| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Automation and Analytical Techniques |
Department of Pathology, University of Michigan Health Sciences Center, Ann Arbor, MI.
aAddress correspondence to the author at: University Hospital, Rm. 2G332, 1500 East Medical Center Dr., Ann Arbor, MI 48109-5054. Fax 734-763-4095; e-mail annesley{at}umich.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Drug stability studies, analysis of biological extracts, mixing experiments, and postcolumn infusions were used to test 9 commercial methanols for ionization differences in liquid chromatography-tandem mass spectrometry assays for immunosuppressants. Area responses for the drugs and internal standards were compared for mobile phases prepared with each selected methanol. Postcolumn infusion experiments were performed to confirm the degree of ionization differences occurring at the ion source, and to evaluate the proportions of ammonium, sodium, and potassium adducts.
Results: The decrease in signal for the immunosuppressant drugs was shown to result from differential ionization associated with the selected methanols. Product ion intensity varied by 10-fold among the methanols tested. For sirolimus, tacrolimus, and mycophenolic acid, the percentage change in ionization was the same for the drug and its corresponding internal standard. Postcolumn sirolimus infusion evaluation revealed that a 1000-fold analyte concentration difference did not affect ionization. The proportions of ammonium, sodium, and potassium adducts of sirolimus precursor ions differed in relation to the source of methanol.
Conclusions: Organic solvents used in mobile phases and extract preparation of biological samples may be associated with ion suppression, affecting adduct formation and assay sensitivity.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One use of LC-MS/MS at my institution is in assays for quantification of immunosuppressants in whole blood and serum. In sample preparation and chromatography for these assays, we use methanol because it is readily available and less costly than acetonitrile. The use of methanol as a mobile phase component in MS methods for immunosuppressants has been frequently reported (2)(10)(11)(12)(13)(14). In our laboratory we observed a slow loss of 32-desmethoxyrapamycin, the internal standard for sirolimus, if the 8 µg/L methanolic working solution was stored at ambient temperature. This solution was very stable if stored at –20 °C, however, so we now store the larger stock bottle at this temperature and pour off a daily working volume. We presumed that the loss resulted from degradation of 32-desmethoxyrapamycin in the Fisher Optima methanol we were using, an effect similar to that reported for the internal standard ascomycin when dissolved in some brands or grades of acetonitrile (15). I investigated whether several different sources and grades of methanol would correct the slow loss of 32-desmethoxyrapamycin at ambient temperature, and also be suitable for use in the mobile phase. When tested as the solvent for preparation of the working internal standard solution, no improvement in the slow loss of the 32-desmethoxyrapamycin was found. However, when used as the organic component of the mobile phase, I noted large differences in the ionization of the immunosuppressants and their internal standards when different sources and grades of methanol were evaluated.
Coeluting components originating from the sample matrix have previously been shown to negatively affect (ion suppression) or positively affect (ion enhancement) the analyte signal in LC-MS analyses. This report describes the phenomenon of ionization changes related to the organic solvent used in the LC-MS/MS analysis.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
commercial methanol sources
Fisher Brand Optima, Fisher Brand ACS, Fisher Brand HPLC, Fisher Brand Optima LC/MS, EMD Omnisolv, Baker Ultra Resi-analyzed, and Baker Analyzed LC/MS methanol were purchased from Fisher Scientific. Riedel-de Haen Chromasolv LC-MS methanol was purchased from Sigma-Aldrich, and Super Methanol was purchased from American Bioanalytical. Two separate lots were also evaluated for 7 of the 9 sources of methanol (see Table 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol53/issue10). My laboratory routinely used Fisher Optima methanol at the time of the evaluation of the different methanol types. Therefore, I chose to use the lot number of Optima methanol in use at that time as the reference solvent to which others were compared, including a subsequent new lot of Optima methanol.
lc-ms/ms analyses
I quantified sirolimus and tacrolimus in whole blood as previously described (16) using commercial calibrators and controls as part of the analyses (17). Sirolimus was analyzed after protein precipitation and solid-phase extraction, and tacrolimus after protein precipitation alone. I quantified mycophenolic acid and mycophenolic acid glucuronide as previously described (18) using calibrators and controls prepared in-house. Each of the immunosuppressant assays referenced above employs gradient chromatography using 2 common mobile phases: 2 mmol/L ammonium acetate plus 1 mL/L formic acid in water (mobile phase A) and 2 mmol/L ammonium acetate plus 1 mL/L formic acid in methanol (mobile phase B).
