MS method for urea quantitation and its application in human respiratory disease studies

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9

Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba

Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies Jianshuang Wang a,∗ , Yang Gao b , Drew W. Dorshorst b , Fang Cai c , Meire Bremer c , Dennis Milanowski b , Tracy L. Staton c , Stephanie S. Cape b , Brian Dean a , Xiao Ding a a

Department of Drug Metabolism and Pharmacokinetics, Genentech Inc., South San Francisco, CA 94080, USA Department of Bioanalytical Chemistry, Covance Laboratories Inc., Madison, WI 53704, USA c Department of OMNI Biomarker Development, Genentech Inc., South San Francisco, CA 94080, USA b

a r t i c l e

i n f o

Article history: Received 2 August 2016 Received in revised form 27 October 2016 Accepted 1 November 2016 Available online xxx Keywords: Bioanalysis Urea LC–MS/MS Volume normalization Respiratory disease

a b s t r a c t In human respiratory disease studies, liquid samples such as nasal secretion (NS), lung epithelial lining fluid (ELF), or upper airway mucosal lining fluid (MLF) are frequently collected, but their volumes often remain unknown. The lack of volume information makes it hard to estimate the actual concentration of recovered active pharmaceutical ingredient or biomarkers. Urea has been proposed to serve as a sample volume marker because it can freely diffuse through most body compartments and is less affected by disease states. Here, we report an easy and reliable LC–MS/MS method for cross-matrix measurement of urea in serum, plasma, universal transfer medium (UTM), synthetic absorptive matrix elution buffer 1 (SAMe1) and synthetic absorptive matrix elution buffer 2 (SAMe2) which are commonly sampled in human respiratory disease studies. The method uses two stable-isotope-labeled urea isotopologues, [15 N2 ]-urea and [13 C,15 N2 ]-urea, as the surrogate analyte and the internal standard, respectively. This approach provides the best measurement consistency across different matrices. The analyte extraction was individually optimized in each matrix. Specifically in UTM, SAMe1 and SAMe2, the unique saltingout assisted liquid-liquid extraction (SALLE) not only dramatically reduces the matrix interferences but also improves the assay recovery. The use of an HILIC column largely increases the analyte retention. The typical run time is 3.6 min which allows for high throughput analysis. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Respiratory diseases have been a major threat to public health. These diseases primarily affect the upper respiratory tract, trachea, bronchi, bronchioles, alveoli, pleura, pleural cavity, along with the nerves and muscles required for breathing. Disease severity ranges from mild, such as a common cold, to life-threatening entities like bacterial pneumonia, pulmonary embolism, and lung cancers. In 2013, chronic lower respiratory disease, influenza and pneumonia together accounted for more than 200 thousand deaths in the United States [1]. In human respiratory disease studies, liquid samples such as nasal secretion (NS), lung epithelial lining fluid (ELF), or upper airway mucosal lining fluid (MLF) are often collected for diagnosis. Unlike blood samples, however, the volume of such a liquid sam-

∗ Corresponding author at: 1 DNA Way, MS412a, South San Francisco, CA, 94080, USA. E-mail address: [email protected] (J. Wang).

ple cannot be directly measured in most cases. For example, in a nasopharyngeal culture test, the NS at the back of nose and throat are typically wiped with a swab and transferred to a tube containing universal transfer medium (UTM) in which bacteria, fungi or viruses are given a chance to proliferate and subsequently detected. Bronchoalveolar lavage (BAL) is a common medical technique for sampling lung ELF from the lower respiratory tract [2]. Typically, fluid (usually normal saline) is instilled through the bronchoscope and fills some of the alveolar space, and then it is aspirated through suction channel. The ELF obtained this way is always diluted. In a recently developed nasosorption technique, the upper airway MLF is collected with a strip inserted into the nasal cavity. The strip is made from the so-called synthetic absorptive matrix (SAM), and it has “wicking” properties which allow for efficient absorption of fluid with minimal invasion [3,4]. After absorption, the MLF is eluted out of the strip with a SAM elution buffer. Due to the unknown sample volume as well as significant sample dilution when using these techniques, it has been difficult to estimate the actual concentration of recovered active pharmaceutical ingredient

http://dx.doi.org/10.1016/j.jpba.2016.11.001 0731-7085/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9 2

