Field-Free Atmospheric Pressure Photoionization ... - ACS Publications

Mar 6, 2017 - A comparison study is presented in which the relative performance of a new orthogonal geometry field-free atmospheric pressure ...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Field-Free Atmospheric Pressure Photoionization−Liquid Chromatography−Mass Spectrometry for the Analysis of Steroids within Complex Biological Matrices Ross D. McCulloch* and Damon B. Robb University of British Columbia, Department of Earth and Ocean Sciences, 2049-2207 Main Mall, Vancouver, British Columbia V6T 1Z4, Canada ABSTRACT: A comparison study is presented in which the relative performance of a new orthogonal geometry field-free atmospheric pressure photoionization (FF-APPI) source was evaluated against both electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) for the analysis of a small panel of clinically relevant steroids, spiked within various complex biological matrices. Critical performance factors like sensitivity and susceptibility to matrix effects were assessed using a simple, isocratic, high-throughput LC-MS workflow. FF-APPI was found to provide the best performance in terms of both sensitivity and detection limit for all of the steroids included in the survey. Order-of-magnitude sensitivity advantages were realized for some low polarity analytes including both estradiol and estrone. A robust linear regression, post extraction addition method was used to evaluate the relative impact of matrix effects upon each ionization method using protein precipitated human serum, plasma and Surine (simulated urine) as standard clinical matrices. Under conditions optimized for sensitivity, both the field-free APPI and APCI sources were found to provide similarly high resistance to matrix suppression effects, while ESI performance was impacted the most dramatically. For the prototype FF-APPI source, a strong relationship was established between optimizable source parameters and the degree of ion suppression observed. Through careful optimization of vaporization temperature and nebulizer gas pressure it was possible to significantly reduce or even eliminate the impact of matrix effects, even for high-throughput LC-MS methods.

A

ion−molecule reactions must occur within the open source environment within the direct vicinity of the MS sampling orifice.15 By contrast, in field-free APPI devices, an extended reaction region is provided. This region is thought to impact analyte sensitivity through improved confinement of the ion beam, as well as an increase in the time provided for ion− molecule reactions to proceed. Ion transmission may also be aided by the unique flight tube geometry provided by FF-APPI sources, potentially impacting the field characteristics responsible for guiding ions toward the MS interface. Regardless of the mechanisms, the performance benefits associated with the enclosed field-free geometry are noteworthy, suggesting that the APPI technique may not yet have reached its full potential. Sensitivity and detection limits typically drive the development and acceptance of new analytical technologies; however, other characteristics such as method accuracy, reproducibility, and throughput are also considered highly advantageous. Matrix effects stemming from cointroduced biological materials are well-known to significantly impact ESI methods, disrupting the vaporization process and, in turn, suppressing ionization

tmospheric pressure photoionization (APPI) has previously been identified as an alternative candidate to electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) for the analysis of clinically relevant analytes within complex biological matrices by LCMS.1−9 However, modern commercially available opengeometry APPI sources often fail to provide significant performance advantages, in terms of sensitivity, for trace level detection of many clinical biomarkers.10−12 As a result, among common LC-MS ionization sources, ESI and APCI remain the methods of choice for the routine determination of steroids, and their metabolites. Recently, we described a new orthogonal geometry, field-free APPI (FF-APPI) source that has demonstrated the potential for order-of-magnitude sensitivity enhancement over commercial open-geometry APPI sources for the analysis of both polar and nonpolar analytes.13 Performance improvements afforded by the new source have been attributed to its geometry; specifically the reintroduction of an enclosed stainless steel flowing reaction chamber−a design element once adopted within the now abandoned first-generation of field-free APPI sources.14 In all modern commercial APPI sources (ex. Photospray and Photomate) an open-geometry configuration is utilized in which primary ionization and all critical secondary © XXXX American Chemical Society

Received: January 13, 2017 Accepted: March 6, 2017 Published: March 6, 2017 A

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 1. MS Parameters Optimized for the Steroid Panel compound

concentration (ppm)