Chromatography and MS detection were performed using an Agilent 1100 solvent delivery system and a Waters Quattro Micro tandem mass spectrometer equipped with a Z-spray ion source operated in the positive electrospray ionization (ESI) mode. The instrument control software was Waters MassLynx 4.0.
comparison of peak area responses
I prepared batches of mobile phase B in 125 mL Nalgene amber high-density polyethylene bottles to evaluate the effect of each commercial methanol on the integrated peak areas for the immunosuppressant drugs and internal standards. The Nalgene bottles were rinsed with distilled-deionized water, dried, and then rinsed twice with the selected methanol before preparation of mobile phase B. Graduated cylinders were also rinsed with the selected methanol before mobile phase B preparation. The same formic acid and ammonium acetate stock solutions were used to prepare each batch of mobile phase B, the only variable being the brand or grade of methanol. Each mobile phase B solution was prepared on the same day of use.
Extracts prepared from calibrators, controls, or deidentified patient specimens (institutional review board approved) were injected into the LC-MS/MS system, and the integrated peak areas for the drug and internal standard were calculated using Waters MassLynx software. The concentration of drug was also quantified by use of peak area ratios. For each series of extracts, the Agilent 1100 system and analytical column were re-equilibrated with new mobile phase B prepared with a different commercial methanol, and then the same series of extracts (same autosampler vials) was reanalyzed with the new mobile phase B preparation. To verify that differences in response for the drugs or internal standards were not due to instrument drift or analyte degradation during the day, on 2 occasions I used mobile phase B preparations from methanol brands in 1 sequence to analyze extracts, and then reversed the order and repeated the analyses with the same mobile phase preparations. Each brand or grade of methanol was also tested again on a separate day, with a new preparation of mobile phase B.
drug stability in commercial methanols
To determine if analyte degradation in commercial methanol contributed to the observed changes in area response, I prepared 8 µg/L working solutions of the internal standards 32-desmethoxyrapamycin and ascomycin in 1 lot of each methanol. These solutions were then used for the precipitation and solid-phase extraction of calibrators, controls, or patient specimens containing sirolimus and tacrolimus (16)(17). Fisher Optima-based mobile phase B was used for the chromatography (16), so the methanol used for the sample preparation was the only variable. Changes in the area response for the drugs or internal standards were compared.
postcolumn infusion
I performed postcolumn infusion studies to evaluate (a) whether the changes in peak area related to the choice of methanol were attributable to ionization differences in the ion source rather than drug degradation during chromatography; (b) whether the concentration of the drug being infused affected the ionization efficiency; and (c) if changes occurred in the relative amounts of the ammonium, sodium, and potassium adduct ions formed during the ionization process.
In the 1st experiment, I infused a 10 µg/L methanolic sirolimus solution, using a tee, into the mass analyzer. The syringe infusion rate was 10 µL/min, and the LC flow rate was 400 µL/min (20% mobile phase A, 80% mobile phase B). The signal for the m/z 931.6
864.5 transition was integrated for 30 s. The variable in this study was methanol used to prepare mobile phase B.
Heller (19) recently showed that the analyte:matrix mass ratio can affect the analyte response and the analyte:internal standard response ratio, and proposed a matrix effect map concept to test for matrix effects in atmospheric pressure ionization MS. To see whether varying the analyte:methanol (i.e., matrix) ratio would highlight the presence of ion loss resulting from the methanolic mobile phase, I performed a 2nd experiment using postcolumn infusions as described above, but with 4 different concentrations of sirolimus (10 mg/L, 1 mg/L, 100 µg/L, and 10 µg/L). In this experiment the variable was the concentration of sirolimus tested against mobile phase B prepared with a selected brand or grade of methanol. The m/z 931.6864.5 transition for sirolimus was monitored.