J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

(API) or biomarkers. Therefore, an endogenous marker is needed to estimate the actual sample volume. To estimate the actual sample volume, many endogenous substrates have been proposed and evaluated, of which albumin and urea are the two most common ones [5–7]. The use of albumin has a disadvantage as the permeability of peripheral lung tissue changes with the disease state making it not suitable for inter-subject comparison [8]. In contrast, the biological concentration of urea is much less influenced by the disease state, and more importantly, urea is thought to be freely diffusible through most body compartments including the lung [9]. The use of urea to normalize sample volumes is based on the assumption that the urea concentration in samples like NS, ELF or MLF is equal to that in serum or plasma. In the nasosorption technique, the urea concentration in MLF (CMLF ) equals the urea concentration in serum (Cserum ) from the same subject. After determining the urea concentration in serum (Cserum ) and buffer eluate (Cbuffer eluate ) from those strips, the volume of MLF can be easily estimated based on simple dilution principles: V MLF = (C buffer eluate /Cserum ) × V buffer eluate The Cbuffer eluate /Cserum is defined as urea concentration ratio (R) and this number can be used for normalizing actual biological sample volumes. Limitations of using urea as a volume normalizer also remained. For example, influx of urea during BAL has been observed, and this leads to a relative overestimation of the volume of ELF [10–12], therefore the urea concentration obtained this way has to be corrected for the influx of urea. However, the method of using urea is commonly considered to be robust enough to give confidence to the general conclusions of studies. Traditionally, urea quantitation has mainly been carried out using colorimetric assays or LC-UV methods [13–15]. These methods have some common disadvantages in that they often require extensive sample preparation and are not applicable to sub␮g/mL quantitation due to the relatively higher limit of detection. Recently, development of urea quantitation methods on LC–MS platforms has been heavily investigated due to its high throughput capability and ability to detect the much lower levels of urea that cannot be detected by traditional methods. In general, when developing an LC–MS method for urea quantitation, sensitivity is not a big concern simply because the mass spectrometric detection limit is 3–4 orders of magnitude more sensitive than colorimetric methods. However, due to its very small molecular weight, urea is vulnerable to interference ions. In addition, it is a highly polar compound that rarely retains on reverse phase HPLC columns. Moreover, the endogenous presence of urea makes it not feasible to prepare a calibration curve by directly spiking urea in biological matrices. Chemical derivatization of urea with malonaldehyde bis(dimethyl acetal) (MDBMA) [16], p-dimethylaminobenzaldehyde [15], or camphanic chloride [17] have been reported for increasing selectivity and column retention, but it also required complicated sample preparation. Direct analysis of urea with normal phase column chromatography has been published [18,19], and both surrogate analyte and surrogate matrix approaches were also reported for the determination of endogenous urea [19–21], however, these methods used only one or no stable-isotope-labeled isotopologue of urea. It’s worthwhile to note that absolute quantitation of urea may not be always necessary because only the urea concentration ratio is needed for sample volume normalization. However, in certain circumstances where blood samples are not collected, such as in the analysis of exhaled breath condensate [21,22] as well as when applying the nasosorption technique in pediatric studies [4], a ratio of biomarker-to-urea is used for normalizing sample dilution. Moreover, urea quantitation is very useful for assessing kidney function, renal ischemia, urinary tract obstruction, and certain extra

renal diseases [17]. These studies necessitate absolute quantitation of urea. Here, we report a multi-matrix LC–MS/MS method for absolute quantitation of urea as a sample volume marker in human respiratory disease studies. The method was qualified in serum, plasma, UTM, SAM elution buffer 1 (SAMe1) and SAM elution buffer 2 (SAMe2). In general, the method employs two stable-isotopelabeled urea isotopologues, [15 N2 ]-urea and [13 C,15 N2 ]-urea, as the surrogate analyte and the internal standard (ISTD), respectively. This approach provides the best measurement consistency across different matrices. The analyte extraction was individually optimized in each matrix. Specifically in UTM, SAMe1 and SAMe2, the unique salting-out assisted liquid-liquid extraction (SALLE) not only dramatically reduces the matrix interferences but also improves the assay recovery. The mass spectrometer is operated in the atmospheric pressure chemical ionization (APCI) − selected reaction monitoring (SRM) mode to enhance the signal/noise ratio through minimizing the background noise. Chemical derivatization of urea is not required, and the use of an HILIC column largely increases the analyte retention. The typical run time is 3.6 min which allows for high throughput analysis. 2. Experimental 2.1. Chemicals and buffers Chemicals, unless otherwise specified, were purchased from Sigma Aldrich (St Louis, MO, USA). [13 C,15 N2 ]-urea was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Proclin 300 and ammonium formate were purchased from Thermo Scientific (Waltham, MA, USA). Type 1 water was generated by ELGA water system (ELGA Labwater, Lane End, UK). Universal transport medium (UTM) was purchased from Becton, Dickinson and Company (Franklin Lakes, NJ, USA). SAMe1 and SAMe2 were prepared according to in-house developed recipes. SAMe1 is composed of phosphate buffer saline, 1% bovine serum albumin, 0.05% Tween® 20, 0.05% Proclin 300. SAMe2 consists of 50 mM Tris HCl, 250 mM NaCl, 5 mM EDTA, 50 mM NaF, 1% NP40, 1 mM sodium orthovanadate, 0.05% Proclin 300 and 1% bovine serum albumin. 2.2. Preparation of calibration standards and quality control (QC) samples Stock solutions of urea, [15 N2 ]-urea and [13 C,15 N2 ]-urea were individually prepared in Type 1 water at a concentration of 125 mg/mL, 125 mg/mL, and 10 mg/mL, respectively. All solutions were stored at refrigerated condition (2–5 ◦ C), and the stability had been established for 32 days. Working solutions were prepared by serial dilution of stock solutions in Type 1 water. Eight calibration standards and five QC samples were prepared by spiking appropriate working solutions into each individual matrix. The levels of calibration standards and QC samples are listed in Table 1 for each matrix. 2.3. Sample preparation UTM and SAMe1: To a clean 96-well plate, 100 ␮L of calibration standard, quality control sample or study sample was added into a pre-designated well. Subsequently, 30 ␮L of internal standard solution was added into each well followed by the addition of 150 ␮L of 2 M sodium carbonate solution. The plate was covered with a mat and vortex-mixed for 2 min. After this, 900 ␮L of acetonitrile was added to all wells, and then the samples were mixed by alternately aspirating and dispensing 20 times on a Nimbus 96 liquid handling platform (Hamilton Company, MA, USA). Following the mix, the plate was sealed and centrifuged at a minimum of 1640 x g

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model PBA-10910; No. of Pages 9

ARTICLE IN PRESS J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

3

Table 1 Levels of calibration standard and quality control samples in five matrices. Matrix

Calibration standards (␮g/mL)a

Quality controls (␮g/mL)a

Serum Plasma UTM SAMe1 SAMe2

50.0, 100, 160, 250, 400, 625, 1000, 1250 50.0, 100, 160, 250, 400, 625, 1000, 1250 0.250, 0.500, 1.00, 1.60, 3.20, 6.40, 10.0, 12.5 1.00, 2.00, 4.00, 8.00, 12.5, 25.0, 40.0, 50.0 1.00, 2.00, 4.00, 8.00, 12.5, 25.0, 40.0, 50.0