Q1

Q3

CE

CXP

DP

EP

RT (min)

cortisone estradiol estrone testosterone progesterone pregnanediol androsterone

1 1 1 0.1 0.1 1 1

361.1 255.2 271.1 289.1 315.2 285.1 255.2

163.1 159.1 133.0 109.1 97.0 81.0 159.1

33 25 35 40 45 51 25

4 3 3 3 3 3 3

50 35 37 46 47 43 35

6 3 4 6 5 5 3

1.51 1.81 1.84 1.95 2.33 3.05 3.22

efficiency.16 As a consequence, time-consuming sample pretreatment methods are often required including: protein precipitation, liquid−liquid or solid phase extractions, analyte derivatization, and lengthy chromatographic separations. Each of these processes has an impact on sample analysis times and accuracy. Although no atmospheric pressure ionization method can be entirely immune to matrix effects, and the need for sample preparation will never be obviated, the development of ion sources that minimize sample pretreatment will inevitably pay dividends. Presently, APCI may be considered a useful alternative to ESI for the analysis of biological samples, demonstrating improved sensitivity for some moderately polar analytes and a reduced susceptibility to matrix ion suppression.17,18 A number of reports have indicated that APPI methods are often less susceptible to matrix effects than ESI-based methods.19−22 For APPI, analyte ion formation is typically the result of gas-phase ion−molecule interactions. Similarly, ion suppression for APPI may be attributed to competitive gasphase ion−molecule reaction pathways.23 In one potential pathway, coeluting matrix elements may abduct charge directly from the chemical reagent ion supply. Depletion of the available reagent ion population through competitive mechanisms will reduce the probability of analyte ionization. Alternatively, charge may be removed from previously ionized analyte molecules through direct interaction with sufficiently basic or electrophilic matrix components. Considering the general complexity of most biological matrices, the relative contributions of these prospective loss mechanisms are unknown. Whatever the pathway, however, it is understood that APPI sources are not immune to ion suppression or enhancement. As a result, the characterization of matrix effects must also be considered during the APPI method development process. In this report we continue to explore the performance potential of the new orthogonal geometry, FF-APPI source as a candidate for the routine detection of clinical biomarkers using simple, high-throughput LC-MS workflows. The position of FF-APPI relative to other modern LC-MS sources such as ESI and APCI has been considered. Additionally, the impact of matrix effects on FF-APPI performance for the analysis of various complex biological sample types has been evaluated.

■

nebulizer gas pressure (GS1) = 40 psi, nebulizer temperature = 400 °C, and discharge current = 4 μA. The prototype field-free APPI source used for all experiments has been described previously.12 Photoionization lamp power was applied using a high voltage power supply originally designed for first generation MDS Sciex coaxial field-free Photospray sources. Offset transfer voltage was applied using a stand-alone Stanford Research (Sunnyvale, CA) PS350 programmable DC high voltage supply. Lamp gas was provided from a high purity nitrogen supply at 0.3 L/min. The FF-APPI source was operated with a heated nebulizer temperature of 275 °C and nebulizer gas pressure of 30 psi, unless otherwise stated. All analyses were performed in positive MS mode with a scan time of 100 ms and a dwell time of 5 ms. Toluene dopant was infused at 20 μL/min (10% of the mobile phase flow rate) using a syringe pump. Increasing the dopant flow rate had no significant impact on assay performance above this flow rate. No other dopants were evaluated in the present study; however, toluene is generally known to provide the best performance for the analysis of analytes ionized through the proton transfer mechanism. Dopant and analytical solvent flows were introduced independently to the source using a modified heated nebulizer probe. The new probe functioned identically to the commercial APCI probe; however, a separate stainless steel capillary was provided for the introduction of the dopant. This feature enabled both solvent flows to be introduced to the heated nebulizer simultaneously yet independently, eliminating the potential for solvent immiscibility and sputtering, a potential source of signal instability and high background noise levels. Optimized MS parameters are summarized in Table 1. For most analytes, [M + H]+ was the most intense ion observed. For androsterone, however, the [M + H]+ ion was observed only at low nebulizer temperature. At elevated temperatures, a base peak corresponding to [M − H2O + H]+ was identified for APCI and ESI, while the most intense ion for APPI, corresponded to a second loss of water [M − 2•H2O + H]+. This ion, however, was identical to the parent ion selected for estradiol, the product of a single water loss. Fortunately, elution times for these compounds were sufficiently different to allow the discrete determination of both analytes. Estrone also exhibited a substantial water loss product ion (m/z 253); however, the strongest signal was observed for [M + H]+. Separations were performed isocratically using a Gilson (Middleton, WI) LC pump to introduce premixed mobile phase solvents at constant flow rate of 200 μL/min. A 90:10 methanol/water, 0.05% formic acid solvent composition was selected to provide retention times suitable for high throughput clinical screening assays. All analytes were eluted in under 4 min. An additional 2 min were provided for column reequilibration between analyses. 10 μL sample injections were made on-column (Phenomenex, Luna C18, 3 μm, 100 mm × 2

