| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Laboratory of Separative Techniques, IRCCS Ospedale San Raffaele, 20132 Milano, Italy
2
Unit of Metabolic Disease, Azienda Ospedaliera e
Universita’, 35100 Padova, Italy.
3
Endocrinology and Diabetes Unit, Istituto Clinico
Humanitas, 20089 Rózzano, Milano, Italy.
4
Institute of Endocrinological Science, IRCCS Ospedale
Maggiore, 20122 Milano, Italy.
a Address correspondence to this author at: Laboratory of Separative Techniques, Laboratory Analysis, H. San Raffaele, via Olgettina 60, 20132 Milano, Italy. Fax 39-02-26432640; e-mail paroni.rita{at}hsr.it
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Serum samples (1 mL), with internal standard added, were purified by solid-phase extraction on OASISTM HLB cartridges (Waters), dried under reduced pressure, and reconstituted with 30–60 µL of acetonitrile:H2O. Analysis was carried out by micellar electrokinetic capillary chromatography in 5 mmol/L borate, 5 mmol/L phosphate, 75 mmol/L sodium cholate, pH 8.5, containing 25 mL/L methanol. Separation was accomplished in a 20 cm x 50 µm (i.d.) silica capillary at 25 °C and a constant voltage of +10 kV. Pharmacokinetics of gliclazide (80-mg tablet) in a diabetic patient were assayed by both HPLC and CE. Two hypoglycemic patients positive by HPLC analysis for unreported gliclazide and tolbutamide overdose were also screened by CE.
Results: Separation of six drugs (including the internal
standard) was accomplished in 5 min plus 5 min rinsing. The between-day
CV of the ratio of the areas of the sulfonylurea drugs to internal
standard was <1% (n = 10). Linearity
(r2 
0.998) and recovery (80%) were good
for all sulfonylurea drugs tested. Pharmacokinetic curves for
gliclazide by CE and HPLC were superimposable. CE analysis confirmed
the HPLC diagnosis of surreptitious abuse of gliclazide and
tolbutamide.
Conclusion: CE is a useful tool in the clinical chemistry and toxicology laboratory for drug monitoring and pharmacokinetic investigations.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
10 mIU/L (10 µIU/mL)
associated with plasma glucose values <2.5 mmol/L (45 mg/dL) suggests
the presence of insulinoma. However, surreptitious administration of
insulin or SU drugs cannot be ruled out, especially in health
professionals or relatives of a diabetic patient (8).
"Factitious hypoglycemia" is confirmed when the origin of the
insulin increase is demonstrated to be exogenous (suppressed plasma
C-peptide immunoreactivity) or when SU drugs (or their metabolites) are
found in serum or urine (9)(10)(11)(12). Timely discovery of an
unreported overdose of SU drugs could avoid unnecessary exploratory
surgery or a partial pancreatectomy. Because of the importance of accurately diagnosing self-administration abuse, we routinely use HPLC for identification of SU drugs in the blood of suspected patients. The procedure is based on previously published methods (13)(14)(15) and has been optimized for identification, in a single analysis, of the most commonly prescribed SU drugs, including tolbutamide (TL), chlorpropamide (CL), glipizide (GP), gliclazide (GL), and glibenclamide (GB). Positive samples are always re-analyzed by an HPLC method based on precolumn derivatization with fluorodinitrobenzene (16)(17). When some doubt remains, confirmation by mass spectrometry is mandatory.
Capillary electrophoresis (CE) is an analytical technique whose popularity is quickly increasing in the clinical chemistry laboratory as an alternative to traditional gel electrophoresis for proteins, peptides, and oligonucleotides. In particular, a micellar electrokinetic capillary chromatography (MEKC) method, based on the use of surfactant agents, has been specifically designed for the separation of neutral and uncharged analytes (18)(19), and its use as an alternative method for identification of SU drugs in urine has been described (20)(21). None of the above-mentioned studies, however, explored the performance of MEKC for determination of SU drugs in serum, and none reported in detail a quantitative evaluation.