50.0, 150, 375, 938, 6250 50.0, 150, 375, 938, 6250 0.250, 0.750, 4.50, 9.38, 62.5 1.00, 3.00, 7.50, 37.5, 150 1.00, 3.00, 10.0, 37.5, 150

a

Concentration is in 3 significant figures for values less than 1000 ␮g/mL.

for approximately 10 min to completely separate the organic (top layer) phase from the aqueous phase (bottom layer). 600 ␮L of organic phase was then transferred to another clean 96-well plate and evaporated to dryness with nitrogen at 45 ◦ C for approximately 20 min. The samples were reconstituted with 100 ␮L of acetonitrile and vortex-mixed for 1 min prior to LC–MS/MS analysis. SAMe2: The sample preparation in SAMe2 is similar to the one in UTM and SAMe1 with an extra cleaning step prior to acetonitrile extraction. After the addition of 150 ␮L of sodium carbonate and vortex-mixing, the plate received 600 ␮L of dichloromethane instead of acetonitrile, and then it was mixed by alternately aspirating and dispensing 20 times. After this, the plate was centrifuged at a minimum of 1640 x g for approximately 10 min to separate the aqueous phase (top layer) from the organic phase (bottom layer). The top aqueous layer was then transferred to another clean 96well plate for receiving 900 ␮L of acetonitrile followed by the rest of extraction steps described above for UTM and SAMe1 samples. Serum and plasma: The preparation procedure for serum and plasma samples is relatively straightforward. 50 ␮L of calibration standard, quality control sample or study sample was added to a clean 96-well plate followed by the addition of 30 ␮L of internal standard solution. 900 ␮L of acetonitrile was added to all wells. The plate was sealed with a mat, vortex-mixed for 2 min, and centrifuged at a minimum of 1640 x g for approximately 10 min 100 ␮L of supernatant was transferred to another clean plate where it mixed with 300 ␮L of acetonitrile. The plate was briefly vortexmixed prior to LC–MS/MS analysis.

[15 N2 ]-urea, and [13 C,15 N2 ]-urea were monitored in SRM transitions of m/z 61 → 44, 63 → 45, and 64 → 46, respectively. 2.5. Method qualification The method was qualified with regard to matrix effect, recovery, precision, accuracy, and dilution integrity. Sample stability was also evaluated. System suitability was assessed prior to each qualification run to confirm the instrument performance in terms of sensitivity, background noise level and carryover. This was done by injecting extracted LLOQ, blank, ULOQ, and blank after ULOQ samples. Additionally, solutions for analyte correction factor determination were injected at the beginning of each batch. Specifically, pure solutions of urea and [15 N2 ]-urea were prepared at an equal concentration with the internal standard, and they were alternately injected in triplicate. The mean peak area ratio was used to establish a correction factor for correcting any difference between the surrogate analyte and authentic analyte. 2.6. Data analysis Mass spectrometric data was acquired and processed using the Sciex proprietary software AnalystTM (version 1.6.2) and MultiQuant (version 3.0.1). Calibration plots of analyte/IS peak area ratio versus urea concentrations were constructed and a weighted 1/x2 linear regression was used. 3. Results and discussion

2.4. LC–MS/MS analysis

3.1. Surrogate analyte, curve range and parallelism

The LC–MS/MS analysis was carried out with a Waters (Dublin, Ireland) Atlantis Silica HILIC Column (100 Å, 3 ␮m, 3 mm × 50 mm) on a Shimadzu (Kyoto, Japan) Prominence HPLC system coupled with a Sciex (MA, USA) 4500 triple quadrupole mass spectrometer. The Shimadzu Prominence HPLC system consisted of a CBM-20A controller, a DGU-20A5R vacuum degasser, two LC-20AD solvent delivery systems, a CTO-20A/20AC HPLC column oven and an SIL20AC autosampler. Chromatographic separation was performed under gradient conditions with a mobile phase composed of 20 mM ammonium formate (Mobile Phase A) and acetonitrile (Mobile Phase B). The mobile phase was delivered at 0.6 mL/min flow rate using a linear gradient elution program from 97 to 60% of B in 1.5 min, hold at 60% of B for 0.5 min and return to the initial 97% of B at 2.2 min. The column is allowed to re-equilibrate for 0.8 min before the next injection. The column oven temperature was set to 35 ◦ C, and the autosampler was maintained at 5 ◦ C. The injection volume was 5–20 ␮L. The total runtime was 3.6 min and the analyte retention time was 1.45 min. The Sciex 4500 Triple quadrupole mass spectrometer was equipped with a Turbo VTM ion source and operated in positive APCI mode to minimize the background level in the urea channel. All parameters were optimized to achieve the best performance: nebulizer current 3 ␮A, temperature 350 ◦ C, collision gas 10, declustering potential 100 V, collision energy 26 eV. Urea,

Employing a surrogate analyte or the use of a surrogate matrix are two common approaches for quantifying endogenous compounds in biological matrices. Due to the environmental presence of urea, it is difficult to obtain a surrogate matrix that is free of urea. In all the buffer solutions (UTM, SAMe1 and SAMe2) that were investigated in this work, we observed variable levels of urea that could affect the quantitation, and therefore the use of an elution buffer alone as the surrogate matrix does not appear to be very attractive for developing the urea method. Alternatively, the approach of using a urea isotopologue as a surrogate analyte provides the best simulation of the authentic analyte especially when a second urea isotopologue is used as the internal standard. The correction factor between the authentic analyte and the surrogate analyte is generally assured after it is determined on a specific LC–MS system. It’s also worth noting that, as a volume marker, the measurement consistency is more relevant than the absolute accuracy when the cross-matrix urea concentration ratios are applied to normalize sample volumes. To this end, we set to develop the urea method using a surrogate analyte approach. [15 N2 ]-urea and [13 C,15 N2 ]-urea are two urea isotopologues (Fig. 1) that are commercially available, and in this method they served as the surrogate analyte and the internal standard, respectively. When using an isotopologue as the surrogate analyte, the isotopic effect must be evaluated at the beginning of method devel-