In the 3rd experiment, I infused a 10 mg/L methanolic sirolimus solution into the mass analyzer. The infusions were performed with 80% mobile phase B and a 400 µL/min flow rate. During each infusion a precursor ion scan from m/z 900 to m/z 990 was performed (30-s integration). The variables studied here were the methanol used to prepare mobile phase B and the precursor adduct ions formed during the ionization process.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because the same concentration of internal standard was added to the sample before extraction and analysis, I compared the mean area responses obtained with LC-MS/MS for each commercial methanol used as the solvent in mobile phase B. The percentage area response of the multiple-reaction monitoring transition m/z 901.5834.4 for 32-desmethoxyrapamycin, the internal standard for sirolimus, as a function of the methanol tested is shown in Fig. 1
. Two things are apparent from the data. First, 10-fold differences in the m/z 834.4 product ion occurred with the various brands of methanol. Second, ion production differed between 2 lots of the same methanol type. The results from a separate multiple-reaction monitoring experiment are shown in Fig. 2
, in which the m/z 901.5834.4 and m/z 809.4756.3 transitions for 32-desmethoxyrapamycin and ascomycin were evaluated for the same biological extracts. The changes in ionization for 32-desmethoxyrapamycin (Fig. 2A
) are of the same magnitude as those observed for ascomycin (Fig. 2B
).
|
|
Although there were differences in ionization efficiency among the methanols, the ionization of analyte in relation to internal standard changed similarly as a function of the quality of the methanol in the mobile phase, so that the area ratios and calibration lines used to calculate drug concentrations in patient specimens were similar (Fig. 3
, Supplemental Data Fig. 1). The 2 exceptions were sirolimus and mycophenolic acid glucuronide when analyzed with the Baker Analyzed LC/MS methanol. The raw product ion signals for both sirolimus and desmethoxyrapamycin were both markedly decreased, but the signal reduction was approximately 84% for sirolimus and approximately 89% for desmethoxyrapamycin. Thus, the area ratios for the sirolimus calibration line were higher. Mycophenolic acid glucuronide eluted at a lower methanol percentage and shorter retention time than the internal standard (4.3 vs 5.8 min) during the gradient chromatography, and the product ion signal was less affected than the internal standard. With the Baker Analyzed LC/MS, the product ion signal for the internal standard was also markedly decreased, a result that was reflected by a change in the area ratios and the calibration line.
|
When sirolimus was infused at concentrations spanning a 1000-fold range, the m/z 931.6864.5 transition yielded a proportional signal response and no apparent change associated with the analyte:matrix ratio (Fig. 4
). The 4 methanol brands used in the comparison were chosen because they represented the reference methanol (Fisher Optima), a methanol brand with a modest increase in signal response relative to the reference methanol (EMD Omnisolv), a methanol that showed a moderate signal decrease (Fisher ACS Grade), and a methanol that showed a larger signal decrease (Baker Analyzed LC-MS).
|
The stability of sirolimus, 32-desmethoxyrapamycin, tacrolimus, and ascomycin during the extraction of whole blood specimens with the 9 methanols was compared. The relative peak area response for the tacrolimus m/z 821.4768.3 transition when these methanols were used during the extraction process (Fig. 5A
) illustrates that degradation of tacrolimus was not a problem during contact with the methanol. However, changes in peak area response were observed for these same blood extracts when analyzed using the same methanol in mobile phase B of the LC-MS/MS system (Fig. 5B
), indicating that the changes in signal occurred during the ESI process. Similar data were obtained for the other drugs and internal standards.
|
To confirm that the differences in ionization were due to methanol-associated ion decreases rather than methanol-associated ion enhancement, I performed a mixing experiment in which I prepared a mobile phase B containing 50% Fisher Optima and 50% Baker Analyzed LC/MS methanol. Using 32-desmethoxyrapamycin as the test compound, the decrease in ionization for this compound was 87% relative to a reference mobile phase B prepared with 100% Optima methanol. This decrease in ionization efficiency was similar to the value shown in Fig. 1
for the Baker Analyzed LC/MS mobile phase B. This result is consistent with an ion reduction phenomenon rather than an ion enhancement effect.
The degree of change in ion formation appeared to correlate with the percentage of methanol required to elute the compound of interest. For example, when Baker Analyzed LC/MS methanol was used to prepare mobile phase B, a 32% decrease in ion signal was observed for mycophenolic acid glucuronide, which eluted 1st from the reversed-phase analytical column during the gradient chromatography at approximately 40% mobile phase B (18). A 67% ion signal reduction was observed for mycophenolic acid, which eluted next at approximately 50% mobile phase B, and a 77% ion signal reduction for mycophenolic acid carboxybutoxy ether, which eluted at approximately 55% mobile phase B. By comparison, the ion reduction for desmethoxyrapamycin, which was chromatographed using an even higher percentage of mobile phase B (17), averaged 87% with this commercial methanol (Fig. 1
).
