CLINICAL CHEMISTRY, Vol. 37, No.

4, 1991 563
pulmonary tuberculosis. ActaMed Scand 1921;54:247-51.
9. Embury DH, Doery JCG. #{176}Iina Quant immunoturbidimetric
kit for C-reactive protein adapted to the Cobas Fara centrifugal
analyzer [Tech Brief]. Clin Chem 1988;34:2381.
10. Beldjord K, BenlatracheK, Colonna R. Assessment of iron
stores in inflammation by assay of serum ferritin concentrations.
Br Med J 1982;284:511.
11. Baynes R, BezwodaW, Bothwell T, Kahn Q, Mansoor N. The
non-immune inflammatory response: serial changes in plasma
iron, iron-binding capacity, lactoferritin, ferritin and C-reactive
protein. Scand J Clin Lab Invest 1986;46:695-704.
12. Krause JR, Stolc V. Serum ferritin and bone marrow iron
stores. Am J Clin Pathol 1979;72:817-20.
13. Milman N, Bangsboll S, Strandberg Pedersen N, Visfeldt J.
CLIN. CHEM. 37/4, 563-565 (1991)
Serum ferritin in non-dialysis patients with chronic renal failure:
relation to bone marrow iron stores. Scand J Haematol
1983;30:337-44.
14. Blake DR, Waterworth RF, Bacon PA. Assessment of iron
stores in inflammation by assay of serum ferritin concentrations.
Br Med J 1981;283:1147-8.
15. Finch CA, Huebers H. Perspectives of iron metabolism. N
EngI J Med 1982;306:1520-8.
16. Blumberg AB, Mart HEM, Graber CG. Serum ferritin and
bone marrowiron in patients undergoing continuous ambulatory
peritoneal dialysis. J Am Med Assoc 1983;250:3317-9.
17. Wood WG. A comparison of eleven commercial kits for the
determination of serum ferritin levels. J Clin Chem Clin Biochem
1981;19:947-52.
Determination of Creatinine in Serum and Urine by Cation-Exchange High-Pressure Liquid
Chromatography
Aimo Harmoinen, Pekka Sillanaukee, and Hannu Jokela
We describe an HPLC method for quantifying creatinine,
separating the analyte from other compounds in serum
and urine by cation-exchangechromatography and mea-
suring its absorbance at 234 nm. The precision of the
method (CV) varied from 2.9% (mean creatinine concen-
tration, 31 pxnol/L) to 1.7% (361 mol/L) within a series of
assays and from 3.9% (34 /LmoI/L) to 2.4% (391 moI/L)
between series. A comparison with the Jaff#{233} method, as
performed with a Technicon SMA analyzer, gave the
regression line YHPLC = 1.OOXjatt#{233} - 12.0 (n = 141, r =
0.998, and S = 19). Results also are comparable with
those of an enzymatic method, if the enzymatic method is
standardized with a serum-based standard when serum
samples are measured. An aqueous standard has to be
used for enzymatic determination of creatinine in urine.
Additional Keyphrases:bilirubin interference Jaff#{233}, enzymic
methods compared . glomerulus function
The concentration of creatimne in serum and creati-
nine clearanceare generallyaccepted as reliableindices
of the glomerular filtrationrate. The methods most
commonly used for quantitativedeterminations of cre-
atinine are based on the Jaff#{233} reaction, involvingalka-
line sodium picrate. This reaction, however, is not
specific, being affected by numerous metabolites and
drugs (1-5). Many attempts to improve the specificity of
the method have not been altogether successful. Some
recently developed enzymatic methods (6-8) are prom-
ising, but a high concentration of bilirubin in the sam-
ple is still a problem (9).
Some scientists have used cation-exchange chroma-
Department of Clinical Chemistry, Tampere University Hospi-
tal, SF-33520 Tampere, Finland.