The aim of this study was to evaluate the potential of MEKC both as a qualitative assay to help in identification of SU drugs in serum for rapid and accurate diagnosis of drug-induced hypoglycemia and as a quantitative technique to be used in pharmacokinetic investigations. For this purpose, we assessed serum pharmacokinetics of GL (80-mg tablet) in a type 2 diabetic patient by both HPLC and CE. Moreover, two patients investigated for factitiously administered SU drugs by HPLC were also screened by CE, which confirmed GL and TL abuse.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
preparation of buffers and calibrators
The buffer for HPLC separation was 10 mmol/L potassium hydrogen
phosphate, pH 3.5. Stock borate buffer (0.5 mol/L), pH 8.5, and
phosphate buffer (0.5 mol/L) for CE analysis were prepared as described
previously (20)(21). We prepared CE running
buffer by diluting the stock buffers 1:100 (by volume) and adding 75
mmol/L sodium cholate and 25 mL/L methanol. The CE washing
buffer was prepared in the same way except that the stock buffer was
diluted 1:10.
Stock drug solutions were prepared at 1 g/L in methanol and kept at -20 °C. Working solutions containing 0.2, 0.05, 0.02 g/L of each drug were prepared by dilution in doubly distilled water.
liquid-liquid serum extraction
Serum samples (1 mL) were added to 10 µL of the internal
standard solution (see below) and acidified with 0.2 mL of 1 mol/L HCl.
Extraction was accomplished by mixing with toluene (5 mL) for 15 min.
After centrifugation for 5 min at 3000g, the organic phase
was transferred into a conical screw-capped glass tube and dried in a
vacuum centrifuge (Savant).
solid-phase serum extraction
Serum samples (1 mL) were acidified with 0.2 mL of 1 mol/L HCl,
added to 10 µL of the internal standard solution (see below), and
diluted 1:1 (by volume) with water. The OASIS HLB extraction
cartridges were connected to Visiprep Solid Phase Extraction Vacuum
Manifolds (Supelco), and after pressure was reduced with a
water pump, the cartridges were preactivated by washing with 1 mL of
methanol followed by 1 mL of water (
1 mL/min). The diluted serum
samples were loaded onto the columns and then washed with water (1 mL),
followed by a series of water:methanol mixtures [1 mL of 95:5 (by
volume), 1 mL of 80:20, 1 mL of 70:30, and 1 mL of 60:40].
After air was flushed through the columns for 1 min, the drugs of
interest were eluted with 1 mL of methanol:acetonitrile (1:1, by
volume) and dried under reduced pressure.
hplc analysis
A Kontron instrument composed of two pumps (model 420), an
autosampler (model 460), and a double-beam ultraviolet detector
(model 430) set at 225 nm was used. Kroma System 2000 software was used
for data handling. The column was a LiChrospher 100 RP-18 [25 cm
x 4.5 mm (i.d.); 5 µm bead size; Merck]. The dried samples were
reconstituted with 120 µL of
K2HPO4:methanol (40:60, by
volume), and 20–100 µL was injected into the column. A good
compromise between rapidity of analysis and resolution of TL, CL, GP,
and GL was achieved simply by eluting the column with 600 mL/L
methanol in potassium phosphate buffer. To detect the presence of GB,
however, a gradient to 710 mL/L methanol was necessary (see Fig. 1
). To
further improve the separation of the first eluting peaks, the analysis
was sometimes started at 580 mL/L methanol.
|
ce
CE separation was carried out using a P/ACE 5010 System (Beckman
Instruments) equipped with a monochromatic ultraviolet detector set at
200 nm and controlled by System Gold 8.1 software. The fused-silica
capillary [27 cm x 50 µm (i.d.); 20 cm to the detector] was
assembled in a Beckman cartridge (200 x 400 µm slit aperture).
For a typical analysis, the following procedure was used: 3-min
prerinse with the running buffer (20 psi, 138 kPa), 2- to 5-s injection
of the sample at low pressure (0.5 psi, 3.4 kPa), 1-s pressure
injection of running buffer, separation at +10 kV (37 µA), 1-min
rinse with the washing buffer, followed by 1 min with 0.1 mol/L NaOH.