When 1 lot of Riedel-de Haen Chromasolv LC-MS methanol, which had shown a mean 16% signal reduction for 32-desmethoxyrapamycin (Fig. 1
), was used, the percentage of reductions were 8% for mycophenolic acid glucuronide, 9% for mycophenolic acid, and 10% for mycophenolic acid carboxybutoxy ether. Although these small changes were not statistically significant, they follow the same trend noted above.
The sirolimus adduct ion profiles for selected lots of 5 methanols used to prepare mobile phase B are shown in Table 1
. As shown by the data, the relative ratios of the ammonium, sodium, potassium, and an as yet unidentified m/z 972.7 ion of sirolimus differed for the 5 methanols tested. At a mobile phase composition of 80% methanol, the sodium adducts were 130% and 114% that of the ammonium adducts for the Baker Analyzed LC/MS and Riedel-de Haen Chromasolv LC-MS methanols, 48% and 33% for the Fisher Optima and American Bioanalytical methanols, and 14% for the EMD Omnisolv methanol.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
99.8% pure, I infer that the ionization changes are caused by trace contaminants in the solvent.
Using acetonitrile as the electrospray LC-MS solvent, Hewavitharana and Shaw (20) demonstrated that the relative intensities of the molecular ion, ammonium adduct, and sodium adduct for a model test compound, BAY 11-7082, changed with the acetonitrile composition in the mobile phase. In the immunosuppressant assays, ammonium acetate is added to the mobile phase (2)(11)(12)(13)(14)(21)(22) to promote formation of the ammonium adduct, so that the molecular ion should not be observed. My data show that the relative intensities of ammonium, sodium, and potassium adduct ions differed with the methanol type used in the mobile phase. The shift toward higher percentages of sodium adducts with some methanol types (Table 1
) could contribute to the lower ion response for the ammonium precursor ion that is used in the immunosuppressant assays. The higher percentages of sodium adducts may result from sodium contamination or the presence of some compound that facilitates formation of sodium adducts. However, a shift toward sodium adduct ion formation cannot fully explain the losses in the intensities of the precursor and product ions. The American Bioanalytical methanol showed no increase in the percentage of m/z 936.6 sodium adduct ion for sirolimus compared to the Optima methanol (Table 1
), yet ion intensities decreased for the m/z 834.4 product ion of 32-desmethoxyrapamycin (Fig. 1
), the m/z 756.3 product ion for ascomycin (Fig. 2
), the m/z 864.5 product ion for sirolimus, and the m/z 768.3 product ion for tacrolimus (Fig. 5
). In contrast, EMD Omnisolv methanol showed the lowest percentage of nonammonium adducts, but did have higher intensities for these same product ions. This result highlights that the ionization of compounds in ESI MS is affected by many unidentified subtle factors that require further study.
In the case of sirolimus and tacrolimus, the structural analog internal standards eluted at the same time, or very close to the specific drug of interest (i.e., same percentage of methanol). Thus, the percentage loss of ions was similar for both compounds and therefore minimally affected the peak area ratios. This similar change in ionization emphasizes the importance of using an internal standard that behaves as closely to the compound of interest as possible (23).
Ion suppression is an important phenomenon in LC-MS analyses (24)(25)(26)(27). It is associated with endogenous compounds in biological matrices, as well as ion-pairing agents used to improve liquid chromatographic separation. Various protocols have been developed for evaluating the presence of ion suppression in LC-MS assays. These include postcolumn infusion (28), postextraction supplementation studies (25), standard line slope comparisons (26), and analyte:matrix ratios (19). In the case of ion suppression associated with the solvent used in the LC-MS, these protocols may not detect ion suppression because they primarily detect ion suppression from the biological matrices. However, there are analytical parameters that can be monitored to evaluate whether solvent-related ionization effects are occurring in LC-MS/MS. Useful information may be gained by comparing assay performance with several commercial sources of solvents to determine if one solvent provides improved assay sensitivity. A comparison of in-house assay characteristics with those of published LC-MS/MS assays for the same compound may help identify solvent-related ionization effects on the limit of quantification or imprecision of the assay.