Received May 10, 1990; accepted January 10, 1991.
tography to separate creatinine from interferents (10-
13); HPLC methods for creatininedetermination have
also been described (14-19). Unfortunately, most of
these methods are very laborious (14-16), and some
require a complicated assay system (14, 17). The HPLC
method we describe has some advantages over the
methods published earlier. The simple sample handling,
the isocratic buffer system, and the shortretentiontime
of creatinine in the proposed method resultin a rapid
assay that is therefore also suitable in routine work.
Materials and Methods
Reagents. All chemicals were reagent grade. Creati-
nine standard was purchased from NIST (U.S. Depart-
ment ofCommerce National Institute of Standards and
Technology, Gaithersburg, MD 20899). The stock solu-
tion of creatinine (10 mmol/L) was prepared by adding
113.1 mg of creatinine to 100 mL of 0.1 molIL HC1
reagent. The working standard was prepared by dilut-
ing the stock solution10-foldwith distilled water. The
mobile phase was 75 mmol/L lithium acetate (Fluka
Chemie AG, Buchs, Switzerland) buffer, pH 7.1. The
proteins of the serum samples were precipitated with
trichloroacetic acid (TCA), 100 g/L.
Instruments. A Model 5000 liquid chromatograph
with a variable-wavelength UV-100 detector (Varian
Instruments, Walnut Creek, CA) was used. Sample (20
pL) injection was carried out with an MSI 660 autosam-
pler (Kontron AG, Zurich, Switzerland). A Model C-lB
Chromatopac (Shimadzu Corp., Kyoto, Japan) was used
as recorder. A cartridge column system (Chrompack,
Middelburg, The Netherlands) was used for separation:
The glass column (100 mm x 3 mm) was packed with
lonospher C, a silica-based cation-exchanger with sul-
fate as functionalgroup.
Sample preparation. Vortex-mix 300 pL of serum
Fig. 1.Two typicalchromatograms: (left) a serumdialyzed overnight
against distilled water, (right) the same serum before dialyzation,
containing 110 zrnoI of creatinine perliter
564 CLINICAL CHEMISTRY, Vol. 37, No. 4, 1991
sample 20s with 600 j.L ofTCA solution and remove the
precipitated proteins by centrifligation for 2 mm at
10 000 x g. Dilute urine samples 20-fold and then
proceed as for serum samples.
Standards. Because we did not find any matrix effect
from proteinsin the HPLC method, we used an aqueous
creatinine standard, 1000 jmoIJL. The same standard
was also used in the enzymatic method when urinary
creatinine was measured. The enzymatic determina-
tions in serum were standardized with SeronormtM (Ny-
comed AS, Oslo, Norway), containing creatinine at 177
.molJL.
Comparison methods. The Jaffe reaction was mea-
sured with a continuous-flow system (SMA 12/60 ana-
lyzer; Technicon, Tarrytown, NY). We used a serum-
based calibrator containing 395 mol of creatinine per
literas standard. The enzymatic test (no. 12320, Creat-
mine PAP; E. Merck, Darmstadt, F.R.G.) was carried
out with a Hitachi 704 analyzer (Boehringer Mannheim
GmbH, Mannheim, F.R.G.).Because the reactioncurve
is not strictly linear, we used a two-point method,
measuring the absorbance change between 100 and 220
s after additionof the start reagent.
Results and Discussion
At first, we used the two-buffer system reported by
Ambrose et al. (17). However, this buffer system was not
suitable with our lonospher C column and we could not
optimize the baseline the way they reported. After
numerous trials we chose an isocratic system with
lithium acetate buffer, 75 mmol/L, pH 7.1. When the
flow rate was 2 mL/min, the retention time of creatinine
was only 1 mm (Figure 1). The sensitivity of the method
was adequate for measuring creatinine in serum and
urine. The detection limit, which was <1 prnollL, could
easily be improved by reducing the attenuation of the
instrument or by introducing a bigger sample-volume
injection loop. The method was linear at least up to 2000
molIL: y (peak height) = 244x + 1.3 (r = 0.999).
