324
Anal. Ch8m. 1987, 59, 324-327
Liquid Chromatographic Assay of ,8-Lactarnase Inhibitors in Human Serum and Urine Using a Hollow-Fiber Postcolumn Reactor Jun Haginaka,* Junko Wakai, and Hiroyuki Yasuda Faculty of Pharmaceutical Sciences, Mukogawa Women’s University, 4-16 Edagawa-cho, Nishinomiya, Hyogo 663, J a p a n
A rapid, sensitive reversed-phase high-performance llquld chromatographk method using a hoHow-ftber postcolumn reactor Is described for the determinatlon of @-iactamaseinhlbttors (ciavulank ackl and sulbactam) in human serum and urine. After uHraflHratlon of serum samples and filtration of urine samples, an aliquot of the filtrates was accurately loaded onto a C,, column. Then the eluent was introduced Into a sulfonated hollow-fber postcolumn reactor suspended In a sodium hydroxide sohrtlon. Detectlon was based on the ultraviolet absorbance of the degradation product(8 ) of clavuianlc acid at 272 nm and sulbactam at 278 nm. At clavulank acld and subactam concentratlons of 0.5 pg/mL in serum samples, within- and between-run precisions (relative standard devlatlon) were 1.26-2.86% and 2.00-3.04%, respectlvely. The lower ilmHs of accurate determination of ciavulank acid and sulbactam were as low as 25 ng/mL for plasma samples and 50 ng/mL for 10-fold diluted urlne samples with a 20-pL injectlon.
The lack of suitable detectors in high-performance liquid chromatography (HPLC) for trace and ultratrace analysis in complex matrices prompted the development of postcolumn derivatization ( I ) . However, the conventional postcolumn derivatization technique requires an additional pump(s), mixing tee, and reaction coil. They lead to band broadening, dilution, and noisy and drifting base lines (which reduce the sensitivity gained through derivatization). Hollow-fiber membranes have been used for ion chromatography as “suppressors” (2-4). Recently, they have been applied to the reactor for HPLC (5) and flow injection analysis (6). The hollow-fiber membrane reeactor can eliminate or control the above-mentioned problems associated with the conventional reactor design. In previous papers (7-9), we reported that clavulanic acid and sulbactam (which are /3-lactamaae inhibitors being used for clinical chemotherapy) are rapidly degraded in alkaline methanolic solution to yield products having ultraviolet (UV) absorption maxima at around 270-280 nm, and we developed HPLC methods for determining these p-lactamase inhibitors in human serum and urine using the above reactions for detection. This paper deals with the HPLC assays of clavulanic acid and sulbactam in serum and urine obtained by using a hollow-fiber postcolumn reactor.
EXPERIMENTAL SECTION Reagents and Materials. Potassium clavulanate and sodium sulbactam were kindly donated from Beecham Yakuhin Co., LM. (Tokyo, Japan), and Pfizer-Taito Co., Ltd. (Tokyo, Japan). Tetra-n-butylammonium bromide (TBAB) and other chemicals of reagent grade were purchased from Nakarai Chemicals, Co. (Kyoto, Japan). Deionized, glass-distilled water and glass-distilled methanol were used for the preparation of HPLC eluents.