EXPERIMENTAL SECTION

Methods and Materials. All characterization experiments were performed upon an Applied Biosystems/MDS Sciex (Concord, ON, Canada) API 3200 series triple quadrupole mass spectrometer. An unmodified Turbo V source was utilized in both ESI and APCI modes. ESI determinations were made using the following ion source conditions: probe voltage = 5500 V, nebulizer gas pressure (GS1) = 45 psi, source gas pressure (GS2) = 50 psi, and source temperature = 500 °C. APCI determinations utilized the following source conditions: B

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry mm) using a manually activated switching valve integrated within the mass spectrometer. For post-column infusion (PCI) matrix effect evaluation experiments, a second syringe pump was used to infuse analyte standards at 10 μL/min. The PCI flow was combined with the LC effluent using a low dead volume T-union prior to entering the ion source probe. Analyte detection limits (DL) were calculated using the relationship described by eq 1.

DL = 3σblank /s

(1)

Analyte sensitivity, s, was calculated based on the average signal intensity determined for four consecutive 10 μL injections of single component 100 ppb analyte standards prepared in pure mobile phase. Blank standard deviation, σblank, was determined through the injection pure mobile phase. Chemicals. All analytes were purchased as dry solids at no less than 97% purity. Cortisone, estradiol, testosterone, and progesterone were purchased from Sigma-Aldrich (St. Louis, MO). Estrone, pregnanediol, and androsterone samples were kindly provided by Sciex, predissolved in methanol to produce single component solutions, each 1 mg/mL. HPLC grade methanol, toluene, and formic acid were purchased from Fisher Scientific (Fair Lawn, NJ). Deionized water was produced using an in-house Millipore Milli-Q (Billerica, MA) generator. The 2× charcoal stripped human serum and plasma (k2 EDTA) matrices were obtained from Bioreclamations Inc. (Westbury, NY), while Surine, synthetic urine, was obtained from Dynatek Industries (Lenexa, KS). Sample Preparation. Cortisone, estradiol, testosterone, and progesterone solids were individually dissolved in methanol to produce 1-mg/mL stock solutions from which serial dilutions were made. Single component standards for each of the analytes were diluted in mobile phase to 1 μg/mL. These standards were used to optimize MS and ion source conditions. Finally, a consolidated steroid panel was created with final analyte concentrations described in Table 1. Further dilutions were made as required for robust regression experiments. Human serum and plasma matrices were protein precipitated using the following procedure: 600 μL of matrix was placed in a conical vial to which 1200 μL of methanol was added. The vial was capped and vortexed for 30 s. The solution was then chilled for 30 min at 4 °C. Each sample was then centrifuged for 30 min at 9000 rpm. The supernatant was retained and filtered at 0.22 μm. The protein precipitated samples were then spiked with the steroid panel to produce solutions with a range of analyte concentrations. The solvent composition of each sample was adjusted to match the isocratic mobile phase conditions. Standard samples containing only the steroid panel in pure solvent were produced to match the concentration range and solvent composition of the matrix containing samples. Simulated urine samples were filtered at 0.22 μm and then diluted in pure solvents to match mobile phase conditions. A 10:1 dilution factor was achieved with respect to the concentration of the stock Surine matrix. Surine samples were then spiked with the steroid panel to produce a series of samples with a range of analyte concentrations. A second series of pure solvent standards was produced with equivalent analyte concentrations.

Figure 1. Overlaid chromatograms (the sum of all analyte MRMs) obtained for the analysis of a panel of steroids spiked within two different matrices: (i) pure solvents only (solid trace) and (ii) protein precipitation human plasma (dotted trace). Note: the intensity scale for the FF-APPI results is 10-fold larger than for APCI or ESI.

panel of steroids within a pure solvent matrix (solid line) for ESI, APCI, and FF-APPI. Each chromatographic trace represents the sum of all MRMs obtained for each analyte in the panel. The intensity scales for the chromatograms provided for the ESI and APCI sources are equivalent; however, the scale for the FF-APPI source has been expanded by a factor of 10, indicating significant sensitivity gains. Formic acid was found to be critical to the performance of ESI; however, it had negligible impact on performance for both APCI and FF-APPI. The introduction of a dopant was of course a requirement for the ionization of polar compounds by APPI. Toluene was found to offer excellent sensitivity and was introduced for all APPI experiments. The most significant performance advantage was observed for estrogens including estradiol and estrone. These results are particularly noteworthy, as estrogens are often considered to be difficult to ionize by electrospray due to a lack of highly acidic or basic functionalities and a localized steroidal ring structure.24,25 Time-consuming derivatization pretreatment methods are often required in order to enhance sensitivity, complicating the sample preparation process and introducing additional measures of uncertainty. The ability to provide improved performance for the clinical determination of trace estrogenic biomarkers is a significant advantage associated with FF-APPI sources. A similar performance advantage was also recognized for the analysis of androsterone. As identified previously, at high temperature androsterone was observed to

