the final method which uses a collision energy of 28.6 V, and method accuracy may be affected if the collision energy was changed significantly from the optimum value. The use of LC columns and filter plates from multiple product lots were not found to impact method accuracy.

Low amounts of HZ, 0.23 ± 0.18 ng/mL, were detected in 99 of the 100 urine samples, and HZ was detected at 40.3 ng/mL in one of the samples. Low concentrations of HZ in the urine samples could be introduced from drinking water decontamination remnants [23–27] or in the case of the 40.3 ng/mL sample, a byproduct of pharmaceutical use [13–22]. The sample containing 40.3 ng/mL HZ in urine is more than an order of magnitude less than the lower end of the range previously reported for occupational exposure which was (2.7 ± 2.0) × 10³ ng/mL. [29] Based on the same previous results, upper concentrations of occupational excretion could reach 9,000 ng/mL at three standard deviations of the average reported. Excretions exceeding 10,000 ng/mL are therefore expected in event-level exposure to hydrazine. Quantification of HZ concentrations in urine due to occupational exposure or an event-level exposure are anticipated to require a dilution of samples as described in this method. To quantify concentrations of HZ that are greater than the highest calibrator, a dilution of the sample urine with water was required. Dilution was determined to be appropriate for samples up to at least 1.23 × 10⁵ ng/mL HZ. For 1.23 × 10⁵ ng/mL, the %RE was ≤ 5% after dilution with water (Table 3). The corresponding %RSD for the 1.23 × 10⁵ ng/mL sample was ≤ 5% (n=8, over 2 days). In the convenience set of individual samples, a 1:10 dilution was needed on one of the samples, which was determined to have a concentration of 40.3 ng/mL.