Anal. Chem. 1994,66, 244-248

Column-Switching Techniques for Screening of Diuretics and Probenecid in Urine Samples Pilar Campins-Falc6,’ Rosa Herrhez-Hernhdez, and Adela Sevlllano-Cabeza Departamento de Quimica Ana fitica, Facultad de Quimica, Universidad de Valencia, Dr. Moliner 50, 46 100 Burjassot, Valencia, Spain

A method based on high-performance liquid chromatography using column-switching is described for the screening of diuretics and probenecid in urine samples. The system uses a 20- X 2.1-mm i.d. precolumn, packed with a Hypersil ODS- C18, 30-pm stationary phase, for the on-line sample cleanup and enrichment. Untreated urine samples are directly injected, and the precolumn is flushed for 1 min with water to eliminate polar matrix components. The retained analytes are then back- flushed by means of a six-port switching valve onto a Hypersil ODS-C18 analytical column ( 5 pm, 250- X 4-mm i.d.), where they are separated using an acetonitrile/phosphate buffer (pH = 3) gradient elution. Under these conditions, the separation and identification of diuretics and probenecid can be achieved with satisfactory selectivity and sensitivity. The described procedure is very simple and rapid since no off-line manipulation of the sample is required, the total analysis time being ca. 15 min.

Liquid-liquid extraction followed by solvent evaporation is the traditional method for sample preparation in the chromatographic analysis of diuretics and probenecid in biological samples. These procedures are usually very time- consuming and often imprecise, as many sample manipulations are usually involved. In addition, multistep extractions under different pH conditions may be required for screening tests, due to the wide differences in the polarities of these drugs.],* Solid-phase extraction on disposable cartridges has been reported to simplify chromatographic quantification of dif- ferent diuretics. We have recently found that this technique can be successfully applied for sample cleanup in the screening of diuretic^.^ The employment of apolar (C8 or C18) solid- phase extraction columns is clearly advantageous over liquid- liquid extraction procedures because a unique extraction can provide acceptable recoveries for the most common diuretics.

In an effort to reduce the time required for the chromato- graphic assay of these compounds in biological fluids, different direct injection techniques have been e ~ p l o r e d . ~ The em- ployment of special packing^^,^ or the addition of a surfactant as a mobile-phase modifier (micellar c h r ~ m a t o g r a p h y ) ~ ~ ~ has also been reported to allow the direct injection of biological samples in the quantification of some diuretics.

* To whom correspondence should be addressed. (1) Fullinfaw, R. 0.; Burry, R. W.; Moulds, R. F. W . J . Chromatogr. 1987,415,

(2) Cooper, S. F.; MassC, R.; Dugal, R. J . Chrornatogr. 1989, 489, 65. (3) Campins-Falc6, P.; Herrbez-Hernbndez, R.; Sevillano-Cabeza, A. J. Liq.

(4) Farthing, D.; Gehr, T. W. B.; Fakhry, I.; Sica, D. A. LC-GC 1991, 9, 478. (5) Pinkerton, T. C.; Perry, J. A.; Rateike, J. D. J. Chromatogr. 1986, 367,412.

347.

Chromatogr. 1991, 14, 3575.

244 AnalyticalChemistry, Vol. 66, No. 2, January 15, 1994

I Waste

Figure 1. Scheme of the column-switching system used for the screening of diuretics and probenecid: position 1 (-), position 2 (- - -).

As an alternative to these methodologies, in recent years, an increasing number of HPLC methods incorporating on- line sample cleanup by solid-phase extraction using column- switching have been developed for the assay of several drugs. Switching devices permit the off-line multistep methods for sample treatment to be transformed into single-step procedures by the on-line purification of the samples.

Since rapid and sensitive methods are required for screening of diuretics, especially in therapeutic drug monitoring and in control of doping, we have evaluated the potential of column- switching techniques for sample cleanup and enrichment of these compounds in urine samples. Probenecid, a uricosuric acid, has also been included in this study because it has a weak diuretic activity and has been used as a masking agent in sports to decrease urinary excretion of anabolic steroids.

EXPER I MENTAL SECTION Apparatus. A schematic representation of the chromato-

graphic system used is shown in Figure 1. This system consisted of two quaternary pumps (Hewlett-Packard, 1050 series (Palo Alto, CA)), an automatic sample injector (Hewlett-Packard, 1050 series) with a sample loop injector of 100 pL, and a high-pressure six-port valve (Rheodyne model 7000). A diode array detector (Hewlett-Packard, 1040 series) linked to a data system (Hewlett-Packard HPLC Chem Station) was used for data acquisition and storage. The detector was set to collect a spectrum every 640 ms (over the range 200-400 nm). The identity of each compound was established by comparing the retention times and UV spectra in real samples with those previously obtained by injection of standards.