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9

J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

4

3.2. Instrument platform

Fig. 1. Structures of urea and its isotopologues: [15 N2 ]-urea and [13 C,15 N2 ]-urea.

opment. Our preliminary data showed that the interference from [15 N2 ]-urea or [13 C,15 N2 ]-urea was small enough and did not affect the quantitation of urea; on the other hand, for an extremely small molecule like urea which has a molecular weight of only 60, the 2- or 3-mass unit difference was actually large enough to eliminate the isotopic contribution from urea to its isotopologues. The cross-contribution between [15 N2 ]-urea and [13 C,15 N2 ]-urea, however, is significant and cannot be ignored. Naturally, [15 N2 ]-urea has 1.1% isotopic contribution to the SRM transition, m/z 64 → 46, for monitoring [13 C,15 N2 ]-urea. Due to the presence of isotopologue impurities, we also observed a small interference peak in the transition of m/z 63 → 45 for monitoring [15 N2 ]-urea when injecting [13 C,15 N2 ]-urea alone, and this small peak accounted for about 1.3% of the [13 C,15 N2 ]-urea response in its own transition, m/z 64 → 46. The cross-contribution puts a limit on the possible range of a urea calibration curve. General selectivity criteria that governs a method validation were followed here to determine the maximum curve range: 1) the blank sample extract should demonstrate a response equal or less than 20% of the LLOQ response at the retention time of the analyte; 2) the blank sample extract should demonstrate a response equal or less than 5% of the ISTD response at the retention time of the ISTD. In this case, these criteria can be expressed by the following two inequations with respect to the analyte-ISTD cross-contribution: 1.3% RISTD([13 C,15 N

2 ]−urea)

1.1% RULOQ of Analyte([15 N

≤ 20% RLLOQ of Analyte([15 N

2 ]−urea)

2 ]−urea)

≤ 5%RISTD([13 C, 15 N

2 ]−urea)

(1) (2)

Based on these two inequations, one can readily obtain ULOQAnalyte([15 N

2 ]−urea)

≤ 70 LLOQAnalyte([15 N

2 ]−urea)

(3)

The above inequation 3 indicated that the curve range of [15 N2 ]urea cannot be greater than 70 times. Luckily, the biological concentration of urea is relatively constant and allows us to set a narrow curve range. In the present work, the curve range was set to be 25 times in serum and plasma, and it was increased to 50 times in diluted matrices: UTM, SAMe1 and SAMe2. Human serum and plasma were directly used for the method development. Due to the fact that NS and MLF samples are always diluted in elution buffers, method development for these samples was actually carried out in the corresponding elution buffers: UTM, SAMe1, and SAMe2. The parallelism was evaluated by the determination of bioQCs for which the authentic analyte, urea, was spiked in each biological matrix (plasma, serum or pooled buffer-eluted NS or MLF samples from healthy donors) at concentrations spanning the corresponding curve range. The plasma or serum bioQCs were calculated against the [15 N2 ]-urea/plasma or serum calibration curve, respectively; the NS or MLF bioQCs were determined by the [15 N2 ]-urea calibration curve in each corresponding elution buffer. The precision and accuracy of bioQCs was within the typical ±15% acceptance criteria (Table 2) confirming the parallelism between the authentic analyte and the surrogate analyte in each matrix.

Due to the small molecular weight and the environmental presence of urea, significant background noise was observed in the SRM transition of m/z 61 → 44 for monitoring urea. Therefore, finding a suitable instrument platform became critical for the successful determination of urea. To this end, a series of Sciex mass spectrometers were evaluated for the assay suitability. The high level of background noise in an API 5000 or Triple Quad 5500 mass spectrometer is presumably due to the increased orifice size comparing to that in an API 4000 or Triple Quad 4500. While the increased orifice size may boost the total ion signal, it could be detrimental for a small molecular weight analyte like urea when a lot more interfering substances are introduced into the mass spectrometer and subsequently contributes to the background noise. Indeed, our preliminary data showed that an API 4000 or Triple Quad 4500 mass spectrometer had outperformed an API 5000 or Triple Quad 5500 in terms of controlling the background noise level. A Triple Quad 4500 mass spec was chosen as the ideal platform for the best balance of minimizing noise level and maintaining assay sensitivity. 3.3. Sample preparation and SALLE For serum and plasma samples, diluting samples with acetonitrile provided satisfactory results simply because the concentration of urea in these two matrices were very high, and therefore large dilution dramatically reduced the matrix suppression. The use of a Triple Quad 4500 mass spectrometer operated in the APCI mode also helped in minimizing the background noise level. A signal to noise ratio of 50 can be readily obtained at the LLOQ level which enabled the successful quantitation of urea in serum and plasma. In UTM, SAMe1, and SAMe2, however, dilution of samples did not produce satisfactory results. Both acetonitrile and methanol were evaluated as a diluent, but neither gave acceptable results due to significant matrix suppression. Support-liquid extraction (SLE) was also evaluated but failed due to insufficient recoveries that were less than 1% in most cases. The extremely low recovery is because urea is highly water soluble and the organic solvent used in the typical LLE is not polar enough to extract urea out of the aqueous phase. SLE uses water immiscible solvents, in which urea does not have good solubility. Therefore, increasing the ionic strength in the aqueous phase becomes a promising strategy as it not only decreases the solubility of urea in the aqueous phase, but also allows using extraction solvent of higher polarity, such as acetonitrile. This modified liquid-liquid extraction approach is termed as salting-out assisted liquid-liquid extraction (SALLE) [23–25]. In order to screen out the best condition for SALLE, a number of volatile and non-volatile salts were tested. The salt solution was prepared at a concentration of 2 M and mixed with samples at a 1:1 (v:v) ratio. The final salt concentration was sufficient for clear and clean phase separation. As an example, the results of recovery and matrix effect obtained by SALLE in UTM were summarized in Table 3. All salts that were tested in method development provided sufficient recovery and acceptable matrix effect, but surprisingly variable amount of interference was observed right next to the analyte peak in the chromatogram of [15 N2 ]-urea as shown in Fig. 2. This interference peak, however, was not observed in a reagent blank, suggesting that it was from the matrix and was more pronounced from certain types of salt. According to the different types of salt evaluated, no obvious trend can be traced down to correlate the extent of interference peak to the physical properties of the buffer solution (i.e.: pH, volatility, counter ions, etc.). However, from the practical point of view, 2 M sodium carbonate solution offered the best result with minimal interference peak presence, comparatively high recovery for both analyte and internal standard, along with minimal matrix effects.