■

RESULTS AND DISCUSSION Detection Limits. The results presented in Figure 1 demonstrate the sensitivity performance for the analysis of a C

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry lose two water molecules (M − 2•H2O) during ionization, resulting in the formation of a product molecule that exhibited MRM transitions identical to estradiol. Sufficient separation through chromatography enabled independent detection of both compounds. Detection limits were calculated for each analyte and are summarized in Table 2. In general, the field-free

headspace of aqueous protein samples using gas chromatography.31 They offered that if signal intensities obtained for two different sample sets agree within a confidence limit of 95%, regression should yield an mRLR that approaches unity. Figure 2 demonstrates RLR results obtained for the analysis of cortisone within a complex human plasma matrix using a

Table 2. Steroid Panel Limits of Detection (3σblank) LOD (pg) compound

APCI

ESI

FF-APPI

cortisone estradiol estrone testosterone progesterone pregnanediol androsterone

4.3 3.8 5.3 1.1 1.0 50 49

2.1 8.1 8.3 0.6 0.9 32 35

1.3 1.0 0.7 0.2 0.2 15 6.2

APPI source was shown to provide the best sensitivity for each of the steroids evaluated. Since a relatively low-performance mass spectrometer was employed, the relevance of the absolute values is limited; however, the results do indicate that relative sensitivity gains translate well to detection limits. Matrix Effects. The relative impact of matrix effects was assessed for each ion source for the analysis of complex biological samples, under high throughput conditions. Referring to Figure 1, overlaid chromatograms obtained with (dashed line) and without (solid line) the addition of proteinprecipitated plasma demonstrate the impact derived from a particularly complex matrix. APCI was the source least impacted by the introduction of matrix, with FF-APPI and ESI sources following, respectively. Although FF-APPI experienced significant ion suppression, particularly for early eluting analytes, the duration of sensitivity loss was relatively short-lived compared to ESI. In contrast, coeluted plasma matrix components generally enhanced signals obtained using the APCI source. Signal enhancement with APCI is not uncommon and may obviously lead to false positive reporting.26,27 Post-extraction addition (PEA) is frequently used to provide a quantitative measure for evaluating matrix effects in LC-MS methods. This method has been thoroughly described throughout the literature.28 In order to build further confidence in the outcome of PEA analysis, a robust linear regression (RLR) method may be applied.29,30 RLR provides a means for comparing two sets of data or analytical methods, where the results are known to have a high degree of variability; the evaluation of matrix effects represents an ideal example. In its application, the error associated with random fluctuations in a data set may be reduced. In a standard calibration curve two sets of data are related: one well-known and carefully controlled (the x axis) and the other measured, containing a high degree of uncertainty (y axis). In RLR, both sets of data are uncertain. Peak areas are plotted for signals obtained for analysis of analyte standards against those obtained for samples that include a biological matrix. This process is repeated for a range of analyte concentrations, while the matrix composition is held constant. Following linear regression, the slope obtained (mRLR) presents a useful measure for appraising matrix effects with increased confidence. In a previous application, Strassnig and Lankmayr utilized RLR to quantify the impact of matrix effects on the measurement of ethanol present within the

Figure 2. Robust linear regression data for cortisone extracted from the analysis of a steroid panel spiked within a human plasma matrix. A simple high-throughput LC-MS workflow was utilized. Results compare the relative performances of ESI, APCI, and FF-APPI. Dotted line represents no significant matrix effects (mRLR = 1). Ion suppression indicated by mRLR < 1, while enhancement shown by mRLR > 1.