(6) Santasania, C. T. J . Liq. Chromatogr. 1990, 13, 2605. (7) Dadgar, D.; Kelly, M. T. Analyst (London) 1988, 113, 1223. (8) Bonet Domingo, E.; Medina Hernlndez, M. J.; Ramis Ramos, G.; Garcfa

Alvarez-Coque, M. C. J . Chromatogr. 1992, 582, 189.

0003-2700/94/0366-0244$04.50/0 0 1994 American Chemical Society

Reagents. All the reagents were of analytical grade. Methanol and acetonitrile were of HPLC grade from Scharlau (Barcelona, Spain). Water was distilled, deionized, and filtered in nylon membranes, 0.45 pm, from Teknokroma (Barcelona, Spain). Diuretics standard solutions were pre- pared by dissolving in methanol pure compounds: amiloride hydrochloride, hydrochlorothiazide, and chlorthalidone from ICI-Pharma (Pontevedra, Spain), acetazolamide from Cy- anamid IbCrica (Madrid, Spain), triamterene, bendroflume- thiazide, ethacrynic acid, and probenecid from Sigma (St. Louis, MO), furosemide from Lasa (Barcelona, Spain), cyclothiazide and bumetanide from Boheringer Ingelheim (Barcelona, Spain), and spironolactone from Searle IbCrica S. A. (Madrid, Spain). Propylamine hydrochloride from Fluka (Buchs, Switzerland) and sodium dihydrogen phosphate monohydrate from Merck (Darmstadt, Germany) were also used.

Columns and Mobile Phases. The precolumn (20- X 2.1- mm i.d.) was dry-packed with a Hypersil ODS-C18, 30 pm (Merck) or SynChropack 30-70 pm (Hewlett-Packard) stationary phase. A Hypersil ODS-C18,5 pm column, 250- X 4-mm i.d., (Merck) was used as the analytical column.

Water was used as the washing solvent for the precolumn (solvent 1). As in previous s t ~ d i e s , ~ an acetonitrile/0.05 M phosphate buffer (pH = 3) mixture (solvent 2) in gradient elution mode was used for theanalytical separation. Phosphate buffers were prepared by dissolving the appropriate amount of sodium dihydrogen phosphate monohydrate in 500 mL of distilled and deionized water; 0.7 mL of propylamine hydro- chloride was added to this solution, and then, the pH was adjusted by adding the minimum amount of concentrated phosphoric acid. All the solutions were prepared daily, filtered with a nylon membrane (0.45 pm from Teknokroma), and degassed with helium before use. The flow rate was 1 mL/ rnin for both pumps, and all the assays were carried out at ambient temperature.

Standard Solutions. The standard solution of each diuretic was prepared by dissolving 50 mg of the pure compound in 25 mL of methanol (2000 pg/mL); the triamterene standard solution was prepared by dissolving 100 mg of the pure compound in 250 mL of methanol (400 pg/mL). All the solutions were stored in the dark at 2 OC. Working solutions were daily prepared by dilution of these stock solutions with the appropriate volumes of distilled, deionized, and filtered water.

Column-Switching Operation. At the beginning of the assay (time zero), 50 pL of the sample is injected from the injector, the switching valve being in position 1 (see Figure 1). By pumping solvent 1, the polar components of the matrix are directly wasted-out, whereas the analytes are trapped in the precolumn. At the same time, the analytical column is being reequilibrated by solvent 2. When the switching valve is rotated (position 2), solvent 2 elutes the retained compounds in back-flush mode, from the precolumn to the analytical column. Different times for flushing the precolumn were investigated in order to optimize the selectivity and the sensitivity. A gradient was used to increase the acetonitrile content in solvent 2 from 15% at 0-1.5 min to 80% at 9.5 min.

(9) HerrBez-Hernhdcz. R.; Camplns-Falc6, P.; Sevillano-Cabcza, A. J . Llq. Chromatogr. 1992, I S , 2205.

After 9.5 min, the acetonitrile content was kept constant. Therefore, the retention and resolution of diuretics are similar to those previously obtained for a conventional chromato- graphic process using a similar analytical column (the volume between the point of mixing of the phosphate buffer and acetonitrile and the precolumn inlet is less than 0.1 mL).9 At 12 min, the switching valve is turned back to the original position to regenerate and reequilibrate both the precolumn and the analytical column.