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9

J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

5

Table 2 Parallelism assessment in five matrices using bioQC (concentration unit: ␮g/mL). Matrix

Mean endogenous urea conc.

Spiked urea conc.

Theoretical urea bioQC conc.

Observed urea conc.

Replicate 1

Replicate 2

Replicate 3

Replicate 4

Replicate 5

Replicate 6

Mean

Accuracy (%)

Precision (CV%)

Plasma

140.1

150 250 400

290.1 390.1 540.1

289.9 402.7 514.1

296.0 393.0 589.1

300.8 411.2 555.1

251.3 412.7 551.6

319.9 405.8 558.7

324.5 359.3 527.3

297.1 397.5 549.3

102.4 101.9 101.7

8.8 5.0 4.8

Serum

187.8

150 250 400

337.8 437.8 587.8

326.1 420.0 558.9

342.6 407.8 549.2

332.8 421.8 582.6

348.8 446.6 614.6

349.9 446.6 600.0

358.5 453.5 628.4

343.1 432.7 589.0

101.6 98.8 100.2

3.5 4.3 5.3

UTM

2.67

0.75 2.5 5

3.42 5.17 7.67

3.74 5.49 8.40

3.78 6.06 8.21

3.96 4.94 8.15

3.33 5.67 8.21

4.06 5.28 7.92

3.88 5.39 8.03

3.79 5.47 8.15

110.8 105.8 106.3

6.7 6.9 2.0

SAMe1

1.34

3 9 25

4.34 10.34 26.34

3.62 10.32 29.49

3.98 8.99 27.91

3.67 9.39 28.11

4.59 11.02 30.59

3.85 9.91 30.31

3.38 9.10 28.87

3.85 9.79 29.21

88.7 94.7 110.9

10.8 8.0 3.8

SAMe2

2.18

3 9 25

5.18 11.18 27.18

4.69 9.52 24.47

4.94 10.18 25.06

3.71 9.22 25.86

4.34 11.07 27.99

5.30 10.44 27.51

4.66 11.08 28.35

4.61 10.25 26.54

88.9 91.7 97.6

11.8 7.6 6.1

Fig. 2. Representative [15 N2 ]-urea chromatograms of LLOQ (0.25 ␮g/mL) from sodium carbonate extraction (A), and of matrix blanks from different salt extractions: sodium carbonate (B), sodium bicarbonate (C), ammonium acetate (D), ammonium bicarbonate (E), potassium chloride (F).

SALLE with 2 M sodium carbonate also succeeded in measuring urea concentration in SAMe1 samples; however, significant matrix suppression was seen when applying the same extraction protocol for SAMe2 samples. A matrix factor of 0.0256 was observed in this matrix as opposed to 1.1 in SAMe1. Q1 scan showed that a large amount of interference peaks co-eluted with the analyte (Fig. 3B) which could account for the suppression. Changing the type or concentration of salt buffer did not reduce the matrix

effect. These observations suggested that: 1) salts were likely not the source of matrix suppression; 2) some components in SAMe2 that were extracted by SALLE could be the culprit of severe matrix suppression. In order to remove those SAMe2 components that contributed to the suppression, washing samples with non-polar solvent dichloromethane and methyl tert-butyl ether (MTBE) were tested. Our results clearly demonstrated that after washing samples with dichloromethane, the interference peaks were significantly

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9

J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

6

Fig. 3. Extracted mass spectrum of Q1 scan m/z 600–1000 at the retention time of the [15 N2 ]-urea: (A) extracted reagent blank; (B) extracted SAMe2 blank sample; (C) extracted SAMe2 blank sample with additional dichloromethane wash.

Table 3 Recovery and matrix effect of [15 N2 ]-urea in UTM with different salts. Salts

Analyte recovery (mean ± SD%)

ISTD recovery (mean ± SD%)

Normalized matrix factor

2M 2M 2M 2M 2M 2M 2M 2M

40.1 ± 4.07 29.3 ± 5.40 42.3 ± 2.92 38.6 ± 3.20 41.5 ± 2.81 30.9 ± 2.41 33.1 ± 5.94 48.1 ± 5.61

37.2 ± 5.11 29.1 ± 5.33 36.5 ± 2.67 38.1 ± 2.71 41.4 ± 2.72 30.9 ± 2.55 35.0 ± 6.73 49.4 ± 3.40