high throughput LC-MS workflow. Cortisone was the fastest eluting component of the steroid panel (RT = 1.51 min) and was the most impacted by the presence of coeluting endogenous compounds. Each data point represents the average of three trials, alternating between pure analyte standards and spiked plasma. The analysis was repeated for each ion source. The dashed line (slope, mRLR = 1) represents the theoretical scenario in which there is no matrix effect observed. For the analysis of cortisone, the most severe matrix effect was observed for the ESI and FF-APPI sources, producing slopes of 0.21 and 0.18, respectively. This indicates that, under the selected sample preparation and separation conditions, cortisone signals were suppressed approximately 80% for both ionization methods. Conversely, the result obtained for APCI (mRLR = 1.41) indicates significant signal enhancement. Evidently, none of the ionization methods were free from matrix interferences; however, the results suggest that APCI could provide the most accurate results under the selected high throughput conditions. Each ionization method could of course benefit from the use of internal standards, additional sample cleanup or simply modified chromatographic conditions. Expanding the RLR method to the entire steroid panel, Figure 3 summarizes regression slopes obtained for each component of the steroid panel, across a variety of clinical matrices. The results are presented as a function of retention time and each analyte may be identified with reference to Table 1. The matrices evaluated included (a) diluted synthetic urine, (b) precipitated human serum, and (c) precipitated human plasma. A dashed line once again represents the theoretical D

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 3. RLR slopes obtained for each component of the steroid panel, presented as a function of retention time. Each analyte may be identified with reference to Table 1. Results compare the relative performances of ESI, APCI, and FF-APPI. Matrices evaluated include (a) diluted Surine, (b) precipitated human serum, and (c) precipitated human plasma.

To summarize the initial RLR results, across each of the matrices and analytes surveyed, ESI was far and away least resistant to matrix effects with only 2 of the 21 analytes responding favorably (0.9 < mRLR < 1.1) under the selected high-throughput LC conditions. FF-APPI and APCI responded the best to matrix effects with, respectively, 10 and 11 of the analyte signals being insignificantly impacted. These results are of course limited to a restricted set of chromatographic conditions, as well as the stipulated RLR selection criterion. In reality, the impact of matrix effects is expected to vary not only between analytes, but also among samples (including factors such as patient gender), sample preparation procedures, and MS instrumentation. The results, however, do suggest that similar to modern commercial open-geometry APPI sources, FF-APPI should be considered less susceptible to matrix effects than ESI, and roughly comparable to APCI. Parametric Evaluation of Matrix Effects. In this section, we examine how parametric factors such as volumetric flow rate and vaporizer temperature impact the matrix effects observed for the FF-APPI source. To aid this effort, post-column infusion (PCI) was used to visualize and characterize the impact various operating parameters have upon the extent and duration of signal suppression. Citing electrospray as an example, it has been demonstrated that user optimizable parameters such as nebulizer and drying gas pressures, as well as capillary voltage, have a significant impact on the extent of matrix induced ion suppression.33 Similarly, user optimizable source parameters may significantly affect APPI source performance. With regard to the analysis of steroids, the FF-APPI source tended to optimize for sensitivity toward reduced nebulizer temperatures and similarly toward lower gas flow rates. Since the reaction pathways that are understood to govern FF-APPI performance may be altered by these parameters, it also seems likely that the pathways responsible for gas-phase matrix ion suppression/ enhancement effects may also be impacted. We know of no previous reports linking matrix effects in APPI sources to parametric source conditions. Figure 4 presents normalized XIC chromatograms obtained for the postcolumn infusion of a testosterone standard, following the injection of human plasma at various vaporizer temperatures. At low temperatures (275 °C), the degree of ion suppression was minimized. Signal suppression began at around 1 min post-injection and was sustained for approximately 2 min. At elevated temperature (475 °C), however, the degree and duration of suppression was amplified, extending for nearly 10 min post-injection before finally returning to the baseline