Urine Samples. Volumes of 1 mL of untreated urine samples were placed into glass injection vials, and 50 pL was directly injected into the chromatographic system.

Recovery Studies. Urine samples were spiked with diuretic standard solutions to give a concentration of 10 pg/mL, and volumes of 50 pL were processed. The percentage of the drug recovered for a particular procedure was calculated by comparing the peak areas obtained for each diuretic in the spiked samples with the values obtained for a direct injection of 50 pL of an aqueous solution containing 10 pg/mL of each compound into the analytical column. Recoveries from aqueous solutions containing 10 pg/mL were also investigated. Each concentration was assayed in triplicate.

Limits of Detection. The limit of detection was estimated by analysis of decreasing concentration solutions of each diuretic in urine. It was established as the concentration required to generate a signal-to-noise ratio of 3.

Human Studies. Real studies were performed with healthy volunteers after a single administration of the lowest recom- mended dose of probenecid (500 mg), ethacrynic acid (25 mg), and chlorthalidone (25 mg). Urine samples were collected at appropriate time intervals postdose and processed as described above.

RESULTS AND DISCUSSION Precolumn Packing. In previous studies with different solid-

phase extraction cartridges,3 we found that the strongest retentions for most of the interesting compounds can be obtained with apolar materials, such as octyl or octadecyl silica packings. This later packing has been recommended for screening purposes.10 Consequently, we have selected octadecyl-bonded materials as the stationary phase for the precolumn. Two different C18 packings have been tested, Hypersil ODS-C18, 30 pm, and SynChropack C18, 30-70 pm. The retention of all the investigated compounds is greater when a Hypersil ODS-C18 precolumn is used. Therefore this material offers the possibility of increasing the time for flushing the precolumn. As a result, greater selectivity in the cleanup process can be achieved.

Switching Parameters. As can be seen in Figure 2, when water is used as solvent 1, a vast majority of the matrix components are eluted from the precolumn in retention times lower than 1.5 min. On the other hand, the recovery of each compound for various durations of the flushing stage is shown in Figure 3. The recoveries obtained for most of the diuretics tested are almost complete and independent of the duration of the flushing in the 1 .O-2.0-min interval. Only furosemide and bendroflumethiazide show values lower than 90% when 2.0 min is used to wash the precolumn. Acetazolamide and

(10) Campfns-Falc6, P.; HcrrBcz-HernBndez, R.; Scvillan&abcza, A.; J . Chro- matogr. 1993, 612, 245.

Analytical Chemistry, Vol. 66, No. 2 January 15, 1994 245

! 000-

3 . 5 - i m e ( r n i n . )

Flgure 2. Chromatogram of a urine sample obtained with the precolumn directly connected to the detector. Conditions: detection wavelength, 230 nm; precolumn packing, Hypersil ODS-C18,30 wm; flushing eluent, water.

-Arni lor ice 120

I Acetazc av ic ie

7001 + hvdro:qlorotpiaz O E

8 80; v

+ Tr,arnterene

+ Chiort-al iaone

6 60- \ + FJrcsemid? > \

+ Cyc'otn i iz ?e

+ BencroflLmethiaz d e

+ Bume'aiide

* Ethac,jn : A c o

* P r c o e r e c 3

8 401 I,\

,I 1 5 2 +?- SDiwnc actcne

cf

20 1

0-

Time (min) Flgure 3. Recoveries of diuretics and probenecid from the precolumn for different durations of flushing. Conditions: precolumn packing, Hypersil ODs-Cl8, 30 pm; flushing eluent, water.

hydrochlorothiazide are partially recovered, which can be explained because of their greater polarity. Similar behavior was observed when solid-phase extraction cartridges were used for sample purification3 These drugs cannot be detected when the duration of the flushing is extended to 2 min. Since an acidic medium can increase the neutral state of acetazola- mide,*J a 0.01 M phosphate buffer was tested as solvent 1. However, no significant improvement in the recovery of any of the investigated compounds was observed. Consequently, water was preferred for washing the precolumn.

In order to obtain a suitable sensibility and selectivity, the duration of the flush step was set a t 1 min. Under these conditions, the recoveries of acetazolamide and hydrochlo- rothiazide are acceptable, whereas the matrix components retained in the precolumn do not affect the stability of the analytical column.

