Determining Urea Levels in Exhaled Breath Condensate with Minimal Preparation Steps and Classic LC–MS

Abstract

Exhaled breath condensate (EBC) provides a relatively easy, non-invasive method for measuring biomarkers of inflammation and oxidative stress in the airways. However, the levels of these biomarkers in EBC are influenced, not only by their levels in lung lining fluid but also by the volume of water vapor that also condenses during EBC collection. For this reason, the use of a biomarker of dilution has been recommended. Urea has been proposed and utilized as a promising dilution biomarker due to its even distribution throughout the body and relatively low volatility. Current EBC urea analytical methods either are not sensitive enough, necessitating large volumes of EBC, or are labor intensive, requiring a derivatization step or other pretreatment. We report here a straightforward and reliable LC–MS approach that we developed that does not require derivatization or large sample volume (∼36 µL). An Acclaim mixed-mode hydrophilic interaction chromatography column was selected because it can produce good peak symmetry and efficiently separate urea from other polar and nonpolar compounds. To achieve a high recovery rate, a slow and incomplete evaporation method was used followed by a solvent-phase exchange. Among EBC samples collected from 28 children, urea levels were found to be highly variable, with a relative standard deviation of 234%, suggesting high variability in dilution of the lung lining fluid component of EBC. The limit of detection was found to be 0.036 µg/mL.

Introduction

Respiratory diseases, such as chronic obstructive pulmonary disease (COPD), asthma and other allergic respiratory diseases, are on the rise throughout the industrialized nations (1, 2). For example, asthma morbidity has increased disproportionately among minority children in western countries, especially those residing in low income, urban communities in the USA (3–5). Measurement of biomarkers of inflammation and oxidative stress in the airways could improve our understanding of respiratory disease pathogenesis and ultimately be used to guide treatment. Several recently developed methods, e.g., fractional exhaled NO and exhaled breath condensate (EBC), have attracted attention due to their sensitivity and non-invasiveness. For example, the biomarker 8-isoprostane in EBC has been shown to be elevated with inflammatory disease states of the lung (e.g., asthma and COPD) as well as with oxidative stress-inducing exposures (e.g., endotoxin, asbestos and tobacco smoke) (6–10).

EBC involves the condensation of exhaled water droplets, including those formed in the airways, and water vapor. A key issue related to the EBC method is the dilution of the lung lining component of EBC, which contains the biomarkers of interest, by condensed water vapor. Exhaled droplets of lung lining fluid are condensed along with water vapor during tidal breathing, and the dilution of the biomarkers from the lung lining fluid by water vapor can vary by orders of magnitude between study subjects (11). An ideal dilution marker would be one that is evenly distributed throughout the body and is not directly related to lung inflammation so that its concentration in the respiratory fluid may be compared with its concentration in the plasma to determine dilution in the respiratory fluid. Several biomarkers have been proposed to correct dilution effects, including urea (12, 13), tyrosine (14), K⁺ and Na⁺ and conductivity (11, 15). K⁺ and Na⁺ measurements are used as approximation for total cations in the respiratory fluid and conductivity used as an estimate of respiratory electrolytes. One complication in using conductivity as a dilution marker is the high levels of NH₄⁺ present in the majority of EBC samples; a product of NH₃ gas in the mouth, it makes up 93 ± 3% of all cations in the EBC. Using conductivity as a dilution marker, then requires the extra steps of lyophilization and passage through an ion exchange column to remove NH₄⁺ from the samples (11). The amino acid, tyrosine, has the advantage of occurring at easily detectable levels in most EBC samples; however, more validation studies are needed. Currently, the validity of its use as a dilution marker is debated because tyrosine concentrations may vary with different states of inflammation (14).

Several characteristics of urea make it a promising dilution biomarker. First, it is evenly distributed throughout the body and readily diffuses between the plasma and airway fluids (16). Secondly, urea is not an inflammatory mediator. Though urea may be produced in the airways by some microorganisms that thrive in certain diseases (17), urea's marked diffusability should offset these microbial inputs. Thirdly, it has a relatively slow turnover in the normal human body and relatively low variation over a 24-h period within an individual (18). Fourthly, urea is relatively non-volatile and, therefore, urea in EBC likely originates from the lung lining fluid droplets rather than as an exhaled gas that dissolves in EBC.