absence of matrix effects. FF-APPI provided the highest resistance to matrix effects for the complete analysis of the steroid panel within the simulated urine sample. Based upon the arbitrary condition 0.9 < mRLR < 1.1, only the estrogens eluting at around 1.9 min were subject to sufficient ionization suppression to warrant further sample pretreatment. APCI again demonstrated significant ionization enhancement in response to the coeluting urine matrix, while ESI responded the least favorably, fluctuating between analyte signal enhancement and suppression. For the analysis of the human serum sample (Figure 3, panel b), ESI signals were suppressed by more than 20% with the exception of androsterone, the latest eluting compound. APCI and the field-free APPI source responded both similarly and favorably. In each case, only cortisone experienced significant ionization enhancement, with all other analytes being largely resistant to any matrix effects. Protein precipitated plasma (Figure 3, panel c) was the most complex of the matrices evaluated. Similar to human serum, APCI responded very well to the introduction of plasma matrix. Signals obtained for FF-APPI and ESI, however, were significantly suppressed indicating that additional sample preparation measures are required. Worthy-of-note, with respect to the observation of signal enhancement, blank injection of charcoal stripped serum and plasma matrix yielded no detectable signals for cortisone. Thus, it may be assumed that the increase in signal can be attributed to matrix-induced enhancement mechanisms and not simply the introduction of additional analyte retained within the precipitated matrix. The root cause of ionization enhancement in APCI and APPI remains poorly understood; however, it is likely the result of gas-phase ionization pathways. Ionization enhancement in APCI has been shown to be more prominent under elevated organic solvent compositions, similar to the conditions used in this study.32 Conversely, ion suppression could result from various gas-phase mechanisms. The post-vaporization formation of solid analyte precipitates with coeluted matrix components could result in the reduced potential for analyte ion formation. Alternatively, suppression could simply arise strictly from the competition for charge between analyte neutrals and suitably basic endogenous matrix elements. Competitive ionization pathways may lead to a significant reduction in the available reagent ion population, in-turn impacting analyte ionization efficiency. In any case, matrix effects likely result from a combination of sources and are expected to be both compound and assay condition dependent. E

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

In order to build confidence in these observations, the effect of nebulizer gas pressure upon the impact of matrix ion suppression was further characterized using the robust regression method. The first panel in Figure 6 demonstrates the effect of nebulizer gas pressure on mRLR obtained for the analysis of cortisone spiked within human plasma using the complete high-throughput LC-MS workflow. The mRLR is shown to increase with decreased nebulizer gas pressure, approaching unity at 80 psi, indicating that a very high resistance to matrix effects may be achieved. The right panel extends the regression analysis to the remainder of the steroid panel, showing mRLR as a function of nebulizer gas pressure for each analyte in the panel. The general trend for each analyte is a consistent improvement in the resistance to ion suppression with increasing nebulizer gas pressure. Under the selected highthroughput isocratic conditions, at a nebulizer gas pressure of 80 psi, it was determined that each of the seven analytes exhibited no significant ion suppression or enhancement (satisfied by the condition: 0.9 < mRLR < 1.1) for the analysis within the human plasma matrix. Figure 7 presents chromatograms obtained for the analysis of pure standards overlaid with traces obtained with spiked plasma matrix included. The results compare both the maximum and the minimum nebulizer gas pressure settings. At low pressure (20 psi), the signals were significantly suppressed, particularly for the first four analytes to elute. Moving to high pressure (80 psi), each of the analyte signal intensities decreased by approximately 50%. This apparent trade-off in sensitivity, however, was balanced by the near-complete elimination of ion suppression. At this point the underlying mechanisms responsible for this effect are uncertain, however, these results highlight an important new potential. For the FF-APPI source matrix effects may potentially be mitigated through parametric optimization, however, possibly at the expense of overall sensitivity. Provided that the sample analyte concentration is appreciably above the detection limit, an FF-APPI-MS method may be tailored to provide fast, accurate determinations that are minimally susceptible to matrix effects. This is particularly important for high throughput clinical screening applications where both sensitivity and accuracy are critically important. Under conditions where the matrix effect is found to be negligible, nebulizer gas pressure may be optimized to increase analyte sensitivity and improve detection limits. For the FF-APPI source, analyte sensitivity tends to improve at lower nebulizer gas pressures. Lower pressure results in an effective increase in the gas-phase neutral analyte concentration and in-turn improved analyte ionization efficiency is expected. We propose the same relationship exists with regards to a reduction in reagent ion population through competitive pathways involving endogenous matrix components. At high nebulizer gas pressure the concentration of matrix components is effectively diluted, while ideally, reagent ions remain in excess supply. As a consequence, a decrease in the consumption of reagent ions through interaction with the matrix is expected. Correspondingly analyte ionization efficiency and thus sensitivity should also decrease at higher flow rates, which is also in agreement with our observations. This presents one plausible mechanism, which could explain the reduced impact of matrix effects at elevated nebulizer gas pressures; however, further investigation will be required in order to confirm this hypothesis.