1.13 0.99 1.05 1.00 1.06 0.85 0.86 1.06

Ammonium Acetate Ammonium Formate Ammonium Carbonate Ammonium Bicarbonate Sodium Carbonate Sodium Bicarbonate Monosodium Phosphate Potassium Chloride

of sodium carbonate and dichloromethane was chosen because it provided the most consistent results and created phase separation with the aqueous phase on the top, which made the subsequent supernatant transfer easier. It should be noted that sodium carbonate is a non-volatile salt, and it has the potential to contaminate the mass spectrometer source. Presumably due to a poor efficiency of extracting sodium salts by acetonitrile, no obvious signal variation or suppression was noted on the same instrument running the assay for weeks, yet it is still recommended to clean the ion source periodically to maintain the instrument performance. 3.4. Matrix effect and extraction recovery

reduced to the level comparable to the reagent blank (Fig. 3A and C), and the absolute matrix effect was dramatically bounced up to over 0.9 as shown in Table 4. In the final method, the combination

Matrix effect was evaluated by comparing extracted matrix blanks and reagent blank with the surrogate analyte added post-

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model

ARTICLE IN PRESS

PBA-10910; No. of Pages 9

J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx Table 4 Absolute matrix effect of [15 N2 ]-urea and [13 C, 15 N2 ]-urea in SAMe2 with different combinations of salt and washing solvent.

Buffer: 2 M Ammonium Formate Wash: MTBE Buffer: 2 M Ammonium Formate Wash: Dichloromethane Buffer: 2 M Sodium Carbonate Wash: MTBE Buffer: 2 M Sodium Carbonate Wash: Dichloromethane

[15 N2 ]-urea matrix factor (mean ± SD)

[13 C, 15 N2 ]-urea matrix factor (mean ± SD)

0.0953 ± 0.0080

0.0947 ± 0.0085

0.271 ± 0.0092

0.279 ± 0.0180

0.871 ± 0.121

0.846 ± 0.112

Serum Plasma UTM SAMe1 SAMe2

Matrix

Serum Plasma UTM SAMe1 SAMe2

Parameters Curve range (␮g/mL)

Slope

Intercept

R-squared

50.0–1250 50.0–1250 0.250–12.5 1.00–50.0 1.00–50.0

0.00644 0.00657 0.0276 0.0717 0.0726

0.0260 0.0214 0.0151 0.0196 0.0145

0.9973 0.9979 0.9931 0.9965 0.9947

Table 7 Summary of QC accuracy and precision in five matrices (concentration unit: ␮g/mL).

0.918 ± 0.0234

Table 5 Matrix effect and recovery of [15 N2 ]-urea in five matrices. Matrix

Table 6 Calibration curve parameters in five matrices.

LLOQ 0.927 ± 0.0205

ISTD normalized matrix factor

Recovery (mean ± SD%)

LQC

HQC

LQC

MQC

HQC

1.05 1.03 0.99 1.01 1.03

1.01 1.01 1.05 1.01 1.06

94.1 ± 10.7 72.6 ± 1.95 50.5 ± 3.07 45.1 ± 3.45 50.1 ± 2.95

119 ± 12.5 77.7 ± 3.21 49.6 ± 2.40 48.7 ± 5.07 52.3 ± 3.43

107 ± 12.1 78.9 ± 2.80 41.9 ± 1.73 41.2 ± 1.08 48.1 ± 4.06

extraction. The overall matrix effect was normalized with ISTD response, and evaluated at both low and high QC levels. It was found in all five matrices that the ISTD normalized matrix effect is approximately 1. This suggests that using the stable isotope labeled internal standard successfully normalized analyte responses in all matrices; and therefore, making the results comparable between matrices. Extraction recovery of [15 N2 ]-urea from various matrices were calculated by comparing the peak area of extracted QC samples to that of blank sample extracts post-spiked with surrogate analyte at the same concentration. Surrogate analyte recovery was evaluated at low, mid and high QC levels. A summary of matrix effect and recovery in five matrices is listed in Table 5. 3.5. Selectivity and linearity The characteristic precursor to product ion transitions, m/z 61 → 44, 63 → 45, and 64 → 46, were used as the SRM transitions to ensure high selectivity. LC–MS/MS chromatograms of the blanks (representing the endogenous level) and qualification samples were visually examined and compared for chromatographic integrity and potential interferences. The selectivity of the method was confirmed by the analysis of blank samples from 6 individual subjects and no interference peak was observed. Linear responses in the plot of surrogate analyte/IS peak area ratio versus the surrogate analyte concentration were observed in each matrix. The correlation coefficient was obtained using a 1/x2 weighted linear regression. Parameters for each calibration curve are summarized in Table 6. 3.6. Accuracy and precision The accuracy and precision of the method was evaluated at the LLOQ, LQC, MQC, HQC and Dilution QC (DQC) levels with six repli-

7

LQC

MQC

HQC

DQC

15

Nominal Conc. Intrarun meana Intrarun CV (%)b Intrarun bias (%)c n

[ N2 ]-urea QC in serum 50.0 150 50.5 156 3.0 1.5 1.0 4.0 6 6

375 386 2.3 2.9 6

938 913 2.6 −2.7 6

6250 6580 1.4 5.3 6

Nominal Conc. Intrarun meana Intrarun CV (%)b Intrarun bias (%)c n

[15 N2 ]-urea QC in plasma 50.0 150 375 50.2 152 382 2.1 1.1 2.9 0.4 1.3 1.9 6 6 6

938 904 1.8 −3.6 6

6250 6350 3.1 1.6 6

Nominal Conc. Intrarun meana Intrarun CV (%)b Intrarun bias (%)c n

[15 N2 ]-urea QC in UTM 0.250 0.750 0.276 0.799 6.6 3.1 10.4 6.5 6 6

2.00 2.11 6.9 5.5 6

9.38 9.38 4.7 0 6

62.5 63.6 4.4 1.8 6

Nominal Conc. Intrarun meana Intrarun CV (%)b Intrarun bias (%)c n

[15 N2 ]-urea QC in SAMe1 1.00 3.00 7.50 0.966 3.09 8.04 4.4 2.8 5.6 −3.4 3.0 7.2 6 6 6

