| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1 Servicio de Análisis Clínicos, Hospital San Agustín, Avilés, and Departamento de Bioquímica y Biología Molecular, Universidad de Oviedo, Camino de Heros 4, 33400 Avilés, Asturias, Spain. Fax 34-985-123-025; e-mail r.venta.000{at}recol.es.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Method performance was continuously assessed by recovery and precision studies with urine samples and controls, respectively. Healthy individuals were prospectively analyzed throughout a 5-year period. Excretion of individual amino acids, expressed as mmol/mol of creatinine, was included in six age-related groups for random urine samples (0–1 month, 1–12 months, 1–3 years, 3–8 years, 8–16 years, and >16 years) and in two groups for 24-h urine collections (8–16 years and >16 years).
Results: Over a 1-year period, CVs for retention times were <0.5% and 3.3% for within- and between-run imprecision, respectively. For amino acid concentrations, within-run CVs were 2.9–17% and between-run CVs were 7.1–46% for the same period. Amino acid recoveries were 78–122%. Reference intervals for 35 amino acids were calculated and compared with the concentrations observed in patients diagnosed with specific pathologies. A few statistically significant differences were found between the reference intervals derived using random and 24-h urine collections.
Conclusions: Long-term reliability of the RP-HPLC method for urine amino acid analysis has been demonstrated. Representative age-related reference intervals for the RP-HPLC method in both random urine and 24-h urine collections have been established, and their feasibility for diagnosis of aminoaciduria has been shown. These intervals could serve as a guide for laboratories changing to HPLC methods.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Physiological amino acids have been determined extensively in clinical laboratories by automated ion-exchange chromatography in combination with postcolumn ninhydrin detection (3)(4). This technique provides simplicity in sample preparation and highly reproducible results (5)(6) and has been considered as the reference method (2). On the other hand, reversed-phase HPLC (RP-HPLC)1 has progressively become an alternative in the last decade for both protein hydrolysates and biological fluids (7)(8).
RP-HPLC offers higher sensitivity, versatility, and speed of analysis than classical ion-exchange separations (9)(10). The more usual methods involve precolumn derivatization with phenylisothiocyanate (PITC) (11)(12) or o-phthalaldehyde (OPA) (7)(13). Although OPA methods show a higher sensitivity and more suitable features for automation than PITC methods, OPA derivatives are highly unstable, and direct detection of secondary amino acids (proline and hydroxyproline) and sulfur amino acids (cysteine and homocysteine) is not possible (3). OPA methods have been recommended for quantification of primary amino acids in physiological fluids, whereas PITC methods have been proposed when both primary and secondary amino acids are to be determined (9).
Every laboratory should always provide method-related reference values. Amino acid concentrations in biological fluids are related to different factors, such as diet, protein metabolism, circadian rhythms in metabolism, and excretion of each amino acid (1). Particularly, urinary amino acid excretion shows age-related physiological variations mainly related to the progressive maturation of the tubular reabsorption systems (1)(14). This leads to an age-dependent decrease of amino acid excretion, which prompts laboratories to define reference values for several classes (15)(16).
In analysis of urinary amino acid excretion in infants, incontinent patients, or general outpatients, it often is very difficult to obtain 24-h urine collections. Therefore, random urine samples frequently are collected for amino acid analysis, and the amino acid values are expressed relative to creatinine (15)(16). The excretion rates of amino acids may vary independently of creatinine over a 24-h period. Nevertheless, creatinine has been shown to be the reference compound with the lowest variance both within and between individuals (17)(18). Moreover, the variance of urinary amino acid excretion has been shown to improve when the concentrations in urine are related to creatinine (17).
