Clinical Chemistry 45, No. 8, 1999
Gas Chromatography–Mass Spectrometry Analysis of Organic Acids: Altered Quantitative Response for Aqueous Calibrators and Dilute Urine Specimens, Alain Kumps,* Pierre Duez, Janine Genin, and Yves Mardens (Laboratoire de Biochimie Me´dicale, Institut de Pharmacie, Universite´ Libre de Bruxelles (ULB), Campus Plaine 205/3, B-1050 Brussels, Belgium; * author for correspondence: fax 322 6505324, e-mail [email protected]
) Urinary organic acid analysis is a key analytical procedure for laboratories involved in the diagnosis of inherited metabolic disorders (1 ). Gas chromatography–mass spectrometry (GC-MS) is by far the most widely used method, and it is generally performed after oxime-trimethylsilyl derivatization of an ethyl acetate extract. A previous report briefly pointed out an erratic analytical phenomenon we observed with this method for some acids extracted from physiologically diluted urine specimens (with a low creatinine content) or from aqueous solutions (2 ). This so-called “dilution effect” has also been observed in other laboratories (e.g., by N. Chamoles, Fundation para el Estudio de las Enfermedades Neurometabolicas, Buenos Aires, Argentina). The analytical method we used for profiling urinary organic acids has been described in detail (2 ), including the sources of chemicals, the preparation of stock solutions, the ethyl acetate extraction, the oxime-trimethylsilyl derivatization, and the GC-MS operating conditions. Briefly, the extraction and derivatization procedures are as follows: Urine or aqueous solutions (1.5 mL) with internal standards tropate and 2-ketocaproate added are treated with hydroxylamine hydrochloride (30 min at 60 °C), acidified to pH 1 with HCl, saturated with NaCl, and extracted three times with 6 mL of ethyl acetate. The three successive organic phases, each supplemented with 100 mL of a 1 mol/L ammonia solution in ethanol, are evaporated to dryness in the same tube, under nitrogen at 50 °C. The residue is taken up in ethanol, re-evaporated to dryness, and silanized with a mixture N,O-bis(trimethylsilyl)trifluoroacetamide–trimethylchlorosilane–pyridine (99:1:20, by volume) in a stoppered tube for 30 min, 60 °C. GC-MS is operated as described under electron-impact fragmentation and with the scan mode in data acquisition. Quantitative results obtained for control urines and aqueous solutions allowed us to compare the recoveries of acids extracted from water instead of urine. Control urines were prepared by combining aliquots of different stock aqueous solutions of selected organic acid and then diluting them with a previously analyzed urine from a healthy individual; aqueous solutions were similarly prepared, substituting urine for water, to contain the same added organic acid concentrations. All solutions contained low pathological concentrations of the individual acids; the concentrations were corrected for the endogenous concentrations of the organic acids present in the urines used as diluent. The relative recoveries were calculated on the basis of the monitored ion abundance expressed as a proportion of the internal standard (tropate, m/z 280). The responses of some compounds that
have been proposed as internal standards were investigated in this study. The internal standard tropate was considered as a general reference because its absolute m/z 280 abundance remains remarkably stable for all investigated occurrences (mean abundance 6 relative SD in 10 urine samples, 9 477 700 6 14%; in 10 aqueous solutions, 9 983 500 6 13%). When aqueous solutions were analyzed rather than urine samples, recoveries significantly lower than 100% (Student t-test, P ,0.05) were observed for lactate, 3-hydroxybutyrate, p-hydroxyphenylpyruvate, 2-keto-3-methylvalerate, 2-ketoisovalerate, and the candidate internal standards 2-phenylbutyrate, 2-ketocaproate, and 2-hydroxyvalerate (Table 1, columns 2–5). Preliminary testing for glycolate and 2-ketoisocaproate