I did not investigate what impurities are responsible for ion suppression and alteration of the types of ions present in the spectrum, nor the types of additional purification steps that might alleviate the matrix effects associated with the solvents used in chromatographic mobile phases. These variables might be studied by full-scan mass spectra and perhaps other analytical methods to characterize the impurity content in commercial brands or grades of methanol, and the development of subsequent solvent purification schemes to remove these compounds. This type of information may further our understanding of the fundamental chemistry and physics of the ESI process.
Because the methanol-associated ion reduction was greatest with the high-percentage organic solvent used to analyze the immunosuppressant drugs, the effect may not be as important or evident for LC-MS/MS assays that require a lower methanol percentage for elution. Because of the specificity of MS, however, many assays use a simple sample cleanup and a ballistic gradient. Furthermore, newer columns such as phenyl-hexyl (29) and pentafluorophenyl-propyl (30) are being used because they require a higher percentage of organic solvent to elute compounds, which purportedly provides better desolvation and increased response for compounds. Hydrophilic interaction chromatography columns (31)(32) are being applied for the analysis of polar compounds. This type of column requires a high percentage of organic solvent in the initial mobile phase to retain compounds of interest, which again should be associated with better desolvation and increased response. If the mobile phase organic solvent itself is associated with ion losses, however, then the actual analytical response in the ESI MS may be compromised.
Although my experiments demonstrate that the methanol used in the chromatographic solvent can be associated with variations in ionization efficiency, it should be noted that this phenomenon was evaluated with 1 type of ion source and interface, with mobile phases containing 2 mmol/L ammonium acetate plus 1 mL/L formic acid, and with LC-MS/MS assays that require high organic solvent concentrations for the chromatographic portion of the analyses. It is possible that other source designs, or different source operating conditions, may be less susceptible to methanol-associated ionization effects. Other variables such as decreased flow rate (ultra-performance LC, capillary or nano-ESI) or postcolumn flow splitting may also diminish this effect. Thus, the degree of ionization efficiency that I observed may not be relevant or as significant with other instruments or analytical conditions. Nonetheless, my experience shows that an assumption that any high-purity methanol will behave similarly may not be valid, and that important information may be gained by comparing assay performance in LC-MS/MS with several commercial sources of solvents to determine if one works better for the assay in question.
|
Acknowledgments |
|---|
Financial disclosures: None declared.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  N. Badawi, K. W. Simonsen, A. Steentoft, I. M. Bernhoft, and K. Linnet Simultaneous Screening and Quantification of 29 Drugs of Abuse in Oral Fluid by Solid-Phase Extraction and Ultraperformance LC-MS/MS Clin. Chem., November 1, 2009; 55(11): 2004 - 2018. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Boyd, N. Rifai, and T. M. Annesley Preparation of Manuscripts for Publication: Improving Your Chances for Success Clin. Chem., July 1, 2009; 55(7): 1259 - 1264. [Full Text] [PDF]
|
|
|
|
|
|
K. L. Napoli More on Methanol-Associated Matrix Effects in Electrospray Ionization Mass Spectrometry Clin. Chem., June 1, 2009; 55(6): 1250 - 1252. [Full Text] [PDF]
|
|
|
|
|
|
V. R. De Jesus, H. Zhou, R. F. Vogt, and W. H. Hannon Changes in Solvent Composition in Tandem Mass Spectrometry Multiplex Assay for Lysosomal Storage Disorders Do Not Affect Assay Results Clin. Chem., March 1, 2009; 55(3): 596 - 598. [Full Text] [PDF]
|
|
|
|
|
|
S. O'Halloran and K. F. Ilett Evaluation of a Deuterium-Labeled Internal Standard for the Measurement of Sirolimus by High-Throughput HPLC Electrospray Ionization Tandem Mass Spectrometry Clin. Chem., August 1, 2008; 54(8): 1386 - 1389. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, Fisher Optima; •, Fisher Optima LC/MS grade; 
, Baker Analyzed LC/MS grade; 
, Riedel-de Haen Chromasolv LC-MS; 
, Fisher HPLC grade; 
, EMD Omnisolv; 
, Fisher ACS grade; 
, Baker Ultra Resi-analyzed; 
, American Bioanalytical Super Methanol.