The analytical recovery of creatimne was determined
by adding 500 and 1000 mol of creatinine per liter to
three serum samples already containing creatinine at 0,
111, and 385 molfL. Recovery was essentially com-
plete, varying from 97% to 102%. The precision of the
method was evaluated by 20 repeated analyses of three
serum samples with low, normal, and high creatinine
concentrations. The results are presented in Table 1.
We compared our HPLC method with the automated
Jaff#{233} procedure and with a commercially available en-
zymatic method (Table 2). The proposed method corre-
lated extremely well with the Jaff#{233} procedure. The
correlations with the enzymatic method were also good,
if the enzymatic method was standardized correctly.
When an aqueous standard was used in serum determi-
nations, the results by the enzymatic method were
>10% too low; vice versa, if serum-based standard was
used in urine determinations, the results were too high.
High turbidity or hemolysis of the sample did not
interfere with the HPLC method and only slightly with
the enzymatic method. Bilirubin concentrations s600
Table 1. PrecisIon of the HPLC Method for Creatlnlne
Within-assay (n = 25) Between-assay (n = 20)
Mean (SD), cv, Mean (SD), CV,
pmol/L % pmol/L S
31(0.9) 2.9 34(1.3) 3.9
114 (2.5) 2.2 111(2.8) 2.5
361 (6.1) 1.7 391(9.4) 2.4
Table 2. Correlation of the HPLC Results (y) with
Those by the Jaff#{233} Picrate and Enzymatic Methods for
Creatinine
Sample n r Regression equation S
Serum 141 0.998 y = (1.00 O.Ol)xjass - 12.0 MmOI/L 19
Serum 136 0.997 y = (1.01 O.Ol)xEflZ - 6.7 tanoi/L 23
Urine 36 0.984 y = (0.97 O.O3)Xja#{248}e - 0.02 mmoi/L 0.37
Urine 39 0.997 y = (1.01 O.O2)xEflZ- 0.01 mmol/L 0.37
Slope - mean SE.
tmo1JL did not interfere in the proposed HPLC method.
Although the manufacturer has added potassium
hexacyanoferrate(ll) to the new enzymatic kit to mini-
mize bilirubin interference, bilirubin concentrations >80
.anoI/L lead to decreased values in the enzymatic test.
It is unlikely that many substances would interfere in
the HPLC method because of the combined selectivity of
protein precipitation, cation-exchange separation, and
the detection wavelength used. We checked endogenous
and exogenous substances known to interfere in the
Jaff#{233} methods, structurally related compounds of creat-
mine, and some common drugs. Although some of these
substances (Table 3) absorb at the 234-nm wavelength, The object of this study was initially to develop an
only toxic concentrationsof phenytoin might interfere efficient referencemethod for evaluating routinemeth-
with this method. ods of determining creatimne. However, the proposed
method is so simple that it can also be used in routine
work, when a highly sensitiveand specific technique is
Table 3. Compounds Tested for Interference
required.
Concn, Retention time,
Title mg/L mm
References
Creatinine 1.06
Acetaminophen 250 0.35 1. Blass K, Thibert R, Lam L. A study of the mechanism of the
Jaffe reaction. J ClinChem ClinBiochem 1984;12:336-56.
Acetoacetate 500 nd
2. Bowers L. Kinetic serum creatinine assays. I. The role of
Acetosalicylic acid 1500 nd various factors in determining specificity. Clin Chem 1980;26:
p-Aminosalicylic acid 1250 0.57 551-4.
Ascorbic acid 100 nd 3. Bowers L, Wong E. Kinetic serum creatinine assays. II.A
Bilirubin 350 nd critical evaluation and review.Clin Chem 1980;26:555.-61.
Carbamazepine 14 nd 4. Soldin S, Henderson L, Hill J. The effect of bilirubin and
ketones on reaction rate methods forthemeasurement ofcreati-
Cephalothin 2500 1.33 nine. Clin Biochem 1978;11:82-6.