also indicated decreased recoveries. It is, however, quite puzzling to note that this dilution effect is, as reported previously (2 ), quite erratic from week to week, ranging from dramatically high to quasi absent. As an example, observed relative recoveries for lactate ranged from 0.34 to 1.04. This variability cannot yet be related to any experimental factor (see below). In a second series of experiments, aliquots of urine from a healthy individual (8.2 mmol creatinine/L) were diluted with water (down to 3.1 mmol creatinine/L) to represent six urines of different physiological dilutions or creatinine concentrations. These six solutions were supplemented with fixed amounts of 10 acids and with some internal standard candidates. The six supplemented urines were then treated according to our standard procedure. After correction for the endogenous concentrations, these samples demonstrated a decrease in the measured concentration when the water content increased for lactate, glycolate, 3-hydroxybutyrate, pyruvate, p-hydroxyphenylpyruvate, 2-keto-3-methylvalerate, 2-ketoisovalerate, and the internal standards 2-ketocaproate and heptanoylglycine (Fig. 1 and Table 1, column 6). A similar preliminary experiment with the candidate internal standards 2-phenylbutyrate and 2-hydroxyvalerate also demonstrated the same trend (3 ). The previous experiment was then repeated with the same six water-urine mixtures, using the same procedure but without addition of NH4OH to the ethyl acetate extracts before evaporation. In this case, the dilution effect was also observed to a similar extent for the “dilution-sensitive” organic acids, except for 3-hydroxybutyrate and the internal standards 2-phenylbutyrate and 2-hydroxyvalerate (Table 1, column 7). The same phenomenon, i.e., a significantly decreased recovery for the diluted specimens, was also observed for a constant amount of 2-keto-3-methylvalerate added to nine urine specimens obtained from healthy children and containing different physiological creatinine concentrations (1.15–15.4 mmol/L; data not shown). Observations of patient urine specimens obtained for routine diagnosis purposes confirm these findings. The specimens for which creatinine was ,2 mmol/L (concentration chosen arbitrarily) yield a small but statistically significant (P ,0.001) decreased response for the internal standards 2-ketocaproate and 2-hydroxyvalerate vs that for tropate. For example, the abundance ratio of m/z 247
Table 1. Recovery data for internal standard candidates and organic acids.a Relative recovery as a function of urine creatinined,e
Relative recovery, water vs urine
Internal standards 2-Phenylbutyrate 2-Ketocaproate Undecanedioate 2-Hydroxyvalerate 3,3-Dimethylglutarate Heptanoylglycine Volatile/polar acids Lactate Glycolate 3-Hydroxybutyrate Keto acids Pyruvate 2-Ketoglutarate p-Hydroxyphenylpyruvate 2-Keto-3-methylvalerate 2-Ketoisovalerate 2-Ketoisocaproate Glyoxylate Other Glutarate
0.62 0.78 1.01 0.38 0.97
0.24 0.15 0.02 0.28 0.02
8 11 3 4 4
0.28 0.065 .5 2.14 .5
0.67 0.30 and 0.52 0.62
0.73 0.89 0.58 0.39 0.47 0.40 and 0.61 0.90
0.34 0.24 0.17 0.22 0.25
0.65 0.45 0.41 NS
6 2 6
0.51 0.59 0.66
.5 .5 1.58 0.33 2.84
0.91 NS 0.51 0.41 0.40
0.96 NS 0.60 0.48 0.43
6 6 4 5 4 2 6
Results estimated by comparing concentrations of compounds added to water or diluted urine to those in undiluted urine. b Tropate was the reference compound. c Number of recovery experiments. d Six supplemented urines (urine-water mixtures with constant amounts of added organic acids; see text) were analyzed in duplicate. e Values are the ratios of concentrations measured in 0.5 mmol creatinine/L samples to concentrations measured in 5 mmol creatinine/L samples. This ratio is given only if the relationship between recovery of acid and creatinine concentration is statistically significant (P ,0.05, second polynomial regression). f Preliminary results of an experiment on four samples. g NS, not statistically significant.