Figure 4. Normalized and overlaid XIC traces obtained for the postcolumn infusion of a 1 μg/mL testosterone standard. Results visualize the suppression profile produced from the injection of blank protein precipitated human plasma samples at various heated nebulizer temperatures.

level. This observation further lends FF-APPI source optimization toward reduced vaporizer temperatures, at least for the determination of steroids. One possible explanation for this behavior could involve the competition for reagent ion charge between coeluted matrix elements and the analyte. At elevated temperatures, the in-source fragmentation of labile endogenous matrix components could be enhanced. Increased fragmentation could potentially lead to an increase in the concentration of suitably basic interferences that may be capable of scavenging charge from the reagent ion population, in-turn altering charge competition with the analyte. Although this proposed mechanism has yet been supported experimentally, the resulting affect is certainly noteworthy and should be considered during the method development process. The parametric investigation of FF-APPI source conditions was extended to include nebulizer gas pressure. Figure 5

Figure 5. Normalized and overlaid XIC traces obtained for the postcolumn infusion of a 1 μg/mL testosterone standard. Results visualize the suppression profile produced from the injection of blank plasma samples at various nebulizer gas (GS) pressures (psi).

provides PCI results that monitor the XIC for testosterone while varying nebulizer gas pressure from 20 to 80 psi. It was necessary to reoptimize the offset transfer voltage in order to maintain the maximum possible analyte sensitivity. The degree of ion suppression was shown to decrease with increased nebulizer gas pressure. Conversely, analyte sensitivity tends to improve at reduced nebulizer pressures, presumably the result of increased ion−molecule reaction time or the net increase in the gas-phase concentration of analyte neutrals. F

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 6. RLR data obtained (a) for cortisone as a component of the steroid panel and (b) summarizing the complete analysis of the steroid panel spiked within a human plasma matrix. Results obtained at various nebulizer gas, GS, pressures (psi). Dotted line represents theoretical case with no significant matrix effects.

In a separate suite of experiments, the impact that variable source conditions have upon the extent and duration of the matrix effect was characterized. This study produced two noteworthy results: First, matrix effects were found to be minimized at lower heated nebulizer temperatures for the analysis of the steroid panel spiked within a protein precipitated human plasma matrix; Second, ion suppression was found to be enhanced at lower nebulizer gas pressures; however, the impact was reduced dramatically higher flows, though at the expense of sensitivity for all analytes. These two observations suggest that there may be potential to mitigate the impact of some matrix effects in FF-APPI sources, allowing one to tailor an assay to provide optimal results. This potential was demonstrated using the high-throughput method to analyze the spiked plasma sample at high gas flow rate, resulting in the near elimination of ion suppression effects. This study has established that FF-APPI is a potential alternative to ESI and APCI for the analysis of steroids. With increased application and acceptance, FF-APPI could be considered the ionization method of choice for many high-throughput clinical LC-MS determinations.

Figure 7. Overlaid chromatograms obtained for the analysis of a steroid panel spiked within a protein precipitation human plasma matrix at nebulizer gas pressure settings of 20 psi (top) and 80 psi (bottom). Black traces obtained for standards diluted in pure solvent only. Red traces obtained for standards spiked into plasma.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: (705) 794-9418. E-mail: [email protected].

■

ORCID

CONCLUSIONS The relative performance of a new orthogonal geometry fieldfree APPI source was evaluated against commercially available ESI and APCI sources for the analysis of a panel of steroids spiked within various complex biological matrices, using a rapid LC-MS/MS method suitable for high throughput clinical workflows. In terms of sensitivity, the FF-APPI source provided best performance for the determination of all analytes, with order-of-magnitude enhancements for the analysis of estradiol and estrone. Limits of detection at or below the low picogram range were frequently measured; however, detection limits would be much lower using higher performance MS instruments. The impact of matrix effects was also evaluated using a robust linear regression method to increase confidence in the post-extraction addition technique. Under conditions optimized to provide maximum sensitivity, FF-APPI and APCI were determined to be the least affected by matrix effects under the selected high throughput LC conditions, while ESI was found to be extremely susceptible to signal suppression.

Ross D. McCulloch: 0000-0003-3496-0213 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge the Natural Sciences and Engineering Research Council of Canada and the University of British Columbia for financial support.