The range of any reference interval can be considered as the sum of both biological and methodological variability; therefore, the latter must be continuously assessed by a long-term quality-control program to guarantee the clinical utility of the established intervals (19), and validation of the long-term method performance should always be included during the production of new reference values. Appropriate urinary amino acid reference values for RP-HPLC methods have not been comprehensively reported. On the other hand, good correlation of RP-HPLC data with those derived by the reference ion-exchange chromatography method has been shown (11)(12), but differences in sensitivity and sample preparation procedures may yield potential differences in the reference values. Therefore, we studied the long-term precision and recovery of a commercial PITC RP-HPLC method and derived specific age-related reference values for urine amino acid excretion in relation to creatinine in both untimed and 24-h urine collections. These reference intervals were compared with those published for ion-exchange methods and with the amino acid concentrations observed in patients diagnosed with specific pathologies.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
analytical procedures and sample preparation
Measurements of amino acid concentrations in urine were performed
by RP-HPLC using the Pico-Tag method (Waters) according to
manufacturer’s specifications (20). The analytical system
consisted of a Baseline 810 workstation running on a Hewlett-Packard
Vectra 486 computer, two Model 510 pumps, a Model 712 WISP autosampler,
a temperature control module (set at 46 °C), and a Model 484
absorbance detector (set at 254 nm). All chemicals and solvents were
purchased from Sigma-Aldrich and Merck. HPLC-grade water was generated
with a MilliQ water purification system from Millipore.
On the day of analysis, the samples were thawed at room temperature and centrifuged for 15 min at 3000g. The clear supernatants were further cleaned up by solid-phase extraction. Briefly, 1 mL of each sample, diluted 1:1 with the internal standard (400 µmol/L methionine sulfone in 0.1 mol/L HCl), was applied to a preconditioned (10 mL of methanol, 10 mL of water) Sep-Pack Plus C18 cartridge on a Sep-Pack vacuum manifold (Waters). Purified amino acids were acquired by collecting the initial pass-through sample plus two sequential eluates of 1.5 mL of 0.1 mol/L HCl and 2.5 mL of 300 mL/L acetonitrile in 0.1 mol/L HCl, respectively.
Derivatization consisted of three steps of evaporation to dryness using a Waters vacuum station. The first drying step was carried out on the calibration mixture (25 µL) and urine eluates (50 µL). The second drying step was carried out after treatment with 10 µL of methanol-1 mol/L sodium acetate-triethylamine (2:2:1 by volume), and the third step after treatment for 10 min with 20 µL of derivatization reagent [methanol-triethylamine-water-PITC (Pierce), 7:1:1:1 by volume]. After derivatization, the phenylthiocarbamyl amino acid derivatives were separated on the C18 Pico-Tag physiological free amino acid column [300 x 3.9 mm (i.d.)] using a stated binary gradient of Waters eluents 1 and 2 at a flow rate of 1.0 mL/min.
Urinary creatinine was measured by the kinetic Jaffe reaction in a Hitachi 911 analyzer (Roche Diagnostics).
calibration and quality control
Derivatized calibrators, controls, and patient samples were
dissolved in 100 µL of the sample diluent (Waters), and 10 µL of
these solutions was injected. A calibration mixture was prepared and
derivatized every 2 months from two commercial solutions
(Sigma-Aldrich) of acid-neutral and basic amino acids in 0.1 mol/L HCl.
A separate solution of glutamine and argininosuccinic acid
(Sigma-Aldrich) in water, stored at -20 °C, was also used. The
concentration of each amino acid in the final mixture was 400 µmol/L,
except for cystine (200 µmol/L). Amino acid derivatives were
identified by their relative retention times based on the reference
peaks for retention time (methionine sulfone and norleucine, both of
which were added to the internal standard solution). The ratio of the
peak area of each amino acid to the internal standard (methionine
sulfone) was used to calculate the amino acid concentrations, and the
final results were expressed as millimole of amino acid per mole of
creatinine. Each urine sample was analyzed once.
Within- and between-run precision was continuously monitored with the
Lyphochek Quantitative Urine Control 2 (Bio-Rad). Duplicate control
samples were assayed bimonthly after fresh reconstitution. Within-run
SDs were calculated from replicate differences (D) as paired
SD =
(
D2/2n)1/2
(21), and between-run SDs were calculated as usual from the
first result of each analytical run.
For the recovery study, two urine pools were prepared according to the creatinine concentration measured in the urine samples, and five samples of each pool were enriched with a concentrate of the amino acid calibration mixture (200 µmol/L). All of these samples as well as the calibration mixture, at the same dilution in 0.1 mol/L HCl, were assayed, and the mean recovery percentage, corrected for the internal standard, was calculated.