for 2-ketocaproate to m/z 280 for tropate is 0.41 6 0.028 (mean 6 SD; n 5 26) when compared with more concentrated specimens (abundance ratio, 0.45 6 0.038; n 5 33). This difference is no longer significant when a correction of the urine extracted volume is undertaken as proposed below. Other acids were also included in preliminary experiments: pyroglutamate, glycerate, p-hydroxyphenyllactate, sebacate, citrate, adipate, succinate, glutarate, and oxalate. These demonstrated no dilution effect. As a result, we studied only the acids that were dilution-sensitive; glutarate, however, was retained as a “nonaffected” compound to control for the analytical procedure. To understand the mechanism and origin of the decreased response of keto and polar acids analyzed in diluted urines, we attempted to correlate the occurrence of this phenomenon with other observations. In several experiments, we were unable to establish any reproducible link between this dilution effect and the following variables: urine pH, proteinuria, water quality for reagent preparation, pH of oximation, pH of extracted urine, addition of urea or salts to aqueous organic acids solution, aging of the reagents and solutions, and injector liner
renewal. Only two variables affected the analytical results, i.e., the addition of ammonia to evaporating ethyl acetate extracts and the temperature of this evaporation process. Comparison of calibration urine extracts evaporated at different temperatures demonstrated decreased concentrations for lactate (257%), glycolate (236%), and 3-hydroxybutyrate (240%) when the temperature was 60 °C instead of 50 °C; the concentrations of these compounds increased when the extracts were evaporated at 40 °C (37%, 42%, and 13%, respectively). Avoiding NH4OH during the evaporation step unpredictably reduced the magnitude of the dilution effect in aqueous solutions or in diluted urine specimens for most of the dilution-sensitive acids (Table 1, columns 6 and 7) and consistently suppressed it for 3-hydroxybutyrate. The dilution effect was still present in the absence of NH4OH, except for 3-hydroxybutyrate, 2-hydroxyvalerate, and 2-phenylbutyrate. Thus, an NH4OH-linked origin is only part of the explanation for the dilution effect. It is well known that some degradation of carboxylic acids occurs when aldehydic, oxo, or hydroxy acids are placed in a strong alkaline medium, especially if heated, because they can undergo the Cannizzaro reaction, an
Clinical Chemistry 45, No. 8, 1999
Fig. 1. Alteration of the quantitative response for different dilutions of a urine specimen supplemented with the same concentrations of organic acids. (A), measured organic acid concentrations as a function of the resulting creatinine content. Recovered concentrations were calculated after subtraction of the endogenous concentration. Only significant (P ,0.05, second polynomial regression) relationships are shown. 2-Ketocaproate (as the abundance ratio of m/z 274 to m/z 280) has been included for comparison. Each point is the mean of two determinations. (B), total-ion chromatograms of two supplemented urine dilutions: dilution to 4.1 mmol creatinine/L urine (top) and dilution to 0.31 mmol creatinine/L urine (bottom). Note the stability of tropate (1) and, in the bottom chromatogram, the decrease in size of the peaks for lactate (2), glycolate (3), 3-hydroxybutyrate (4), 2-ketoisovalerate (5), 2-keto-3-methylvalerate (6), p-hydroxyphenylpyruvate (7), 2-ketocaproate (8), and heptanoylglycine (9). Other main peaks are glyoxylate (10), pyruvate (11), urea (12), glutarate (13), 2-ketoglutarate (14; which is lowered not from a dilution effect but by the decreased endogenous concentration), aconitate (15), hippurate (16), and citrate (17).
aldol condensation, or a dehydration, respectively (4, 5 ). The lower concentrations of the hydroxy acids (3-hydroxybutyrate and 2-hydroxyvalerate) are probably attributable to degradation caused by the alkaline medium. Aldehydic and ketonic acids are converted to their oxime derivatives to stabilize the oxo group before extraction and silylation. Thus, in view of the constant physical and chemical conditions in the procedure, the reasons for the dilution-related (amount of water added) degradation of the compounds in urine are unknown. The chemical explanation for the degradation observed by us and other laboratories thus remains unknown. One consequence of our observations is the finding that an aqueous calibration solution is inadequate to standardize the quantification of acids showing a dilution effect unless correction factors based on recovery measurements (Table 1, column 2) are introduced. These relative recoveries may, however, be quite irreproducible from day to day (depending on a still unidentified variable factor). In our opinion, a better calibrator is a supplemented urine sample stored at 220 °C or lower in ready-to-use aliquots. Another consequence is that an important quantitative error may occur in the analysis of some acids (keto or polar) extracted from urine specimens with low creatinine. To remedy this error, we propose the following: (a) The use of 2-ketocaproate as an internal standard for dilution-sensitive acids. This compound is an indicator for the dilution effect and will partly compensate its magnitude. In fact, a previous study demonstrated its usefulness in improving precision of keto acids analysis (2 ). (b) Adjusting, as far as possible, the volume of treated urine so that extract aliquots contain the same amount of creatinine, thus partially controlling the dilution effect. Obviously, very diluted urine (i.e., less than ;1 mmol creatinine/L) cannot be compensated because an impracticably large volume would need to be treated. As a compromise, we now routinely extract 500-mL (with 1000 mL water), 1500-mL, or 2500-mL urine specimens for samples containing .6, ,6, or ,2 mmol creatinine/L, respectively. The volumes of the internal standards solutions, reagents, and ethyl acetate remain unchanged. In conclusion, the extent of the error resulting from the dilution effect may be such that the quantitative response is significantly altered, possibly changing the clinical interpretation when successive samples from a given patient are monitored. However, it is unlikely that this dilution effect could affect the diagnostic interpretation of urinary organic acid profiles.