37.5 38.8 3.7 3.5 6

150 164 4.4 9.3 6

Nominal Conc. Intrarun meana Intrarun CV (%)b Intrarun bias (%)c n

[15 N2 ]-urea QC in SAMe2 1.00 3.00 7.50 0.957 3.18 7.48 6.5 8.0 8.7 −4.3 6.0 −0.3 6 6 6

37.5 36.8 2.7 −1.9 6

150 153 2.7 2.0 6

Concentration is in 3 significant figures for values less than 1000 ␮g/mL. Expressed as [standard deviation/mean] × 100% in one decimal place. c Expressed as [(mean observed concentration)/(nominal concentration)] × 100% in one decimal place. a

b

cates analyzed at each level in each run. Acceptance criteria were set at ±15% bias (±20% for LLOQ) of nominal values. The mean concentrations at all levels were within the acceptance criteria with accuracy ranging from −2.7–5.3% in serum, −3.6–1.6% in plasma, 0–10.4% in UTM, −3.4–9.3% in SAMe1, and −4.3–6.0% in SAMe2, respectively. Precision (% CV) was less than 9% at all QC levels in all matrices tested. Dilution integrity was confirmed up to 100 times. All data are summarized in Table 7. 3.7. Stability The stability of [15 N2 ]-urea and urea was assessed for conditions to which samples could be exposed from collection to post-extraction. The stability of [15 N2 ]-urea was determined by analyzing prepared samples at different conditions while the stability of urea was assessed by reanalyzing incurred samples. Freeze-thaw stability: Samples at the LQC and HQC levels were subjected to six freeze–thaw cycles from storage at −70 ◦ C to thawing at room temperature. Stability acceptance criteria of ±15% was met for both [15 N2 ]-urea and urea.

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model PBA-10910; No. of Pages 9

ARTICLE IN PRESS J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

8

was developed and qualified. The method employs two stableisotope-labeled urea isotopologues as the surrogate analyte and the internal standard to ensure the best cross-matrix measurement consistency. The unique SALLE is simple and efficient for extracting urea out of the aqueous phase. The relatively short run time allows for high throughput analysis up to 400 samples per day. The results reported herein suggest this method is rugged, precise, accurate, and well-suited to support human respiratory disease studies. Conflict of interest The authors declare that there are no conflicts of interest. Acknowledgement This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Fig. 4. Spearman’s rank correlation between the urea concentration ratio, R, and the MLF weight.

Bench-top stability: Samples at the LQC and HQC levels were stored at room temperature for 24 h before being extracted and compared against fresh prepared QCs. All samples met acceptance criteria of ±15%. Long-term stability: Samples at the LQC and HQC levels were stored at −70 ◦ C for 1 week before being extracted against a fresh calibration curve and QCs. All samples met acceptance criteria of ±15%. Post-process variability: Extracted samples were stored in an autosampler at 5 ◦ C, protected from light and were reinjected 24 h after the extraction. Due to the fact that the reconstitute solution is 100% acetonitrile, evaporation of the solvent is the major concern of not going for longer test duration. The reinjected batch met the same accuracy and precision acceptance criteria as the original injection for both [15 N2 ]-urea and urea. 3.8. Method application The method has been applied to the determination of urea in UTM, SAMe1 and SAMe2 as well as in serum and plasma in support of a series of internal respiratory disease studies, and the calculated urea concentration ratio (R) was successfully used as the sample volume normalizer for correcting PK calculation or biomarker concentrations. To better assess how well the urea concentration ratio serves as the normalizer, in one set of human MLF samples (44 samples from 22 subjects) collected by nasosorption, we recorded the weight of each SAM strip before and after sampling with the weight difference corresponding to the weight of MLF sample absorbed onto each strip. The urea concentrations were measured in the SAMe1 eluate from these strips and serum samples from the same subjects, and the urea concentration ratio was calculated as R = CSAMe eluate /Cserum . As shown in Fig. 4, a strong correlation between the MLF sample weight and the urea concentration ratio can be established with a Spearman’s rank correlation coefficient () of 0.77. The MLF sample weight is a natural normalizer for correcting sample volume; however, in clinical studies it’s not feasible to obtain sample weights because most clinical sites do not have required instrument to perform such a measurement. Our results clearly demonstrated that the urea concentration ratio can be used as a volume normalizer in lieu of the sample weight. 4. Conclusion An easy and reliable analytical method for the determination of urea in a variety of matrices involved in respiratory disease studies