statistical analysis
Statistical analyses were performed according to the Expert Panel
on the Theory of Reference Values from the IFCC
(22). The Anderson-Darling test (23) was used to
assess whether the distribution of test results was gaussian. Outliers
were excluded according to the Dixon test (24). Results were
expressed as mean ± 1.96 SD (0.95 reference intervals),
and the estimated confidence intervals (ß = 0.9) were also
calculated. Statistical significance of differences in mean values was
calculated by the unpaired Student t-test. MS Excel 97
software (Microsoft) was used for data processing and statistical
analysis.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Phenylthiocarbamyl derivatives for 39 amino acids and some related
compounds (such as amines and dipeptides) were identified in the
calibration mixture, but only 36 derivatives were regularly identified
in control samples according to their retention times. In the urine
control, phosphoserine and argininosuccinic acid were undetectable, and
-aminobutyric acid and citrulline overlapped, eluting as a single,
small peak with an average retention time assigned for
presentation to citrulline. Within-run CVs were <0.5% for every peak,
whereas the highest between-run CVs were 2.8% and 3.3% for threonine
in calibration and control solutions, respectively. Retention times of
peaks migrating between 8 and 20 min showed poorer precision. In
general, peaks were slightly delayed in control samples, and the
retention time imprecision was slightly higher for control samples.
To estimate accuracy, we performed two recovery studies during the last year using two different columns with 265 and 30 injections, respectively. In the first study, in which a dilute urine was used, mean recoveries were 87–122%, with the exceptions of argininosuccinic acid (132%) and aminobutyric acid (81%). In the second study, in which a concentrated urine was used, mean recoveries were 78–120%, with the exception of phosphoserine (50%) and glutamic acid (67%). In concentrated urine, derivatives migrating in the first 12 min seemed to have poorer recoveries than those migrating later. Tables showing the within- and between-run precision for retention times and the recovery data and within- and between-run precision for amino acid concentrations are included in an electronic supplement in the on-line version of this article (http://www.clinchem.org/content/vol47/issue3/).
Method precision over 1 year of study showed the well-known dependency on amino acid concentration. Within-run CVs ranged from 2.9% for 1-methylhistidine to 17% for arginine, with the exception of hydroxylysine (22%). Between-run CVs ranged from 7.1% for lysine to 46% for hydroxylysine. In general, amino acid concentrations >165 µmol/L showed between-run CVs <10%, and amino acid concentrations between 30 and 165 µmol/L showed CVs <20%, with the exception of ethanolamine (29%). CVs >25% were observed in some amino acids present in concentrations <15 µmol/L (arginine, proline, aminobutyric acid, methionine, cystine, isoleucine, and hydroxylysine).
All amino acids detected in the urine control were also quantified in
urine samples from patients to establish adequate reference intervals,
regardless of the fact that some of them either were not useful for
diagnosis or required special sampling for clinical management
(1-methylhistidine, carnosine, tryptophan, phosphoethanolamine,
ethanolamine). Table 1
shows the reference intervals for amino acid excretion in
selected age ranges for random urine samples and 24-h urine collections
provided by separate groups of subjects. The age ranges were selected
to represent physiological periods such as the newborn period, infancy,
early and late childhood, puberty, and adulthood. A reference interval
for anserine is not included because the peak frequently was
undetectable. Positively skewed distributions were found in almost
every amino acid age group; therefore, log-transformations were
performed on every group of original data. As a result,
log-distributions gave rise to asymmetric intervals around the mean.
Taurine showed extremely skewed distributions with numerous outlying
values; therefore, corresponding reference intervals are presented as
the observed ranges.
|
In general, amino acid excretion decreases with age. Some exceptions include phosphoethanolamine and 3-methylhistidine, which showed a relatively constant excretion throughout childhood. 1-Methylhistidine excretion increases during the first year of life with subsequent fluctuations at higher concentrations. Taurine excretion decreases during the first year of life and afterward remains constant, with a new increase in adulthood. Statistical comparisons of age-matched intervals probed the differences between random and 24-h urine samples in unrelated amino acids (ß-aminoisobutyric acid, carnosine, and arginine for the 8–16 years group; and aspartic acid, ß-alanine, threonine, carnosine, valine, and hydroxylysine for the adult group). Likewise, only isolated statistical sex differences in unrelated amino acids could be observed within groups of subjects older than 8 years (data not shown).
Chromatograms for nonpathological and pathological urine
samples are shown in Figs. 1
and 2
. Table 2
shows the urinary excretion of representative amino acids in
random and 24-h samples from patients with cystinosis, Lowe
syndrome, and phenylalanine-restricted or -unrestricted phenylketonuria
as well as from a random urine sample of a diet-restricted patient with
leucinosis. Table 2
also displays the cystine and dibasic amino acid
excretion from phenotypically homozygous and heterozygous cystinuric
patients. The lowest and highest concentrations are shown as well as
ranges of the multiples of corresponding upper reference limits.