Sulfonated hollow-fiber membrane (AFS-2 fiber) was purchased from Dionex Co. (Sunnyvale, CA). Chromatography. The experimental setup used in this study was illustrated in Figure 1. The following HPLC system was used: an LC-5A pump (Shimadzu Co., Kyoto, Japan) for delivering the eluent; a Model 7125 loop injector (Rheodyne, Cotati, CA) equipped with a 100-pL loop for the loading of the samples; and an SPD-6AV (Shimadzu Co., Kyoto, Japan) spectrophotometer equipped with an 8-pL flow-through cell for detection. The reversed-phase C18columns (15 cm x 4.6 mm id.) used were as follows: column I, Nucleosil 5C18(5 pm) (Macherey-Nagel,Diiren, FRG) for serum samples; column 11, Develosil ODS-5 (5 pm) (Nomura Chemicals, Seto, Aichi, Japan) for urine samples. A guard column (3 cm X 4.6 mm i.d.) packed with the same packing material was used to protect the main column. The eluents used were as follows: eluent A, 15 mM TABA-30 mM NaH2P04-30 mM NazHP04-methanol (l:l:l:l, v/v) for the assay of clavulanic acid in serum; eluent B, 15 mM TBAB-3 mM NaH2P04-3 mM Na2HP04-methanol (1:1:1:1.2, v/v) for clavulanic acid in urine; eluent C, 15 mM TBAB-15 mM NaHzP04-15 mM Na2HPO4methanol (1:1:1:1.2, v/v) for sulbactam in serum; eluent D, 15 mM TBAB-1.5 mM NdzPO4-1.5 mM NazHP04-methanol (1:1:1:1.5, v/v) for sulbactam in urine. The detection was performed at 272 nm for clavulanic acid and 278 nm for sulbactam. A hollow-fiber membrane (0.3 mm i.d. X 1.2 m) was inserted in a PTFE tube (0.86 mm i.d. and 1.46 mm 0.d. X ca. 2 cm) and was attached to suitable connecting fittings. The hollow-fiber membrane suspended in a sodium hydroxide solution in a 50-mL beaker was inserted between the column and the detector. The concentrations of sodium hydroxide solution were 1.0 M for clavulanic acid and 3.0 M for sulbactam. No pressure was applied to the hollow-fiber membrane reactor. All separations and postcolumn reactions were performed at ambient temperature. Comparison of Detection Method. The peak broadening due to the postcolumn reactor design was estimated by the following methods: method A, detection at 230 nm without a postcolumn reactor; method B, detection at 272 nm with an open-tubular postcolumn reactor; method C, detection at 272 nm with a hollow-fiber postcolumn reactor. For method B, the following reaction devices were used: a double plunger pump (NP-DX-2, Nihon Seimitsu Kagaku, Tokyo, Japan) for delivering the postcolumn reagent (0.5 M sodium hydroxide solution) at a flow rate of 0.2 mL/min; a mixing tee; and the reaction coil, made of a 0.5 mm i.d. X 1 m PTFE tube, for reactor. For method C, the hollow-fiber postcolumn reactor (0.3 mm i.d. X 1.2 m) suspended in a 1.0 M sodium hydroxide solution was used. A 10-pL portion of clavulanic acid solution (5 pg/mL) was introduced into the chromatograph under the conditions described above (column I and eluent A). The dispersion):g( and the number of theoretical plates (N) of each peak were calculated. Sample Preparations. For serum samples, a clavulanic acid or sulbactam standard was dissolved in human control serum. The serum samples were ultrafiltered by using an MPS-1 micropartition system (Amicon,Tokyo, Japan) with YMT membrane at 1500g for 10 min. A 20-pL portion of the serum ultrafiltrate was accurately loaded onto an HPLC column. For urine samples, a clavulanic acid or sulbactam standard was dissolved in human control urine. The urine samples were diluted 10-foldwith water and filtered with a 0.45-pm acrylate-copolymer membrane (Gelman Science Japan, Tokyo, Japan). A 20-pL portion of the
0003-2700/87/0359-0324$01.50/00 1987 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 59, NO. 2, JANUARY 1987
325
DETECTOR
INJECTOR
e
WRSTE
COLUMN PUMP ELUENT
HOLLOW-FIBER
RERCTOR
RERGENT
Flgure 1. Experimental setup for the determination of P-lactamase 8-
-E,
* c
2
4-
c Y 0
n
__I_.
0
I 1
2
3
4
Concontratlon of NaOH ( M I
Figure 3. Effect of the concentration of sodium hydroxide and the length of a hollow-fiber membrane reactor on the peak height of the reaction product(s) formed from sulbactam. Hollow-fiber reactor lengths were as follows: 1, 60 cm; 2, 1 m; 3, 1.2 m. Detection was at 278 nm. Sensitivity was 0.064 aufs. 0
1 2 3 Concontratlon of NaOH (M)
4
Table I. Band Broadening in the Postcolumn Reactor method
Flgure 2. Effect of the concentration of sodium hydroxide and the length of a hollow-fiber membrane reactor on the peak height of the reaction product(s) formed from clavulanic acid. Hollow-fiber reactor lengths were as follows: 1, 60 cm; 2, 1 m; 3, 1.2 m. Detection was at 272 nm. Sensitivity was 0.064 aufs.
A B C
ut:
s2
23.8 45.6 29.2
Nb 3940 2350 3390
"The total dispersion in the system. bThe number of theoretical
filtrate was loaded onto the column.
plates.
