CLIN. CHEM.

29/2, 256-259 (1983)

Creatinine Determined by High-Performance Liquid Chromatography

R. T. Ambrose, D. F. Ketchum, and J. W. Smith
We describe

a high-performance liquid-chromatographic

of creatinine in serum. Creatinine is separated from other species by cation-exchange chromatography. Absorbance of the creatinine in the eluate

procedure for determination

is monitored at 234 nm. Results from this procedure on sera correspond closely to those of a manual Jaff#{233}-reaction procedure involving use of a protein-free filtrate with additional cleanup of the sample with Lloyds reagent. None of the 21 endogenous and exogenous substances that we screened interfered. The overall CV at 138 ,umol of creatinine per liter is 1.35% and at 539 moI/L is 0.72%. Additional Keyphrases: chromatography,
economics of laboratory operation cation-exchange

the 21 interferents that we screened. The results for serum assays by this technique were extremely precise, with coefficients of variation from 0.7 to 1.4%, and were in excellent agreement with results obtained by a carefully performed Jaffe/Lloyds reagent procedure. We constructed a semiautomated instrument dedicated to the creatinine determination, in which a simple step-gradient pumping system is used that is much less expensive than conventional gradient equipment for HPLC.

Materials and Methods

Apparatus
We used Eppendorf pipets to take samples of serum specimens and to add trichloroacetic acid to the samples. In the semiautomated HPLC instrument we constructed (Figure 1), the mobile phases are metered by a single-head LDC Model 396 minipump (Milton Roy Co., Riviera Beach, FL 33404). The flow rate is 1.20 mUmin. The flow of mobile phase is smoothed by connecting the pump output to the sampling valve via a restrictor coil (PN25561) and by connecting a pulse dampener (PN98060) (both from Waters Associates, Milford, MA 01757) just before this coil, with a tee joint. The free end of the pulse dampener is fitted with a simple release valve (15MW-A200; Scientific Systems Inc., State College, PA 16801).A Model 7010 valve (Rheodyne, Berkeley, CA 94710) equipped with a 70-01 pneumatic activator and a 50-FL loop is used for sample injection. The mobile phases are selected with a Teflon three-way, 115-V ac solenoid valve (Isolatch; General Valve Corp., East Hanover, NJ 07936) mounted on the pump frame directly below the inlet, with General Valve fittings and a minimum length of Teflon tubing. The instrument is controlled by a four-interval digital valve sequence programmer (Valco Instruments Co., Houston, TX 77055); this is a general-purpose programmable sequencer with associated relays that may be wired by the user to control external instrumentation. Here we used it to control the mobile-phase valve and the sampling-valve

Measurement of creatinine in human serum is useful as a measure of renal function. Traditionally the Jaff#{233} reaction (1, 2) is used on a protein-free filtrate to form a creatininepicrate adduct, the absorbance of the resulting color being measured at 500 nm. This method is not completely specific because several substances that could be in the protein-free filtrate form colored species with the alkaline picrate reagent-e.g., acetone, acetoacetic acid, L-dopa, and cephalothin. The use of a cleanup procedure (3), such as adsorption of the creatinine in a protein-free filtrate onto Lloyds reagent (an aluminum silicate) followed by desorption in alkaline solition, does much to remove interferences, but the procedure is time-consuming and some interferences are not entirely removed (4). Some workers have used cation-exchange chromatography to separate creatinine from potential interferents (5-9). High-performance liquid chromatography (IIPLC) in a reversed-phase mode has been used to separate and determine creatinine in serum (10-13). In the HPLC method we describe here, cation-exchange chromatography is used with elution at neutral pH, which permits measurement at 234 nm. A pH (4.68) below the pKa (5.02) of creatinine is used in the separation, to make the

moleculesufficiently cationic tobe retained by an appropriate cation-exchange resin.

CH pK
=

Reservoirs

5.02

H C-<

3
.