Due to dilution by water vapor, urea concentrations in EBC are much lower that those detected in urine and blood and difficulty in detecting urea at levels typical of EBC has been reported (14). Whereas some commercially available chemical, colorimetric assay kits do not appear to be sensitive enough to detect urea in EBC (14), other methods have been more successful when used in conjunction with various forms of pretreatment and sample concentration (9, 16, 17). Effros et al. (11) measured urea concentrations based on the transformation of urea to ${\text{NH}}_4^{+}$ following treatment with urease, but this required lyophilization and passage through an ion exchange column to remove pre-existing NH₄₊ from the samples before urease treatment. Gas chromatography–mass spectrometry (GC–MS) and liquid chromatography–mass spectrometry (LC–MS) methods have been developed for the detection of urea, many of which involve a derivatization step (e.g., 19–29). Urea derivatization is a time and labor-intensive procedure and requires the use of hazardous materials. Methods used in other studies also involve a 20- to 250-fold concentration of the EBC samples (13, 30), requiring relatively large volumes of EBC samples, which typically are difficult to obtain from young subjects. Indeed, Patel et al. (31) had some difficulty getting reliable urea measurements from the small volumes of EBC they collected from infants and preschoolers. Here, we report on an approach that allowed us to detect urea without derivatization or pretreatment steps in small volumes of EBC (∼36 μL) using a classic LC–MS system.

Methods

Chemicals

Urea was obtained from Sigma-Aldrich (St. Louis, MO, USA). ¹⁵N₂-urea and ¹³C,¹⁵N₂-urea were obtained from Cambridge Isotope Laboratories (Andover, MA, USA). HPLC-grade acetonitrile (ACN) was acquired from Acros Organics (Geel, Belgium). HPLC-grade water (Fisher Scientific, Waltham, MA, USA) was used to make mobile phase B. Milli-Q water (Millipore, Billerica, MA, USA) was used in all other applications where water was needed.

Liquid chromatography–mass spectrometry

An Agilent 1100 Series pump supplied mobile phase to an Agilent 1100 Series autosampler. Mobile phase A consisted of 100% ACN, and mobile phase B was a 10-mM ammonium acetate solution containing 0.025% acetic acid (HAc) (v/v). A Dionex (Sunnyvale, CA, USA) Acclaim Mixed-Mode HILIC-1 column (5 µm, 150 × 4.6 mm) was used for separation. The column was installed with a KrudKatcher Ultra HPLC inline filter guard column (0.5 µL × 0.004 in.), as opposed to the Acclaim Holder V-2 guard column that is the manufacturer-recommended guard column for this column. The lower dead volume achieved with the Phenomenex KrudKatcher Ultra minimizes peak dispersion, thus, resulting in better peak shape. The mobile phases were introduced isocratically, with the flow held at 95% mobile phase A and 5% mobile phase B throughout the 8 min program. The flow rate was 1 mL/min. The detector was a Thermo Scientific TSQ Quantum Classic triple quadrupole mass spectrometer (MS) (Waltham, MA, USA) installed with an electrospray ionization ion source. The MS was operated in positive ion mode. The ion spray voltage was 4100 V, and the heated capillary temperature was 300°C. The sheath gas was set at 40 arbitrary units and the auxiliary gas was set at 25 arbitrary units. The m/z 61.15 center mass was set for MS determination of urea in selected-ion-monitoring (SIM) mode, and the m/z 63.15 center mass was set for MS determination of ¹⁵N₂-urea, which was used as an internal standard, in SIM. A peak width of 0.4 FWHM, scan time of 0.8 s and scan width of 0.3 m/z was used.

Sample collection and preparation

EBC samples were collected from 28 seven to eight year olds living in New York City. The average age of these subjects is 7.41, and 68% of these subjects are male. The children were participants in the NYC Neighborhood Asthma and Allergy study, a case–control study of asthma described previously (32).