RESULTS AND DISCUSSION Reaction Conditions for Hollow-Fiber Postcolumn Reactor. In a previous communication (IO),we reported an HPLC method for the determination of clavulanic acid in serum and urine using a hollow-fiber postcolumn reactor. In this report, the postcolumn reaction conditions for clavulanic acid and sulbactam were precisely examined by changing the concentration of sodium hydroxide solution and the length of the hollow-fiber membrane reactor. Eluents B and D were delivered at 0.8 mL/min for clavulanic acid and sulbactam, respectively. A 10-pL portion of a clavulanic acid or sulbactam standard solution (10 pg/mL) was loaded onto the column (column 11),and the peak height was measured. Figures 2 and 3 show the effects of the concentration of sodium hydroxide and the length of the hollow-fiber membrane reactor on the peak height of the degradation products formed from clavulanic acid and sulbactam, respectively. At a hollow-fiber reactor length of 60 cm, for clavulanic acid the constant peak height was obtained at a sodium hydroxide concentration of 2 M or more, while for sulbactam the increase of the peak height was observed as the concentration of sodium hydroxide was raised. A t hollow-fiber reactor lengths of 1and 1.2 m, the maximum and constant peak height (which was 1.5-times higher than that obtained at a hollow-fiber reactor length of 60 cm) was obtained at a sodium hydroxide concentration of 1 M or more for clavulanic acid and 3 M or more for sulbactam. Thus, the postcolumn reaction conditions were selected for the assays of clavulanic acid a t a hollow-fiber reactor length of 1.2m and a sodium hydroxide concentration of 1 M; sulbactam reaction conditions were 1.2 m and 3 M. As reported previously (7,8), clavulanic acid and sulbactam can be transformed to methyl 8-hydroxy-6-0~0-4-aza-2-octenoate and methyl 5-carboxy-6-methyl-6-sulfino-4-aza-2heptenoate and detected at 272 and 278 nm, respectively.
Comparison of Detection Method. The band broadening of the three detection methods (methods A, B, and C) was compared with respect to the dispersion):a( and the number of theoretical plates (N). Table I shows the results of u t and N for the three detection methods. The increases in the dispersion due to the open-tubular and hollow-fiber postcolumn reactors were 21.8 and 5.4 s2, respectively. The calculated number of theoretical plates indicated 40% and 14% loss of resolution due to the open-tubular and hollow-fiber postcolumn reactors, respectively. Figure 4 shows the comparison of the three detection methods for clavulanic acid in serum samples: parts A, B, and C are corresponding to detection methods A, B, and C, respectively. In method A, clavulanic acid (retention time, 5.2 min) completely overlapped with the background component of serum and was observed only at peak height of 0.3 cm on a chromatogram. These results reveal that selective and sensitive detection of clavulanic acid is not attained at 230 nm. In methods B and C, clavulanic acid was selectively and sensitively detected at 272 nm following the postcolumn alkaline degradation reaction: in method B, clavulanic acid overlapped with one of the background components, while in method C, clavulanic acid was resolved from the component (which is observed on a chromatogram as a shoulder peak). In addition, the peak height of clavulanic acid in method C was 1.5 times higher than that in method B. As described above, these are due to the fact that the band broadening of the hollow-fiber postcolumn reactor is much smaller than that of the conventional postcolumn reactor. Separation of Clavulanic Acid and Sulbactam from Serum and Urine. Under the optimum postcolumn reactor conditions, the separation of clavulanic acid or sulbactam from the normal components of serum and urine was examined by
326
ANALYTICAL CHEMISTRY, VOL. 59, NO. 2,JANUARY 1987 B
A
A
C
- - i
0
5
10
Tlme (mlnl
0
5
10
Time (mln)
0
- -
5 10 Time (mln)
Figure 4. Comparison of the three detection methods for clavulanic acid in serum samples: A, detection at 230 nm without a postcolumn reactor; B, detection at 272 nm wlth an open-tubular postcolumn reactor: C, detection at 272 nm with a holbw-fiber postcolumn reactor. I n parts B and C, dotted lines Indicate the serum ultrafiltrate backgrounds, but in part A clavulanic acid completely overlapped with the serum ultrafiltrate background and was observed only at a peak height of 0.3 cm. Sensitivity was 0.016 aufs.