The capillary temperature was maintained at 25 °C. The dried samples
were reconstituted with 30–60 µL of acetonitrile:water (60:40, by
volume) and loaded into the water-cooled autosampler tray.
calibration curves for ce analysis
Calibration curves were prepared with the working solutions. We
prepared a GL calibration curve in the 1–20 mg/L range by adding to 1
mL of control serum increasing amounts of GL (5–100 µL of a 0.2 g/L
solution), and a fixed amount of TL used as internal standard (10 µL
of a 0.2 g/L solution; total amount added, 2 µg). The samples
were then acidified and loaded on the solid-phase extraction cartridge
as described previously. We prepared a calibration curve for GB in the
0.25–20 mg/L range using the 0.05 and 0.2 g/L solutions. For TL and CL
determination,
N-acetyl-5-(2,3-dichlorophenylureido)benzensolfonamide was
added as internal standard (10 µL of a 0.5 g/L solution; total amount
added, 5 µg). The curves were in the 0.4–20 mg/L range, and the 0.02
and 0.2 g/L solutions were used (20–100 µL). Because of the small
volumes of patients’ sera available, some curves were prepared
as described, but only 0.2 mL of plasma was extracted and analyzed.
pharmacokinetics of gl in serum
To determine whether CE could give reliable results in
identification and quantification of SU drugs, blood from a type 2 male
diabetic patient under chronic treatment with GL (40 mg three times per
day before meals for 1 year) was withdrawn immediately before oral
intake of a 80-mg tablet in the morning, and after 0.5, 1, 2, 3, 4, 6,
and 8 h. During the study the patient had regular meals.
patients
We studied two cases of hypoglycemia after overdose of
factitiously administered SU drugs.
Patient 1 was a 58-year-old woman who attended the outpatient clinic because of severe obesity. She complained of tachycardia, sweating, weakness, and recent hypoglycemic episodes. An insulin-secreting tumor was suspected, but after she was hospitalized for 6 days, the cause of the hypoglycemia had not been diagnosed. The patient’s husband had diabetes and used GL. A screen to confirm surreptitious abuse of SU drugs was requested.
Patient 2 was a 44-year-old woman affected by posttraumatic epilepsy, asthma, chronic gastritis, and depression. She was hospitalized for evaluation of recent hypoglycemic episodes. Her chronic medications included phenobarbital, ranitidine, diazepam, salmeterol xinafoate, and antacid. The patient had familial and psychological problems, and the father was a diabetic being treated with SU drugs. The search for an insulinoma gave negative results, and because of the clinical characteristics of the patient, covert use of SU drugs was strongly suspected. A serum sample was collected and sent for laboratory evaluation.
blood samples
Blood samples were collected in Vacutainer®
Tubes without additives; after separation, the sera were stored at
-20 °C until analysis.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Because of the improved performance, all analyses reported
below (by HPLC or CE) refer to the solid-phase extraction procedure.
|
A typical profile obtained by our reversed-phase HPLC method for
screening of SU drugs is shown in Fig. 1
. The analysis time was 35 min, but good resolution of all
drugs, including the strongly retained GB, was achieved. Fig. 2
shows the CE electropherogram of a mixture of five SU drugs
plus N-acetyl-5-(2,3-dichlorophenylureido)benzensolfonamide,
which has been suggested by Roche et al. (21) as a possible
internal standard for analysis of urine by CE. It is not surprising
that the elution order of the SU drugs obtained by CE is different from
that obtained by HPLC: the resolution in the former technique is based
on the distribution coefficient between the micellar and aqueous phases
(the greater the percentage of the analyte that is distributed into the
micelle, the slower it migrates). In both CE and HPLC, however, GB was
the last compound to elute, likely displaying a higher affinity for
C18-RP resin and higher capacity factor for the cholate micelle. When
the MEKC conditions (slightly modified) described by Roche et al.
(20) were used, the serum components sometimes interfered
with the
N-acetyl-5-(2,3-dichlorophenylureido)benzensolfonamide. We
therefore preferred to use TL as internal standard for quantification
of GL, GB, and GP. Peak identification was carried out in relation to
the migration time (tM) for
the internal standard and by comparison with an external calibrator,
with the area corrected for the tM
always used. Linearity by CE was good for all of the SU drugs. The
regression equations were calculated by plotting the SU drug:internal
standard area ratio vs the serum concentration, and the resulting
correlation coefficients were all 0.998. Examples of the curves for
GL, TL, and GB are reported in Fig. 3
.
|
|
Under the described analytical conditions, blank serum showed no
interference for any of the SU drugs except for GB, for which there was
always a peak present at the same tM,
corresponding to a concentration of 0.060 mg/L, that had to be taken
into account for correct quantification. The minimum detectable
concentrations in serum were 0.2 mg/L for GB and 0.1 mg/L
(signal-to noise-ratio, 3:1) for all other SU drugs. This value could
be improved only minimally by increasing the injected amount because
peak broadening together with comigration of interfering serum peaks
occurred. The HPLC detection limit was much lower (5 µg/L)
because it was possible to inject the entire extracted sample.