References [1] M. Heron, Deaths: leading causes for 2013, Natl. Vital Stat. Rep. 65 (2016) 1–95. [2] R.P. Baughman, Technical aspects of bronchoalveolar lavage: recommendations for a standard procedure, Semin. Respir. Crit. Care 28 (2007) 475–485. [3] B.L.K. Chawes, M.J. Edwards, B. Shamji, C. Walker, G.C. Nicholson, A.J. Tan, N.V. Folsgaard, K. Bonnelykke, H. Bisgaard, T.T. Hansel, A novel method for assessing unchallenged levels of mediators in nasal epithelial lining fluid, J. Allergy Clin. Immunol. 125 (2010) 1387–1389. [4] M.W. Johansson, The pros and cons of airway lining fluid composition analysis, Am. J. Respir. Crit. Care 185 (2012) 237–238. [5] K.P. Jones, J.H. Edwards, S.P. Reynolds, T.J. Peters, B.H. Davies, A comparison of albumin and urea as reference markers in bronchoalveolar lavage fluid from patients with interstitial lung-disease, Eur. Respir. J. 3 (1990) 152–156. [6] H.C. Kaulbach, M.V. White, Y. Igarashi, B.K. Hahn, M.A. Kaliner, Estimation of nasal epithelial lining fluid using urea as a marker, J. Allergy Clin. Immunol. 92 (1993) 457–465. [7] B.C. Mcgorum, P.M. Dixon, R.E.W. Halliwell, P. Irving, Evaluation of urea and albumin as endogenous markers of dilution of equine bronchoalveolar lavage fluid, Res. Vet. Sci. 55 (1993) 52–56. [8] P. Von Wichert, J. Klaus, B. Muller, W.M. Franck, Bronchoalveolar lavage-quantitation of intraalveolar fluid, Am. Rev. Respir. Dis. 147 (1993) 148–152. [9] S.I. Rennard, G. Basset, D. Lecossier, K.M. Odonnell, P. Pinkston, P.G. Martin, R.G. Crystal, Estimation of volume of epithelial lining fluid recovered by lavage using urea as marker of dilution, J. Appl. Physiol. 60 (1986) 532–538. [10] E.A. Vandegraaf, H.M. Jansen, J.A. Weber, M.G.J. Koolen, T.A. Out, Influx of urea during bronchoalveolar lavage depends on the permeability of the respiratory membrane, Clin. Chim. Acta 196 (1991) 27–40. [11] T.W. Marcy, W.W. Merrill, J.A. Rankin, H.Y. Reynolds, Limitations of using urea to quantify epithelial lining fluid recovered by bronchoalveolar lavage, Am. Rev. Respir. Dis. 135 (1987) 1276–1280. [12] S. Bayat, K. Louchahi, B. Verdiere, D. Anglade, A. Rahoui, P.M. Sorin, M. Tod, O. Petitjean, F. Fraisse, F.A. Grimbert, Comparison of Tc-99m-DTPA and urea for measuring cefepime concentrations in epithelial fining fluid, Eur. Respir. J. 24 (2004) 150–156. [13] D. Jung, H. Biggs, J. Erikson, P.U. Ledyard, New Colorimetric reaction for end-point, continuous-flow, and kinetic measurement of urea, Clin. Chem. 21 (1975) 1136–1140. [14] J. Almy, C.S. Ough, Urea analysis for wines, J. Agric. Food Chem. 37 (1989) 968–970. [15] M. Czauderna, J. Kowalczyk, Simple, selective, and sensitive measurement of urea in body fluids of mammals by reversed-phase ultra-fast liquid chromatography, Czech J. Anim. Sci. 57 (2012) 19–27. [16] Z. Quan, C. Purser, R.C. Baker, T. Dwyer, R. Bhagat, Y.H. Sheng, J.R. Leszczynski, Determination of derivatized urea in exhaled breath condensate by LC-MS, J. Chromatogr. Sci. 48 (2010) 140–144. [17] C.L. Bowen, H. Licea-Perez, Development of a sensitive and selective LC-MS/MS method for the determination of urea in human epithelial lining fluid, J. Chromatogr. B 917 (2013) 24–29. [18] H.P. Jiang, Y.H. Zhang, K.L. Yang, J. Zou, Validation of a LC-MS method for the determination of urea contamination in market teas, Food Anal. Method 7 (2014) 13–20. [19] M. Pitiranggon, M.S. Perzanowski, P.L. Kinney, D.Q. Xu, S.N. Chillrud, B.Z. Yan, Determining urea levels in exhaled breath condensate with minimal preparation steps and classic LC-MS, J. Chromatogr. Sci. 52 (2014) 1026–1032. [20] G. Folesani, M. Corradi, M. Goldoni, P. Manini, O. Acampa, R. Andreoli, G. Bertorelli, S. David, I. Franchini, Urea in exhaled breath condensate of uraemics and patients with chronic airway diseases, Acta Biomed. 79 (Suppl. 1) (2008) 79–86.

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

G Model PBA-10910; No. of Pages 9

ARTICLE IN PRESS J. Wang et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2016) xxx–xxx

[21] C.R. Esther, H.M. Jasin, L.B. Collins, J.A. Swenberg, G. Boysen, A mass spectrometric method to simultaneously measure a biomarker and dilution marker in exhaled breath condensate, Rapid Commun. Mass Spectrom. 22 (2008) 701–705. [22] C.R. Esther, G. Boysen, B.M. Olsen, L.B. Collins, A.J. Ghio, J.W. Swenberg, R.C. Boucher, Mass spectrometric analysis of biomarkers and dilution markers in exhaled breath condensate reveals elevated purines in asthma and cystic fibrosis, Am. J. Physiol. Lung C 296 (2009) L987–L993. [23] J. Zhang, H.Q. Wu, E. Kim, T.A. El-Shourbagy, Salting-out assisted liquid/liquid extraction with acetonitrile: a new high throughput sample preparation

technique for good laboratory practice bioanalysis using liquid chromatography-mass spectrometry, Biomed. Chromatogr. 23 (2009) 419–425. [24] E. Loeser, M. DelaCruz, V. Madappalli, Solubility of urea in acetonitrile-water mixtures and liquid-liquid phase separation of urea-saturated acetonitrile-water mixtures, J. Chem. Eng. Data 56 (2011) 2909–2913. [25] Y.Q. Tang, N.D. Weng, Salting-out assisted liquid-liquid extraction for bioanalysis, Bioanalysis 5 (2013) 1583–1598.

Please cite this article in press as: J. Wang, et al., Development of a multi-matrix LC–MS/MS method for urea quantitation and its application in human respiratory disease studies, J. Pharm. Biomed. Anal. (2016), http://dx.doi.org/10.1016/j.jpba.2016.11.001

9