The pH is then increased to 7.1 to elute the neutral creatinine molecule. This mode of separation eliminates as interferents molecules that are uncharged or negatively charged at pH 4.68, because they are not retained by the resin. The pH programming from 4.68 to 7.10 and the wavelength (234 nm) used for detection provide additional discrimination against interferents. Ion-exchange chromatography separated creatinine from

Milton Roy Co. Single Piston Minipump

Research Laboratories, Eastman Kodak Co., Rochester, 14650. Received Aug. 29, 1980; accepted Oct. 21, 1982.
256

NY
Fig. 1. Creatinine HPLC flow chart

CLINICAL CHEMISTRY, Vol. 29, No. 2, 1983

pneumatic activator. The latter is equipped with pneumatic solenoids and a mounting block, also supplied by Valco, and is driven by 550-kPa pressure of compressed air. The digital valve sequence programmer was wired and programmed for the following sequence: Sequence idle 1 2 3
4

Duration, mm
-

1.5 2.0 3.0 unused

pH of mobile phase 4.68 4.68 7.10 4.68

Sampling value load inect inject load

aqueous trichioroacetic acid to cover the range 9 to 663 umol/L. The pH 4.68 lithium acetate eluent A was prepared by adding 2.04 g of lithium acetate . 2H20 to 900 mL of distilled water; then 45 mL of 1 mol/L acetic acid was added. The pH was adjusted with 1 mol/L acetic acid (monitored with a pH meter) to 4.68, and the solution was diluted to 1 L with distilled water. The pH 7.10 eluent B was prepared by adding 30 mL of 0.1 mol/L acetic acid to 7.65 g of lithium acetate . 2H20 in 900 mL of H20. The pH was adjusted to 7.10 with 0.10 mollL acetic acid and the solution diluted to 1 L as before. The eluents were degassed pressure in a vacuum flask, while being magnetic stirrer. under agitated reduced with a

In this sequence, the sample is injected and the protein-free filtrate is passed through the system. The high-pH mobile phase is than injected, which elutes the creatinine. Finally, the low-pH mobile phase flows through the system and the sampling valve is opened, so that at the end of the sequence the column is fully equilibrated and ready for the next sample. We used three different detectors during this work: (a) the LC-55 (Perkin-Elmer Corp., Oak Brook, IL 60521), (b) the LIM (Gilson Medical Electronics, Inc., Middleton, WI 53562), and (c) the Perkin-Elmer LC-75. The small differences observed in sensitivity, signal-to-noise ratio, and linearity of response were negligible for this application. An A-25 linear strip-chart recorder (Varian Associates, Walnut Creek, CA 94598) was used with the detector, and quantification was by measuring peak height. We used the Gilson HM detector for most of this work.

Procedure
Trichloroacetic acid (1.0 mL, 100 g/L) was added to 1.0 mL of human serum calibrator or aqueous standard. The sample was vigorously agitated on a vortex-type mixer and centrifuged at 3000 rpm for 10 mm. The supernate was decanted, and an aliquot was injected into the liquid chromatograph, where the mobile phases for the step-gradient analysis were as described in Apparatus. This sequence maximized sample throughput without compromising the separation of creatinine from potentially interfering components. A calibration curve of peak height vs creatinine concentration was used to quantify the unknown samples. To study precision, we used Versatol normal, Versatol abnormal, and Versatol abnormal-alternate calibrator fluids (General Diagnostics, Morris Plains, NJ 07950). Aliquota of reconstituted and pooled vials of each solution were stored frozen in screw-cap vials. Two 1.0-mL samples from each aliquot were analyzed as described, after thawing; results are the mean determined from two chromatograms for each sample. We determined the analytical recovery of creatinine as follows: Two creatinine solutions (22.1 and 2.21 mmol/L) were prepared in water. Six sample poois (three duplicates) were prepared by pipetting 1 mL of water or 1 mL of one of the creatinine solutions into 25-mL volumetric flasks; and diluting to volume with pooled serum obtained from the Kodak Park Medical Department. Five 1.0-mL aliquots of each of the six serum samples were precipitated with 1 mL

The 50 cm x 2.1mm stainless-steel HPLC columns were packed with a strongcation-exchange material-either Zipax (30 JLm; Du Pont Co., Wilmington, DE 19898), Partisil-lO (SCX (10 tm; Whatman Inc., Clifton, NJ 07014), or Bondapak CX Corasil (35-75 tim; Waters Associates). The small differences in peak shape and retention time that resulted from the use of these different columns were not meaningful for this application, because the creatinine peak was well resolved from other sample components. The columns are stable; after more than 300 samples had been applied to a single Zipax column, there was essentially no change in peak height or retention time. The studies reported here involved use of Zipax columns prepacked by the supplier.