The reproducibility of CE system was determined by analyzing mixtures of three different SU drugs (0.066 g/L) 10 times a day and for 10 consecutive days. The relative standard deviation for the peak area and the tM was 2–5% for within-day and 5–7% for between-day reproducibility. When the SU areas were corrected for the internal standard [TL or N-acetyl-5-(2,3-dichlorophenylureido)benzensolfonamide] the between-day CV for the area ratio dropped to <1% for all SU drugs tested. For quantitative purposes, the use of an internal standard is, therefore, essential.
pharmacokinetics of gl in serum
The pharmacokinetic curves for GL in serum, assayed by HPLC and
CE, are shown in Fig. 4
. Despite some slight differences (1.0 vs 0.87 mg/L and 1.06 vs
0.93 mg/L for CE vs HPLC at 2 and 8 h, respectively), the two
methods produced kinetic curves with the same trend and with the
majority of sampling points almost superimposable, thus supporting the
reliability of CE as a quantitative technique. The GL concentration in
serum of the diabetic patient under examination, who had good metabolic
control, was 0.7–1.2 mg/L (23)(24).
|
identification of gl-induced hypoglycemia
The chromatographic profile obtained from the
solid-phase-extracted serum (0.2 mL) of patient 1 is shown in Fig. 5
. An important peak, not present in the control serum (data not
shown) and corresponding to the retention time
(tr) of GL was evidenced
(11.8 min). Quantification against the GL calibration curve in serum
gave a concentration of 12.1 mg/L. The presence of GL in the serum of
the patient was also confirmed by HPLC analysis with precolumn
derivatization (data not shown). The same serum from the patient (0.2
mL) was reextracted and analyzed by CE. The electropherogram confirmed
the presence of a peak that corresponded to the
tM of GL calibrator (3.29 min) and was
quantified as 12.6 mg/L (Fig. 6
).
|
|
The patient repeatedly denied using SU drugs or any other medication, and we were not able to contact her any more after hospitalization.
identification of tl-induced hypoglycemia
The chromatogram obtained for the serum sample collected from
patient 2 is shown in Fig. 7
. The important peak at tr
7.2 min suggested covert self-administration of TL and was estimated
to have a concentration of 11.4 mg/L. A small but significant peak was
also detected at the tr of GB, and
corresponded to a concentrations of 0.035 mg/L. The presence of TL and
GB in the serum of the patient was also confirmed by HPLC with
precolumn derivatization (data not shown). The CE profile from the same
solid-phase-extracted serum (Fig. 8
) confirmed the presence of TL
(tM = 3.55 min) at a concentration
14.1 mg/L. The peak at tM = 3.77 min,
likely a concomitant therapeutic drug or a more polar metabolite, did
not correspond to any of the SU drugs under evaluation. In the present
setting, the CE technique did not have enough sensitivity to detect the
presence of GB.
|
|
During the period when patient 2 was still hospitalized, some GB pills disappeared from the ward. On this occasion, a second blood specimen was withdrawn, and the presence of GB was also confirmed by mass spectrometric analysis. The patient was questioned but denied antidiabetic medication use. After 3 months of psychiatric therapy and social counseling, this patient’s hypoglycemic episodes disappeared and SU drug tests were negative.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The CE technique seemed to offer several advantages compared with HPLC or mass spectrometry: it is very rapid (analysis time <6 min), relatively robust, inexpensive (low volumes of buffers and reagents), and quite reproducible with the use of a suitable internal standard. Coupling with a diode array detector (21) or with a mass spectrometer may improve the accuracy of the assay. If these instruments are not easily accessible, we suggest use of CE as a secondary test for confirming HPLC-positive results.