The R-Tube system (Respiratory Research, Charlottesville, VA, USA) was used to collect EBC. The condensing is accomplished through an aluminum cold sleeve wrapped around a condensing tube, which provides a collection temperature of about −5°C. The mouthpiece has a two-way valve preventing inhaled air from passing through the condensing tube, but allows the individual to inhale and exhale relatively normally through the mouthpiece, without the use of nose clips. Unlike spirometry or exhaled nitric oxide the EBC sample is collected under tidal breathing conditions and does not require a forced exhalation maneuver, making it easier for children. During collection, the child is asked to breathe for 10 min through the device. Typically, this yields between 700 and 1200 μL of condensate, sufficient for pH and biomarker analyses.

Collected samples were stored at −80°C until analysis. Samples were thawed at ambient temperature, 36 µL was pipetted out and 9 µL of a 2-µg/mL solution of ¹⁵N₂-urea was added. The samples were then left to evaporate to near dryness (1–3 µL left) in a fume hood at ambient temperature following placement in a Multiblok Heater (Lab Line, Melrose Park, IL, USA) set at 30 ± 5°C for 7 h, also in a fume hood. Following evaporation, based on the volume of sample left in the tube, ACN was added to reach a total of 36 μL.

QA/QC

Standard solutions of urea were used for instrument calibration. Calibration solution concentrations ranged from 0.05 to 1.0 µg/mL for urea with constant ¹⁵N₂-urea concentrations of 0.5 µg/mL. A 95% ACN (v/v, aqueous) solution was used as the solvent for all calibration standards. The reliability of the method was also tested using repeated measurement of urea standard solutions as well as multiple aliquots of the same EBC sample. Thirty-six-microliter volumes of 0.05, 0.1, 0.2 and 0.5 µg/mL solutions of urea, dissolved in water, were each combined with 9 µL of a 2 µg/mL aqueous solution of ¹⁵N₂-urea, then subjected to the aforementioned solvent exchange procedure. The EBC samples were treated as described in the sample collection and preparation section above. All standards were tested in duplicate, except for the 0.1 µg/mL standard, which was done in triplicate.

Results and discussion

Contamination issues

Because urea is a commonly used material, we found extra care was needed to eliminate sources of urea contamination during sample preparation and analysis. For example, the caps of some 20 mL scintillation vials are made out of “white urea.” This was the major culprit of contamination during method development. Additionally, cross-contamination during lyophilization was found to be an issue, with urea levels detected at 96–142 µg/mL in half of our blank controls. It may be worthwhile to note that the freeze dryer used in this study is shared by many different research groups, and the contamination may possibly be from residues of others' samples. Contamination from the freeze dryer in our lab was eliminated by avoiding its use entirely, and, instead, a heating block was used for sample evaporation, as described previously. After pinpointing the primary contamination sources, urea levels in the blank samples were not detected or below the detection limit.

Optimization of sample preparation and LC assay

Urea detection was optimized through a careful selection of sample preparation techniques, LC hardware and MS settings. The use of a low-dead volume guard column paired with an HILIC column resulted in better, more focused peak shapes. The selected mixed-mode HILIC column is characterized by a silyl ligand consisting of both hydrophilic and hydrophobic functionalities (33). The combination of both HILIC and reversed-phase features in one column leads to the high resolution of small polar molecules, such as urea, which could not be obtained by HILIC and RP columns alone (34). We also found that operating the MS in SIM mode resulted in greater sensitivity when compared with selected-reaction-monitoring (SRM) mode; a 0.8-μg/mL urea standard run in SRM mode had an integrated peak area of ∼5600 (signal-to-noise ratio of 1376), whereas the same standard run in SIM mode had an integrated peak area of ∼34 000 (signal-to-noise ratio of 8073). This is probably due to the fact that the fragmentation efficiency of urea is low and, therefore, daughter ions are hard to detect.

In our study, ¹⁵N₂-urea was selected as the internal standard mainly due to a much lower instrument background at m/z 63.15 (the protonated molecular ion of ¹⁵N₂-urea) than 64.15 (the protonated molecular ion of ¹³C,¹⁵N₂-urea) (Figure 1). The high background around m/z 64 can be due to the formation of an ionic association between ACN⁻, from the mobile phase and the ubiquitous Na⁺ ion.

Figure 1.