■

REFERENCES

(1) Keski-Rahkonen, P.; Huhtinen, K.; Desai, R.; Harwood, D. T.; Handelsman, D. J.; Poutanen, M.; Auriola, S. J. Mass Spectrom. 2013, 48 (9), 1050−1058. (2) Karuna, R.; von Eckardstein, A.; Rentsch, K. M. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2009, 877, 261−268. (3) Kushnir, M. M.; Neilson, R.; Roberts, W. L.; Rockwood, A. L. Clin. Biochem. 2004, 37, 357−362. (4) Yamamoto, A.; Kakutani, N.; Yamamoto, K.; Kamiura, T.; Miyakoda, H. Environ. Sci. Technol. 2006, 40, 4132−4137. G

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry (5) Zhang, F.; Bartels, M. J.; Geter, D. R.; Carr, M. S.; McClymount, L. E.; Marino, T. A. Rapid Commun. Mass Spectrom. 2009, 23, 3637− 3646. (6) Borges, N. C.; Astigarraga, R. B.; Sverdloff, C. E.; Galvinas, P. R.; Silva, W. M.; da Rezende, V. M. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2009, 877, 3601−3609. (7) Lih, F. B.; Titus, M. A.; Mohler, J. L.; Tomer, K. B. Anal. Chem. 2010, 82, 6000−6007. (8) Matejicek, D. J. Chromatogr. A 2011, 1218, 2292−2300. (9) Zhang, F.; Rick, D. L.; Kan, L. H.; Perala, A. W.; Geter, D. R.; Lebaron, M. J. Rapid Commun. Mass Spectrom. 2011, 25, 3123−3130. (10) Hommerson, P.; Khan, A. M.; Jong, G. J. de; Somsen, G. W. Electrophoresis 2007, 28, 1444−1453. (11) Lien, G.-W.; Chen, C.-Y.; Wang, G.-S. J. Chromatogr. A 2009, 1216, 956−966. (12) Chen, W.-L.; Wang, G.-S.; Gwo, J.-C.; Chen, C.-Y. Talanta 2012, 89, 237−245. (13) McCulloch, R. D.; Robb, D. B.; Blades, M. Rapid Commun. Mass Spectrom. 2016, 30, 333−339. (14) Robb, D. B.; Covey, T.; Bruins, A. Anal. Chem. 2000, 72 (15), 3653−3659. (15) Syage, J. A.; Evans, M. D.; Hanold, K. A. Am. Lab. 2000, 32 (24), 24−29. (16) Annesley, T. M. Clin. Chem. 2003, 49, 1041−1044. (17) Schlusener, M. P.; Bester, K. Rapid Commun. Mass Spectrom. 2005, 19, 3269−3278. (18) King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah, T. J. Am. Soc. Mass Spectrom. 2000, 11, 942−950. (19) Takino, M.; Daishima, S.; Nakahara, T. Rapid Commun. Mass Spectrom. 2003, 17 (5), 383−90. (20) Takino, M.; Daishima, S.; Nakahara, T. Rapid Commun. Mass Spectrom. 2003, 17 (17), 1965−72. (21) Takino, M.; Tanaka, T.; Yamaguchi, K.; Nakahara, T. Food Addit. Contam. 2004, 21 (1), 76−84. (22) Takino, M.; Yamaguchi, K.; Nakahara, T. J. Agric. Food Chem. 2004, 52 (4), 727−35. (23) Robb, D. B.; Blades, M. W. Anal. Chim. Acta 2008, 627 (1), 34− 49. (24) Tai, S. S.-C.; Welch, M. J. Anal. Chem. 2005, 77, 6359−6363. (25) Kushnir, M. M.; Rockwood, A. L.; Bergquist, J.; Varshavsky, M.; Roberts, W. L.; Yue, B. Am. J. Clin. Pathol. 2008, 129 (4), 530−9. (26) Liang, H. R.; Foltz, R. L.; Meng, M.; Bennett, P. Rapid Commun. Mass Spectrom. 2003, 17, 2815−2821. (27) Zhao, X.; Metcalfe, C. D. Anal. Chem. 2008, 80, 2010−2017. (28) Taylor, P. J. Clin. Biochem. 2005, 38, 328−334. (29) Miller, J. C.; Miller, J. N. Statistics for Analytical Chemistry, 3rd ed.; Ellis Horwood Ltd.: West Sussex, England, 1993. (30) Hartmann, C.; Vankeerberghen, P.; Smeyers-Verbeke, J.; Massart, D. L. Anal. Chim. Acta 1997, 344, 17−28. (31) Strassnig, S.; Lankmayr, E. P. J. Chromatogr. A 1999, 849, 629− 636. (32) Gosetti, F.; Mazzucco, E.; Zampieri, D.; Gennaro, M. C. J. Chromatogr. A 2010, 1217, 3929−3937. (33) Kruve, A.; Herodes, K.; Leito, I. Rapid Commun. Mass Spectrom. 2010, 24 (7), 919−926.

H

DOI: 10.1021/acs.analchem.7b00157 Anal. Chem. XXXX, XXX, XXX−XXX