0
clavulanic acid
sulbactam
l b
26 36
0.49 f 2.04% 0.52 f 2.86% 0.50 f 1.29%
0.51 f 1.87% 0.50 f 1.26% 0.49 f 1.95%
between-runc
0.50 f 3.04%
0.50 f 2.00%
5 10 T i m e (min)
0
5
10
T l m e (mln)
Figure 5. Chromatograms of clavulanic acM (1) in 10-fold diluted urine
(A) and serum ultraflltrate (e). The concentration of clavulanic acid was 0.5 gg/rnL in 10-fold diluted wine and neat serum samples. Dotted lines indicate the backgrounds. Sensitivty was 0.016 aufs.
Table 11. Precision of the Assay of Clavulanic Acid and Sulbactam in Serum Samples" assay
L
Y
A
"The concentration of clavulanic acid and sulbactam was 0.50 pg/mL. bMean (pg/mL) f relative standard deviation (RSD) (%) of five replicates. CMean(Kp/mL) f RSD (%) of three replicates. Table 111. Precision of the Assay of Clavulanic Acid and Sulbactam in Urine Samples" assay
clavulanic acid
sulbactam
l b
26 36
4.94 f 1.13% 4.96 f 1.13% 5.02 f 0.91%
5.21 f 2.88% 4.84 f 1.22% 5.00 f 2.63%
between-runc
4.97 f 0.84%
5.02 f 3.70%
- -
"The concentration of clavulanic acid and sulbactam was 5.0 pg/mL. bMean (pg/mL) f RSD (70)of five replicates. cMean (hg/mL) f RSD (%) of three replicates. use of TBAB as an ion-pairing agent, as described previously (7-9). Figure 5 shows the separation of clavulanic acid from the background components of 10-fold diluted urine (part A) and serum ultrafiltrate (part B). Figure 6 shows the separation of sulbactam from the background components of 10-fold diluted urine (part A) and serum ultrafiltrate (part B). Precision and Linearity. Table I1 shows the within- and between-run precisions of the assays of clavulanic acid and sulbactam in serum. Table 111 shows the within- and between-run precisions of the assays of clavulanic acid and sulbactam in urine. The eight-point calibration graphs constructed by peak height vs. concentration for clavulanic acid and sulbactam were linear in the concentration ranges ranging from 0.05 to 10 pg/mL for serum samples and from 5 to 100 gg/mL for neat urine samples and passed through the origin. The lower limits of accurate determination of clavulanic acid
0
5
10
0
Time5 (min) 10
Tlme (min)
Figure 6. Chromatograms of sulbactam (1) in 10-fold diluted urine (A) and serum ultraflitrate (B). The concentration of sulbactam was 0.5 gg/mL in 10-fold diluted urine and neat serum samples. Dotted lines Indicate the backgrounds. Sensitivity was 0.016 aufs.
and sulbactam were as low as 25 ng/mL for plasma samples and 50 ng/mL for 10-fold diluted urine samples with a 20-gL injection. Registry No. NaOH, 1310-73-2; p-lactamase, 9073-60-3; clavulanic acid, 58001-44-8; sulbactam, 68373-14-8.
LITERATURE CITED (1) Frei, R. W. Chemical Derivafizafion in Analytical Chemistry; Frei, R. W., Lawrence, J. F., Eds.; Plenum: New York, 1981: Vol. 1, Chapter 4.
(2)
Stevens, T. S.;Davis,
(3)
Stevens, T. S;Jewett, G. L.; Bredeweg,
1488-1492. 1206-1208.
J. C.; Small, H. Anal. Chem. 1981, 53, R. A. Anal. Chem. 1982, 5 4 ,
Anal. Chem. 1987, 59, 327-333 (4) Hanaoka, Y.; Murayama, T.; Muramoto, S.; Matsuura, T.; Nanba, A. J. ChrOmetOgr. 1982, 239, 537-548. (5) Davis, J. C.; Peterson, D. P. Anal. Chem. 1985, 57, 768-771. (8) Hwang, H.; Dasgupta, P. K. Anal. Chem. 1988, 58, 1521-1524. (7) Haginaka, J.; Yasuda, H.; Uno, T.; Nakagawa, T. Chem. fharm. Bull. 1983, 31, 4436-4447. ( 8 ) Haginaka, J.; Yasuda, H.; Uno, T.; Nakagawa, T. Chem. fharm. Bull. $984, 32, 2752-2758.