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-amino acids such as imino acids and ß-amino acids. Likewise,
>100 compounds and pharmacological metabolites have been
comprehensively studied as potential interferents
(11)(12).
We have provided recovery and precision data obtained over the last
year of our study as representative of the long-term performance of the
method, but equivalent results for recovery and precision were obtained
during the rest of the study. Over this 1-year period using three
different columns, the peak positions for 37 amino acids were
reproducible enough to ensure correct peak identifications in
patients’ samples. Co-elution of -aminobutyric acid and citrulline
has previously been documented (12), but it does not
represent a serious problem because of the low concentrations of
-aminobutyric acid expected. The precision of retention times was
similar to that seen in short-term studies in plasma with PITC HPLC
methods (12), and in plasma and urine with ninhydrin
automated ion-exchange chromatography methods (5).
In concentrated urine samples, based on a high creatinine concentration, peaks eluting from the column in the first 12–16 min showed poorer recoveries. These poorer recoveries may have resulted from insufficient clean-up of the sample with the solid-phase extraction procedure, which is intended mainly to clear the first part of the chromatogram. Insufficient clean-up leads to more complex chromatograms and poorer integration results. As a routine practice, dilution of urine samples with creatinine concentrations >10.0 mmol/L has been performed. The recoveries in the dilute pool were comparable to those for urine samples of the phenylhydantoin HPLC method (65–100%) (25), the 9-fluorenylmethyl chloroformate-OPA HPLC method (83–105%) (26), and the ninhydrin automated ion-exchange chromatography method (94–115% to 75–140%, depending on the de-ammoniation procedure) (6). Likewise, recoveries were equivalent to those obtained for plasma samples using the PITC HPLC method (89–117%) (11) and the OPA HPLC method (79–121%) (9).
The results of our within- and between-run precision studies over 1 year seem better than those described by Feste (12) (5.7–129% for the same method in a short-term scheme). Both studies confirm the poorer reproducibility for concentrations <15–30 µmol/L occasionally found for some amino acids (such as arginine, proline, methionine, or isoleucine) with more efficient reabsorption mechanisms (14). These long-term CVs are poorer than those found with the 9-fluorenylmethyl chloroformate-OPA HPLC method (26) both within (2–24%) and between runs (4–27%) as well as with the most recently developed methods using ninhydrin automated ion-exchange chromatography [within-run CV <4–19.8% and between-run CV <10–23.1% (5)]. Although both studies report short-term CVs, the better precision could be explained by the automation of their procedures as opposed to the manual sample procedure used in the PITC HPLC method. The generalized high imprecision of the amino acid urine analyses highlights the importance of using an appropriate urine control with physiological concentrations rather than calibration mixtures or plasma controls in method validation and quality assurance.
The reference values obtained in random urine samples confirm previous findings of an age-related decrease in the excretion of amino acids adjusted for creatinine concentration (15)(16) attributable to the phenomenon of renal adaptation during childhood (1)(14). The validity of random urine samples for the diagnosis of aminoaciduria has already been established using ion-exchange chromatography methods, although it is commonly accepted that there is a higher variation of amino acid concentrations in untimed samplings (15) and in 12-h urine collections (17) compared with 24-h collections. Nevertheless, no important differences in mean amino acid concentrations of untimed, 12-h, and 24-h paired urine samples have been reported in adults (15)(17). Comparisons of the current means of intervals corresponding to unpaired samplings confirm these findings, demonstrating only a few significant differences between random and 24-h urine samples in age-matched groups. Moreover, higher limits for 24-h intervals have been observed in several amino acids.
The differences observed between specimens could be explained in several ways: (a) as an energy source, amino acids are consumed in bacterial contamination (27), which could be more prominent in the 24-h urine specimens; (b) a diurnal variation has been observed in amino acid excretion (15)(17), with the lowest values in early morning and the highest in late evening; and (c) random urine samples could be more affected by food intake than 24-h urine collections, with early morning, pre-breakfast specimens being the least affected (28). In our study, gross bacterial contamination was visually excluded for each sample at reception. We had little control over the time of sampling in random specimens, but based on the times of their reception in the laboratory, most of them were collected before 1200.