Peaks for (A) 10 µg/mL¹³C,¹⁵N₂-urea and (B) 10 µg/mL ¹⁵N₂-urea.

Urea peak shape changed dramatically with the water levels in the mobile phase; the greater the proportion of ACN in the mobile phase, the narrower the peak width and the better the peak shape (Figure 2) when a mixed-mode HILIC column was used. More water in the mobile phase leads to spiky peak shapes and lower peak area response. Likewise, the organic solvent levels in samples also affect the peak shape. One microgram per milliliter of urea standards were made in different mixes of water and ACN, varying at 5% increments from 100% ACN down to 85% ACN. The standards in 100% ACN exhibited the narrowest peaks, whereas the ones in 85% ACN exhibited the greatest peak tailing, with the mixes in between (95 and 90%) showing little variation in peak shape from the 100% ACN standard. Therefore, there is a need to change phase for EBC samples, which mainly consist of condensed water vapor. Two evaporation methods, lyophilization and slow, overnight evaporation in a fume hood, were compared. A slow but incomplete evaporation (1–3 μL of sample left) process, cannot only minimize the cross-contamination issue related to freeze drying, but also has better recovery rates; the recovery rates of slow evaporation are between 88 and –132% while poorer recoveries (31–79%) with a relative standard deviation (RSD) of 36.1% were observed using lyophilization. Complete evaporation can lead to a low recovery rate as well (23–31%), suggesting urea can also evaporate if there is no water. This method, like some previously published methods, requires time for sample evaporation; however, unlike other methods, it requires a relatively small volume of EBC samples and no further treatment such as passage through an ion exchange column.

Figure 2.

Peaks for 0.800 µg/mL urea using (A) 90%ACN, 10% water in mobile phase and (B) 95% ACN, 5% water in mobile phase.

Reliability of the method

The signal responses were linear over 2–3 orders of magnitude (r² = 0.9996) with a limit of detection of 0.036 µg/mL. 0.1 µg/mL urea standards were run repeatedly between samples to confirm the reproducibility of the method and stability of the instrument. The relative standard deviation of the 0.1 µg/mL urea standard solution, over a total of nine standard injections, is 9.8%. QA/QC experiments described above gave reproducible results with relative standards of deviation <15% for the EBC aliquots and all standards.

Urea levels in pediatric EBC

In a sample size of 28 seven- to eight-year-old children, EBC urea was found to be highly variable (RSD = 234%), ranging from 0.039 to 4.190 µg/mL with an average of 0.350 µg/mL (Figures 3 and 4). These levels are comparable with those found in other studies of EBC urea in children (21). Assuming 0.2 g/L of urea in the plasma (11, 35), and thus the airway fluid, and using the following equation to calculate dilution of the airway fluid in EBC:

$${\text{Dilution}}_{\mathrm{urea}}=\frac{{\text{Urea}}_{\mathrm{airway}\mathrm{fluid}}}{{\text{Urea}}_{\mathrm{EBC}}},$$

Figure 3.

Urea concentration, as measured in EBC in 28 different children.

Figure 4.

“Typical” chromatogram of urea in EBC (0.375 µg/mL).

We find that the urea dilution factor in the 28 EBC samples from children in this study, ranges from 50- to 5000-fold. The high variability of urea indicates the highly variable content of airway secretions in EBC, as has been observed in studies using other dilution markers. This further indicates that including a measure of dilution is necessary for EBC biomarker studies.

Summary

A simple, reliable, highly sensitive LC–MS method was developed for the quantification of urea in EBC. The sensitivity achieved may be attributed to both LC hardware and MS settings. The pairing of a low-dead volume guard column with an HILIC column resulted in higher quality peaks. Additionally, a slow, gentle evaporation method was used to achieve high urea recovery after solvent exchange. We also found that in our experiment settings, operating the MS in SIM mode is much more sensitive than the SRM mode for the analysis of urea. The high variability of urea in EBC illustrates the variability of airway secretions in EBC, thus necessitating the use of a dilution marker in the analysis of biomarkers in EBC.

Funding

This research is supported by NIEHS grants (ES015905, ES014400 and ES009089). This is LDEO contribution number 7734.