Reagents
All The either icals. chemicals used in the procedure were reagent grade. chemicals screened in the interference study were reagent grade or white-label Kodak laboratory chemCreatinine standard solutions were prepared in 50 g/L

67.5mm

68mm

675mm

I6ZSmm

J
b
-

Fig. 2. Effect of pH and molarity of lithium acetate solution on retention time and baseline stability in determinations of creatinine, 88 moI/L. (a) EluentA 20 mmol/L, pH 4.6: eluent B 80 mmoVL, pH 7.1. (b) Eluent A 10 mmol/L, pH 4.4;eluent B 10 mmol/L. pH 7.2. (C) Eluent A 20 mmol/L, pH 4.68; eluentB 75 mmot/L, pH 7.1

Fig. 3. Creatinine peak-height reproducibility, indicated heights of repeated injections of same specimen

by measured

CLINICAL CHEMISTRY, Vol. 29, No. 2,

1983

257

Table 1. Creatinine Precision Study

Within-day precision Creatinine, mol/L Overall precision SO, imoIIL
1.9 CV, % 1.35

Sample
Normal

n
64

Mean
138

Range
133-141 355-364

SD, zmoi/L
1.7

CV, %
1.23

Abnormal Abnormal Alternate

68 68

360 539

1.6
1.7

0.44
0.31

2.3
3.9

0.64
0.72

532-548

aVersatol calibrat ors (General Diagnostics).

of 100 g/L trichloroacetic acid. Results for a single injection of each aliquot into the liquid chromatograph were compared with those for creatinine standards prepared in 50 gIL
trichloroacetic acid.

Results and Discussion

Development of the Method

At the initial pH of 4.68, creatinine is cationic and is by the cation-exchange resin. Uric acid (p1 5.7) is not cationic at this pH, is not retained, and is eluted at the solvent front. The detection of creatinine is most sensitive at neutral pH (molar absorptivity = 6471 L cm mol at pH 7.1), because it is the uncharged molecule that shows an absorption maximum at 234 nm.
retained The importance of optimizing eluent strengths is seen in Figure 2. We chose acetate as the eluent anion because it is a good buffer at pH 4.68. The control of the pH of eluent A is very important. The amount of cationic character and therefore the retention time for creatinine are governed by

The pH and concentration of eluent B must be high enough to elute the creatinine from the columxi and to present the neutral creatinine molecule to the detector. At pH 7.1, less than 1% of the creatinine molecules are protonated, and the sensitivity is therefore maximized. The concentration of the buffer is such as to minimize baseline shift during the chromatographic run. Chromatogram C in Figure 2 has minimal baseline distortion. The concentration of eluent B must be high compared with that of eluent A because lithium acetate has essentially no buffer capacity at pH 7.1. Figure 3 shows the peak-height reproducibility of chromatograms obtained for repeated injections of 44 p.mol/L solution of creatinine. Analytical
Precision

Variables

this pH. The retention time in Figure 2 decreases from 7.1 mm at an initial pH of 4.4, to 5.9 mm at pH 4.6, to 3.2 mm at pH 4.68. The desired retention time is the minimum time
that will resolve creatinine from materials at the solvent front and from other interfering substances; minimizing retention time increases peak sharpness (improving the precision of measurements of peak heights) and yields maximum sample throughput.
I I

and accuracy. The precision of the method was over a 66-day period. Pools with three different concentrations were prepared from commercially calibrator materials and kept frozen until analyzed. Duplicate samples were taken and analyzed on 20 days during this 66-day period. Results, summarized in Table 1, agree well with those reported by other techniques

measured creatinine available

(4).

0.60

The standard curve for the procedure, prepared from 50tL injections of the aqueous standards, is linear up to a concentration of at least 0.66 mmol/L: y (peak height) = 346x + 2.96 (r2 = 0.998). Calibration is done daily to ensure maximum precision. The sensitivity of the method is more than adequate for analyzing serum creatinine. A detection limit of 0.01 mol/L

0.50
-J
0

Table 2. HPLC Interference

Substances tested

Screen
Concn tested,
mg/L

E 0,40 E
C C

Acetylaminophenol Acetylsalicylic acid

p-Aminosalicylic acid

0.30

Ascorbic acid
Bilirubin

50 300 230 20
200
1000

0 C

Cephalothin 0.20
0

Chlorothiazidea
0

30
15

Creatine L-Dopaa
Gentisic acid

80 50
3000

Glucose
Hemoglobin Isoniazid
I I I

0.10

2000

4
20

Meprobamate 6-Mercaptopurine
0.50 0.60 Phenytoin Sulfathiazolea

Th

0,10 Joff

0.20

0.30

0.40 mmol/L

15 20
60 220 1500 1200

creotinirie,

Tolbutamide
=

Fig. 4. Comparison of HPLC and Jaff#{233} determinations of creatinine: y 1.0040 (SE = 0.0153) x - 0.00017 (SE = 0.0030),

Triglycerides Urea

Uric acid

70

S,,,= 0.0128
258 CLINICAL CHEMISTRY, Vol. 29, No. 2, 1983

Substances known to intertere in the Jaffe/Lloyds reagent procedure.

the chromatographic procedure, none affected the quantification of the creatinine peak. Figure 5 shows the noninterference of (e.g.) p-aminosalicylic acid on the determination of creatinine. We thank Dr. D. W. Grisley of the Health, Safety, and Human Factors Laboratory, Eastman Kodak Co., for his work with the Jaffe/Lloyds reagent procedure. We also thank Dr. D. N. Baxter of the Analytical Sciences Division, Kodak Research Laboratories, for his help in the design and construction of the equipment.