Some amino acids, such as phosphoethanolamine, methylhistidines, and taurine, show clearly different age-dependent patterns of excretion. Constant or increased excretion of these amino acids during childhood probably is related to food intake; for example, methylhistidine excretion can be increased by a diet rich in poultry, rabbit, or pork (1)(27). 3-Methylhistidine excretion is related to muscle protein breakdown (29), and creatinine is related to muscle mass, supporting the age-constant pattern of the ratio. The decreases in taurine excretion during the first year of life could be related to the progressive substitution of taurine-rich maternal milk for other foods (1)(30), whereas increased ingestion of meat and shellfish in adulthood could increase taurine concentrations (27). These distinct excretion patterns have been described previously by Parvy et al. (16), using a ion-exchange chromatography method.
ß-Aminoisobutyric acid excretion showed an extremely skewed distribution, especially in children younger than 8 years. The use of cutoff values to identify high excretors (1)(16) in ion-exchange chromatography methods has been common. In our study, no clear cutoff values were found, but we tentatively used 113 and 57 mmol/mol of creatinine for children younger and older than 8 years, respectively (i.e., 1000 and 500 µmol/g of creatinine), according to the highest excretions observed. These values would provide a frequency of high excretors in children of 8%. Similar frequencies between 4% and 10% have been reported for Western countries (1).
Because of differences in age categories, direct comparison with other studies is not possible. Nevertheless, a general overview of our reference values suggests intervals similar to those reported by Tsai et al. (15) in untimed samples, particularly in children and adults. For glycine, we found lower excretion in infants and children younger than 8 years. With regard to the intervals reported by Parvy et al. (16) in pre-breakfast samples, we found higher reference intervals for almost every amino acid, particularly in newborns and infants. However, the means of the intervals for some of the most prominent amino acids, e.g., threonine, serine, glutamine, glycine, alanine, tyrosine, and phenylalanine, were similar. For 24-h urine collections in adults, the reference intervals for our HPLC method were also greater than intervals reported previously (31). All of these cited studies were performed with the ninhydrin ion-exchange chromatography method.
The observed differences in method-related reference intervals could be explained in several ways: (a) the increased sensitivity and better resolution of RP-HPLC columns (3)(9) as well as the differences in the sampling period; (b) the commercial origins of the calibration solutions, which have been considered responsible for dispersion of results in quality-control schemes (4) and therefore could partially explain the differences in reference intervals; and (c) the skewed distributions of amino acids in both plasma (32) and urine (1), which have long been known; therefore, the use of recommended statistical treatments of the data could yield reference limits different from the ranges reported in other studies.
Some examples of representative amino acid excretion in patients with aminoaciduria support the reliability of the reference intervals obtained. A partial overlapping with the corresponding reference intervals was found for the first sample of a 3-month-old patient with Lowe syndrome, but a subsequent sample at 17 months of age showed a characteristic generalized hyperaminoaciduria. This pattern has been described elsewhere (33). According to the recent studies of Guillén et al. (34), using the Waters Pico-Tag method, cystinuric patients can be classified as phenotypic homozygotes and heterozygotes by the use of a cutoff of 113 mmol/mol of creatinine for cystine excretion. From our 26 diagnosed cystinuric subjects, 24 patients could be correctly classified by cystine excretion. Our results confirm the conclusion of Guillén et al. regarding lysine as the better discriminator between non-type I heterozygotes and the general population, because it always shows values above the upper reference limit.
In conclusion, the advantages of the PITC HPLC method, such as high resolution, speed, and sensitivity for quantifying of imino acids, sulfur, and ß-amino acids, make it suitable for urine amino acid analysis, and its reliability and feasibility for clinical use has been demonstrated. However, its performance in terms of precision should be improved to achieve the precision of the reference method, automated ion-exchange chromatography using ninhydrin. The prospective selection of the reference patients in this study probably yielded wider and more practical reference intervals than those obtained using more stringent conditions. On the other hand, small deviations from reference intervals can be common and generally have no clinical significance (3). Although RP-HPLC methods to analyze plasma amino acids are currently in use in some clinical laboratories, the measurement of urinary amino acids is most often performed by ion-exchange chromatography. Hopefully, these reference intervals will serve as a guide for laboratories moving to this newer technology.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