When a suitable internal standard was used, CE was shown to be valid as a quantitative technique, displaying an excellent linear response for all of the drugs tested in the therapeutic and above-therapeutic range. CE also provided satisfactory performance when compared with HPLC, although because only very small volumes of patients’ sera were available, this evaluation was based on only one aliquot per sample extracted and analyzed with both methods. In the present setting, the only true limitation of this method appears to be the detection limit, which is too high for pharmacokinetic studies on the third generation of SU drugs, which because of their greater potencies are prescribed at very low dosages. Moreover, this could yield false-negative results if patients are investigated more than 24 h after ingestion, when SU drugs have been mostly metabolized and excreted in urine. Kunkel and Wätzing (19), using direct injection of plasma samples for drug quantification by MEKC, found a dramatic improvement of the detection limit (to 0.005–0.010 mg/L) for acetaminophen and salicylate when they added 75–100 mL/L isopropanol to both the running and washing buffers. Despite the fact that we had already modified the buffer used by Roche et al. (21) by adding 25 mL/L methanol and that we performed preanalytical purification and concentration of the sample by solid-phase extraction, the possibility claimed by Kunkel and Wätzing (19) of avoiding comigration with endogenous plasma components and thus increasing the injection time to 15–20 s appears quite promising and merits testing, especially for GB analysis. More efforts in this direction are advisable, however, to look for improved and more sophisticated preanalytical and analytical procedures.
Because of their extensive hepatic metabolism, SU drugs of the third generation are excreted mainly in urine as hydroxylated metabolites (25)(26), and screening for drug abuse in this matrix may sometimes be complex because of the number of different oral hypoglycemic compounds commonly prescribed and the ease of availability for people willing to use these drugs covertly. Hence, despite the advantage that urinary excretion of metabolites occurs for several days after intake, we chose to look for the presence of parent drugs in serum. In addition, urine collection may be difficult in patients who sometimes have extensive psychological and social problems. Our hospitalized patients "suspected" of drug abuse are under strict control, and blood is collected along with the "suppression test" and/or whenever blood glucose concentrations are <2.2 mmol/L (40 mg/dL). It is worth noting, moreover, that factitious disorder often occurs after ingestion of above-therapeutic dosages of the drug.
In this report, we presented two cases of hypoglycemia after an
overdose of SU drugs administered factitiously, disclosed by HPLC and
confirmed by CE. However, when GB concentrations measured by HPLC were
in the therapeutic range (0.030 mg/L), CE failed to validate the
result.
In conclusion, we have demonstrated that CE provides correct quantification of SU drugs in serum and can easily replace HPLC in pharmacokinetic studies (taking into account detection limits). We do not propose CE alone as a technique able to give a definitive diagnosis of drug abuse, even when coupled to diode array scanning. The availability of a method in addition to HPLC based on a different principle, however, may be pivotal in unmasking the illicit use of hypoglycemic agents. To help physicians in insulinoma diagnosis, we currently follow a multistep identification approach that includes different techniques such as reversed-phase HPLC with direct ultraviolet detection (13)(14)(15), reversed-phase HPLC with precolumn derivatization (16)(17), and CE (18)(19)(20)(21). Only when results are still ambiguous or do not agree with biochemical testing and clinical features do we ask for confirmation by the unequivocally most specific (but less readily available) mass spectrometry.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  K. Aaboe, F. K. Knop, T. Vilsboll, A. Volund, U. Simonsen, C. F. Deacon, S. Madsbad, J. J. Holst, and T. Krarup KATP Channel Closure Ameliorates the Impaired Insulinotropic Effect of Glucose-Dependent Insulinotropic Polypeptide in Patients with Type 2 Diabetes J. Clin. Endocrinol. Metab., February 1, 2009; 94(2): 603 - 608. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Hoizey, D. Lamiable, T. Trenque, A. Robinet, L. Binet, M. L. Kaltenbach, S. Havet, and H. Millart Identification and Quantification of 8 Sulfonylureas with Clinical Toxicology Interest by Liquid Chromatography-Ion-Trap Tandem Mass Spectrometry and Library Searching Clin. Chem., September 1, 2005; 51(9): 1666 - 1672. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
).
) corresponded to
a concentration of 0.035 mg/L. Analysis started with 60% methanol
(%B); gradient began at 9 min and increased to 71%
over 4 min.