References
1. Jaffe M. Ueber der Niederschlag, welchen Pikriksaure in normalen Ham erzeugt und uber eine neue Reaktion des Kreatinins. Z Physiol Chem 10, 391-400 (1886). 2. Folin 0, Wu H. A system of blood analysis. J Biol Chem 38, 81110 (1919). 3. Henry RJ. Clinical Chemistry: Principles and Technics, Harper and Row, New York, NY, 1974, pp 543-550. 4. Grisley DW, Ambrose RT, Ketchum DF, et al. Investigation of reference methodology for creatinine. Clin Chem 25, 1064 (1979). 5. Weatherburn MW, Trotman EBB, Jackson SH. Specific method for serum creatinine determination based on ion exchange chromatography and an automated alkaline picrate reaction-a proposed reference method. Clin Biochem 11, 159-166 (1978). 6. Adams WS, Davis FW, Hansen LE. New method for deterrnination of creatinine in urine by ion exchange separation and ultraviolet spectrophotometry. Anal Chem 34, 854-856 (1962). 7. Adams WS, Davis FW, Hansen LE. Determination of serum creatinine by ion exchange chromatography and ultraviolet spectrophotometry. Anal Chem 36, 2209-2211 (1964). 8. Chiou WL, Gadalla MAF, Peng GW. Simple, rapid, and micro high-pressure liquid chromatographic determination of endogenous true creatinine in plasma, serum, and urine. JPharm Sci 67, 182187 (1978). 9. Chiou WL, Peng GW, Gadalla MAF, Nerenberg ST. Comparison of plasma creatinine levels in patients determined by high-pressure liquid chromatography, automated analysis, and boiling alkaline picrate method. J Pharm Sci 67, 292-293 (1978). 10. Brown ND, Sing HC, Nelley WE, Koetitz SE. Determination of true serum creatinine by high-performance liquid chromatography combined with a continuous-flow microanalyzer. Clin Chem 23, 1281-1283 (1977). 11. Lim CK, Richmond W, Robinson DP, Brown SS. Towards a definitive assay of creatinine in serum and urine: Separation by high-performance liquid chromatography. J Chromatogr 145, 4149 (1978). 12. Soldin SJ, Hill JG. Micromethod for determination of creatinine in biological fluids by high-performance liquid chromatography. Clin Chem 24, 747-750 (1978). 13. Spierto FW, MacNeil ML, Culbreth P, et a!. Development and validation of a liquid-chromatographic procedure for serum creatinine. Clin Chem 26, 286-290 (1980).

C)
C

Fig. 5. Chromatogram of creatinine (140 mol/L) acid (230 mg/L)

and p-aminosalicylic

is easily obtained if a 500-L sample is chromatographed. Analytical recovery of creatinine added to four human serum pools, measured in two identical experiments, ranged from 99.3 to 100.8%, i.e., was essentially complete, CVs for these experiments ranged from 0.48 to 1.04% for the 2.21 mmollL creatinine addition and from 0.30 to 0.95% for the 22.1 mmolIL creatinine addition (for both, n = 5 each). Using 80 human-serum samples, we compared the HPLC method with a manual Jaff#{233} procedure in which Lloyds reagent was used as a cleanup for the protein-free filtrate. The regression line for this comparison shows the agreement between these two procedures (Figure 4). Interfere nces. It is unlikely that many substances would interfere in this procedure because of the combined selectivity of the mobile-phase pH, the cation-exchange column stationary phase, and the 234-nm detection wavelength used. Materials that are uncharged or anionic at the initial pH of 4.68 would not be retained by the separating column and would elute in the void volume. The 21 endogenous and exogenous materials we tested (Table 2) did not interfere. Although several of the drugs (paminosalicylic acid, chlorothiazide, L-dopa, sulfathiazole, tolbutamide, and cephalothin) gave observable responses in

CLINICAL CHEMISTRY, Vol. 29, No. 2, 1983

259