
weight according to the equation of Dubois and
Dubois [7].
2.3. Methods, reagents and instruments
Serum, plasma and urinary creatinine was mea-
sured by the conventional kinetic Jaffe
´(Jkin) meth-
od, the new Jcomp method and by a modified
version of the established enzymatic creatinine PAP
method called Creatinine Plus (Crplus). All reagents,
including calibrators and controls were provided by
Roche Diagnostics (Mannheim, Germany). Measure-
ments were carried out on a Hitachi 717 analyzer
throughout.
The Crplus method is based on the same reaction
sequence as the creatinine PAP method consisting of
four consecutive enzymatic steps via creatininase,
creatinase, sarcosine oxidase and peroxidase [8], but
using a different chromophore (2,4,6-triiodo-3-
hydroxybenzoic acid), and a modified detergent com-
position. Endogenous creatine is completely removed
during a preincubation step. Lipemic samples are
rapidly cleared by the detergent system, which is
effective at 37 jC only. Bilirubin interference is
eliminated by the combined effect of detergents and
the use of a new chromophore. The concentration of
reagents was optimized for completion of the reaction
after 10 min. The assay was standardized with a
isotope dilution mass spectrometry method for creat-
inine. Measuring range from lower detection to dilu-
tion limit is 0.03–30 mg/dl creatinine.
In both, the Jkin and the Jcomp method a reduction
of bilirubin interference is achieved by means of a
rate-blanked mode during the preincubation period.
To further increase the specificity, an additional cor-
rection is made in the Jcomp method for non-creati-
nine Jaffe
´-reactive components, e.g. proteins, glucose,
ascorbic acid, ketone bodies, by subtraction of 0.3 mg/
dl from the result obtained for all Jaffe
´-reactive serum
components. This numerical term reflects the average
contribution of the serum matrix to the Jaffe
´reaction
signal as estimated in a comprehensive multicenter
study [9] and later confirmed by Mazzachi et al. [1].
This means that creatinine-free serum samples will be
measured by the Jkin method as having on the average
0.3 mg/dl creatinine; the existence of such a ‘‘blank
value’’ has been reported earlier [10,11], but never
been quantified.
The calibrator (c.f.a.s.) set points for both Jaffe
´
procedures are determined by measuring several hu-
man serum pools with the reference and the Jkin
method. As preliminary set point the creatinine con-
centration of the calibrator as determined with the
reference method is used. Due to the presence of non-
creatinine Jaffe
´reactive constituents in the calibrator
matrix, a distinct difference is found between the
matched data pairs as illustrated in Fig. 1a.The
correct set point for the Jkin method is then calculated
from the regression equation of that curve.
If the Jkin values of the serum pools are divided by
the slope of the regression line shown in Fig. 1a and
plotted versus the reference method values, the result-
ing regression curve parallels the identity line, but in a
distance which is determined by the non-creatinine
Jaffe
´reactants in the serum pools (Fig. 1b). Division
of the preliminary calibrator value by the slope of the
regression line in Fig. 1a then defines the Jcomp set
point. However, the results of serum samples obtained
with this set point have finally to be corrected for the
non-specificity of the Jaffe
´method by subtraction of
0.3 mg/dl.
The measuring range for the Jcomp method is
0.2– 25 mg/dl.
2.4. Statistics
Reference intervals were calculated using an Excel
compatible non-parametric program for descriptive
statistics. Significance was tested with a ‘‘Wilcoxon
two-sample test’’ for the differentiation between the
male and the female group and with a ‘‘Wilcoxon
two-sample test in case of dependent samples’’ for
the differentiation between the results obtained with
the two creatinine assays for the same reference
groups. Regression analysis in method comparisons
was done with non-parametric Passing/Bablok calcu-
lation and a linear regression model to establish
correlation coefficients.
3. Results and discussion
3.1. Analytical variability
Standard deviations of both Jaffe
´methods are the
same, CVs necessarily higher for the Jcomp due to the
W. Junge et al. / Clinica Chimica Acta 344 (2004) 137–148140