Chlonazepam 130 nd 5. Swain R, Briggs S. Positive interference with the Jaff#{233} reaction
Chiorothiazide 150 0.35 by cephalosporin antibiotics. ClinChem 1977;23:1340-2.
Creatine 80 nd 6. Yasuhara M, Fujita S, KohdaK, Hayashi C. A new enzymatic
method to determine creatinine. Clin Chim Acta 1982;122:181-8.
Digoxin 50 1.64 7. Fossati P, Lorenzo P, Giovanni B. Enzymatic creatinine assay:
Disopyramide 500 nd a new colorimetric method based on hydrogen peroxide measure-
L-Dopa 400 0.36 ment. Clin Chem 1983;29:1494-6.
Ethosuximide 130 nd 8. Artise J, McEnroe R, Zak B. Bilirubininterference in a perox-
Furosemide 500 0.35 idase-coupled procedure for creatinine eliminated by bilirubin
oxidase. ClinChem 1984;30:1389-92.
Gentamicin 8 nd 9. LindbackB, Bergman A. A new commercial method for the
Gentisic acid 250 nd enzymatic determinationofcreatinine in serum and urine evalu-
Glucose 15000 nd ated: comparison witha kinetic Jaff#{233} method and isotopedilution-
Histidine 1000 nd mass spectrometry. ClinChem 1989;35:835-7.
10. PolarE, MetcoffJ. True creatinine chromogen determination
Hydrochlorothiazide 500 0.36
in serum and urine by semiautomated analysis. Clin Chem
Lidocaine 500 nd 1986;11:763-70.
Meprobamate 100 nd ii. Teger-Nilsson A. Serum creatinine determination using an
6-Mercaptopurine 80 0.44 ion-exchange resin. Scand J Clin Lab Invest 1961;13:326-31.
Methenamine 500 nd 12. Rockerbie R, RasmussenK. Rapid determination of creatinine
Nalidixic acid 500 nd by an ion-exchangetechnique.Clin Chim Acta 1967;15:475-9.
13. MitchellR. Improved method for specific determination of
Nitrofurantoin 500 0.59 creatinine in serum and urine. ClinChem 1973;19:408-10.
Phenobarbital 40 nd 14. Brown N, Sing H, Neeley W, KoetitaE. Determination of
Phenytoin 100 1.05 true serum creatinine by high-performance liquid chromatogra-
Pivmecillinam 500 nd phy combined with a continuous-flow microanalyzer. Clin Chem
1977;23:1281-3.
Pnmidone 15 nd 15. Spierto F, MacNeil M, Culbreth P, Burtis C. Development and
Procainarnide 500 nd validation of a liquid-chromatographic procedure forserum creat-
Propranol 500 nd mine. Clin Chem 1980;26:286-90.
Quinidine 500 0.61 16. Paroni R,Arcelloni C,Fermo I, Bonini PA. Determination of
creatinine in serum and urine by a rapid liquid-chromatographic
Spironolactone 500 nd method. Clin Chem 1990;36:830-6.
Suifathiazole 300 0.77 17. Ambrose R, Ketchum D, Smith J.Creatimne determined by
Theophyllin 500 0.63 high performance liquid chromatography. Clin Chem 1983;
Tnmethopnm 29:256-9.
Tolbutamide 1100 0.64 18. Nishimaki S, SudoJ, Tanabe T. HPLC determination oftrue
creatinine in ratplasma. Dcv Toxicol Environ Sci 186:251-4.
Urate 350 nd 19. Bjerve K, Egense J, Lampinen L, Masson P. Evaluation of
Urea 6000 nd several creatinine methods in search of a suitable secondary
Vaiproic acid 130 nd reference method: report from the subcommittee on reference
method for creatinine, Nordic Society for Clinical Chemistry.
nd = riot detected. Scand J Clin Lab Invest 1988;48:365-73.
CLINICALCHEMISTRY, Vol. 37, No. 4, 1991 565