We thank W. Greig-Neil for revision of the manuscript translation. References 1. Sweetman L. Organic acid analysis. In: Hommes FA, ed. Techniques in diagnostic human biochemical genetics. A laboratory manual. New York: Wiley-Liss, 1991:143–76. 2. Duez P, Kumps A, Mardens Y. GC-MS profiling of urinary organic acids evaluated as a quantitative method. Clin Chem 1996;42:1609 –15. 3. Mardens Y, Kumps A, Duez P. Quantification of urinary organic acids
analyzed by gas chromatography mass spectrometry (GCMS). Sixth International Congress on Inborn Errors of Metabolism, Milan, Italy, May 27–31, 1994. 4. Finar IL, ed. Organic chemistry 1: the fundamental principles. London: Longmans, 1967:966 pp. 5. Chalmers RA, Lawson AM, Watts RWE. Extraction of urinary organic acids by combined barium salt precipitation-anion exchange [Reply to Letter]. Clin Chem 1977;23:903– 8.
Extra Leader Sequence Affects Immunoactivity of Cardiac Troponin I, Shigui Liu,* Min Yuan Zhang, Qianli Song, Xiaochen Zhang, Lilly Kadijevic, and Qinwei Shi (Spectral Diagnostics, Inc., 135-2 The West Mall, Toronto ON M9C 1C2, Canada; * author for correspondence: fax 416-6263651, e-mail [email protected]
) Cardiac troponins are considered the preferred cardiac markers for diagnosis of acute myocardial infarction (1 ). The development and evaluation of cardiac troponin I immunoassays require purified and immunochemically stable antigen for the production of calibrators and controls. Cardiac troponin I (cTnI) purified from human heart has traditionally been used for such purposes. However, cTnI is unstable and prone to degradation during the process of purification from human heart tissue. Recombinant protein is favored over the protein purified from tissue because it is easier to purify, safer to handle, and has unlimited availability and good reproducibility. In physiological conditions, troponin C (TnC) and TnI along with troponin T (TnT) form an integral protein, designated as troponin complex. To study the immunoactivity of TnI in a clinically relevant form, it is imperative to put TnI in the context of the other two subunits, TnC and TnT. In this study, human cTnI, cTnC, and cTnT cDNAs were amplified from the Human Heart Quick-Clone cDNA (Clontech), using gene-specific primers designed from published cDNA sequences (2–5 ) by PCR. Initially, the cDNA sequence for human cTnI was inserted into a pET vector without modification. Very little cTnI was expressed with this construct. To increase the expression of human cTnI in bacterial host cells, three different modifications were made to the cDNA sequences separately. The first modified clone, designated as rcTnI-0, was prepared with two NH2-terminal codon mutations without changing any amino acid residues, as described by Al-Hillawi et al. (6 ). The second and third modified clones, designated as rcTnI-6 and rcTnI-14, were constructed with added bacteria-favored NH2-terminal leader sequences (7 ); the leader sequences for rcTnI-6 and rcTnI-14 are MASMGS and MASMTGGQQMGRGS, respectively. The human cTnT cDNA sequence was modified as described previously (8 ) to increase its expression in Escherichia coli. TnC was expressed without modification to the cDNA sequence. All constructs were confirmed by DNA sequencing. Expression constructs for each protein were engineered, transformed, and expressed according to Shi et al. (9 ). The purification processes for these expressed proteins were essentially described by Shi et al. (9 ). The concentrations of all troponin subunits after