327
(9) Haginaka, J.; Wakai, J.; Yasuda, H.; Uno, T.; Nakagawa, T. J. Llq. ChrOMtGgr. 1985, 8 , 2521-2534. (10) Haginaka, J.; Wakai, J.; Yasuda, H. Chem. fharm. Bull. 1986, 3 4 , 1850-1852.
for review
9, 19g6* Accepted September 16,
1986.
Ultrasonic Micronebulizer Interface for High-Performance Liquid Chromatography with Flame Photometric Detection J. F. Karnicky* and L. T. Zitelli Varian Research Center, 611 Hansen Way, Palo Alto, California 94303 Sj. van der Wal* Varian Instrument Group, 2700 Mitchell Drive, Walnut Creek, California 94598
An ultrasonic micronebuilrer Interface for effective nebuliration at 2-20 pL/min liquid was developed to allow the determination of nonvolatlle analytes by mlcrobore HPLC wlth virtually unmodified gas chromatographic detectors. A detectablllty of 50 pg/s phosphorus was obtalned wlth a dynamlc range of 3 decades and acceptable peak broadening using a flame photometrlc detector (FPD) as a detector. Solvent compatlblllty appears mainly determined by the choice of chromatographlc detector. Dual-wavelength operatlon can improve the signal-to-noise ratio by a factor of 5, improve the preclslon from 6 to 2 % , and eliminate base-line shifts. The utility of the micronebulirer-FPD as a selective HPLC detector Is demonstrated in the analysis of phospholipids and sugar phosphates.
Detectability with the flame photometric detector (FPD) (1) in gas chromatography is approximately 1 pg/s phosphorus. With the thermionic specific detector (TSD) (2) detectability is better than 0.1 pg/s phosphorus and 0.1 pg/s nitrogen. These excellent mass detectabilities together with the P and N selectivity make the FPD and TSD attractive detectors for micro-HPLC. Unfortunately, direct introduction of flow rates of even a few microliters of liquid per minute into an unmodified gas chromatographic detector is restricted to volatile mobile phases and solutes (3,4). Even for these volatile mobile phases and solutes differential volatilization of solutes and solvents leads to a narrow range of operating conditions. Too little volatilization (short residence time or low temperature in the direct interface tube) results in a noisy base line when the (first) flame of the detector receives a high load of mobile phase; too much volatilization causes precipitation of less volatile components in the interface tube, noticeable as spiking, and eventually clogging of the tube. An interface for effective nebulization at 2-20 pL/min liquid was therefore developed. Initial attempts to use concentric or cross-flow pneumatic nebulization failed to yield reliable results for truly nonvolatile solutes, i.e., solutes that cannot be analyzed by gas chromatography in an underivatized form, cf. ref 5. Cross-flow nebulization on a sintered disk (6) was not successful below
Table I. Apparatus component
model
manufacturer
conditions
HPLC
5560
Varian
split flow, 2-20
injector rf signal generator
7520 185
Rheodyne
1-pL injection swept from 3.25 to 3.45 MHz
rf amplifier
240L
E.N.I.
power meter
43
Bird
pL/min
(2)
Wavetek
at 10 Hz continuous wave, 7-15-W output
frequency 5381A Counter thermocouple 151-786 temperature controller liquid N2 D-6000 Dewar FPD (1) K-filter S-10-767-F K-PMT R910 gas flowmeters 602/60l
Cole Parmer
modified
Varian Corion Hamamatsu Matheson
dual-beam mode
hydrogen air
liquid air liquid air
105, 175
helium (for iniector)
liquid air
90 psi
Hewlett-Packard Cole Parmer
-114 "C to -70 O C
105-175 mL/min 135 mL/min mL/min
20 pL/min aqueous mobile phase, which precludes its use with an unmodified FPD or TSD. An alternative to developing a low flow rate nebulizing interface would be adaptation of the detector to the standard HPLC flow rates (ca.1mL/min). This is naturally much more costly. it has been applied to the FPD by Julin et al. (7) and Chester (8) and has resulted in detectabilities of more than 1 ng/s phosphorus, which renders it less attractive. The objective of this paper is to report on the nebulizer interface we have developed and its utility in allowing the use of an FPD as a micro-HPLC detector. EXPERIMENTAL SECTION A summary of the main equipment components is given in Table I. A Model 5560 high-pressure liquid chromatograph
0003-2700/87/0359-0327$01.50/00 1987 American Chemical Society