Figure 4 illustrates the chromatograms obtained under the proposed conditions for an aqueous standard solution con- taining a mixture of diuretics and probenecid (Figure 4a) and for a blank urine sample (Figure 4b). By comparing these figures, it can be derived that the chromatograms are free from endogenous components which may interfere with the identification of the drugs screened in this study. Aceta- zolamide and caffeine present close retention times (5.35 and 5.48 min, respectively), although they can be distinguished by

a 4 0 0 j

2 4 6 T i m e ( m i n . )

b 2 5 0 1

U

1 0 0 = E l S O 1

,-

8

ia

0 2 4 6 8 10 T i m e ( m i n . )

Flgure 4. Chromatograms of an aqueous standard solution containing a mixture of diuretics and probenecid (a) and a blank urine sample (b). Peaks: 1, amiloride; 2, acetazolamide; 3, hydrochlorothiazide; 4, triamterene; 5, impurity; 6, chlorthalidone; 7, furosemide; 8, cyclothlazlde stereoisomers; 9, bendroflumethiazide; 10, ethacrynic acid; 11, bumetanide; 12, probenecid; 13, spironolactone. Conditions: detection wavelength, 230 nm; precolumn packing, Hypersil ODS-Cl8, 30 pm; flushing eluent, water; duration of flushing, 1 min; concentratlon of each compound, 10 wg/mL.

Table 1. Recovery Percentages of Dluretlcs and Probenecid from Aqueous Solutlons and Spiked Urine Samples'

recovery recovery compound in water (% ) in urine (%)

amiloride acetazolamide hydrochlorothiazide triamterene chlorthalidone furosemide cyclothiazide bendroflumethiazide ethacrynic acid bumetanide probenecid spironolactone

88.2 f 0.6 47 f 5 63 f 2 98 f 2 95 f 2

102 f 2 99 f 3 9 9 f 2 99.0 f 0.5 98 f 1

105 f 2 103 f 2

93 f 4 46 f 3 6 1 f 2

101 f 4 92 f 3 99 f 3 99 f 4 97 f 5

101 f 3 94 f 2

103 f 1 99 f 4

Conditions: precolumn packing, Hypersil ODs-C18, 30 wm; flushing eluent, water; duration of flushing, 1 min; concentration of drugs, 10 pg/mL.

their spectra. With the proposed procedure, the selectivity is better than the previously observed for liquid-liquid extraction using ethyl acetate, and comparable to that obtained when octadecyl silica solid-phase extraction cartridges are used.3

Recoveries and Precision. Table 1 summarizes the overall recoveries obtained for aqueous solutions of the compounds tested. It can be observed that the recoveries for most of the compounds tested are ca. 100%. Acetazolamide and hydro- chlorothiazide are not completely recovered, as has been previously indicated. However, the efficiency of the proposed method is comparable to that found for solid-phase extraction techniques using a similar packing. The recoveries obtained for spiked urine samples are also showed in Table 1. As can

246 AnalyticalChemistry, Vol. 66, No. 2, January 15, 1994

Table 2. LhHs of Detection Obtalned at 290 nm for Diuretics and Probonedd wlth the Propowd Procedure

limit of limit of detection detection

compound (ng/mL) compound (ng/mL) amiloride 20 cyclothiazide 20 acetazolamide 40 bendroflumethiazide 20 hydrochlorothiazide 7 ethacrynic acid 200 triamterene 7 bumetanide 4 chlorthalidone 4 probenecid 200 furosemide 2 spironolactone 20

1400:

1200:

3 I,,,!

E e007

600

be observed, these values are similar to those found in aqueous solutions, then indicating that the matrix of the sample does not affect the recoveries of diuretics or probenecid.

The precision is excellent, with coefficients of variation below 6% in spiked urine samples. The greatest imprecision is obtained for acetazolamide. This can be attributed to the switching step, as in our case it is manually effected. However, the reproducibility is greatly improved in comparison to liquid- liquid or solid-phase extraction on disposable cartridge technique^,^ as no off-line operation is involved. No differences in the precision were observed when an internal standard was used. Therefore, the addition of a standard is not necessary, then avoiding sample dilution.

Sensitivity. A back-flush mode has been preferred to minimize thedispersion of the sample into the chromatographic system. In such a way, no peak broadening was observed in comparison with direct injection of the samples into the analytical column.

In Table 2 are summarized the limits of detection found under the proposed conditions. The lowest limits of detection for screening tests of diuretics in doping control have been obtained with gas chromatography" or liquid chromatogra- phy12 coupled with mass spectrometric detection procedures, with reported limits of detection for most of the investigated compounds in the 2.0-0.04 pg/mL range. These values are comparable to those observed in Table 2. The sensitivity reached under the described conditions is also similar to that previously found using C 18 solid-phase extraction cartridges for the sample cleanup,1° but in such a prooedure, a greater volume of sample (2.0 mL) was required. The values listed in Table 2 can be decreased if volumes of samples greater than 50 pL are injected. However, urinary endogenous compounds are also concentrated. As a result, the enrichment factor of the analytes is limited, especially for diuretics eluted at short retention times, where most of the matrix components are also eluted. Moreover, the injection of grater volumes of samples decreases the operative life of the precolumn.

Utility. Although urine samples are directly injected, more than 50 injections can be made without increasing the back- pressure, neither in the precolumn nor in the analytical column.

In the described system, the limiting step is the time required for chromatographic analysis (approximately 1 1 min after the injection of the sample). The precolumn and the analytical column are effectively regenerated and reequilibrated in a few minutes, and baseline drifts are not observed (see Figure 4). The total analysis time is ca. 15 min, which makes it

(1 1) Lisi, A. M.; Trout, G. J., Kazlauskas, R. J . Chromarogr. 1991, 563, 257. (12) Ventura, R.; Fraisse, D.; Becchi, M.; Paisse, 0.; Segura, J. J . Chromarogr.

1991, 562, 723.

3 E

b 160

1 140j

b

140

1204

100

6 0 L 1 0 M "

20 401

800-j

WI /I

800

600

4 00

200

D

d P

0 A 2 4 6

D

Tlme ( m l n . )

Figure 5. Chromatograms at 230 nm of positive findings of chiorthaiidone (a), ethacrynic acid (b), and probenecid (c) obtained 120, 6.5, and 30 h after dosing, respectively. Dose administered: probenecid, 500 mg; ethacrynic acid, 25 mg; chiorthaiidone, 25 mg.

useful for doping control tests, where large series of samples must be processed.

The studied compounds show a wide variety of pharma- cokinetic properties. For example, the period of activity for ethacrynic acid is about 6-8 h, whereas chlorthalidone is effective for several days after dosing.13 However, under the described conditions, the investigated compounds are detect- able within their respective periods of activity after a minimum single dose administration.10J3

On the other hand, most of thediuretics tested are primarily excreted in urine as unchanged drugs.I3 Only spironolactone ( f R = 10.8 min) is rapidly converted to the metabolite canrenone, which is the major circulating form of the drug.14 However, we have found that canrenone is eluted at a retention time slightly lower than that of spironolactone; thus, canrenone does not interfere with the identification of hydrochlorothiazide ( t ~ = 6.3 min) or chlorthalidone ( t ~ = 7.9 min), which are the diuretics most frequently used in combination with spironolactone.

(13) Rimbau, V. Offarm 1991, 10, 57. (14) Sherry, J. H.; ODonnell, J . P.; Colby, H. D. J . Chromatogr. 1986,374, 183.

247 Analytical Chemistty, Vol. 66, No. 2, January 15, 1994

A validation of the usefulness of this procedure was accomplished by analyzing real urine samples obtained after a minimum single dose administration of chlorthalidone, ethacrynic acid, and probenecid. The detection of these compounds can be achieved for time intervals comparable to their periods of activity." Figure 5 shows positive findings of chlorthalidone (a), ethacrynic acid (b), and probenecid (c) obtained from samples collected a t 120, 6.5, and 30 h after drug administration, respectively. The presence of these compounds can be easily confirmed by comparing the UV spectra of the samples and the standards.

CONCLUSIONS The screening of diuretics and probenecid in urine samples

is greatly simplified with the described system. The off-line multistep sample pretreatments usually required for screening

tests are avoided, as untreated urine samples are directly injected onto the chromatographic system. The efficiency of the cleanup process is satisfactory, with almost complete recoveries for most of the compounds tested, the precision being excellent. The method is selective and sensitive enough to detect diuretics and probenecid in real urine samples. In addition, the proposed column-switching system leads to a considerable saving of both chemicals and time.

ACKNOWLEDGMENT

received from the realization of Project SAF 92-0655. The authors are grateful to the CICyT for financial support

Received for review June 4, 1993. Accepted September 28, 1993."

e Abstract published in Aduance ACS Absrracts. November 1, 1993.

248 Analytical Chemistry, Vol. 66, No. 2, January 15, 1994