Creatinine determination in urine samples by batchwise kinetic procedure and flow injection analysis using the Jaffé reaction: chemometric study

Talanta 55 (2001) 1079– 1089 www.elsevier.com/locate/talanta

Creatinine determination in urine samples by batchwise kinetic procedure and flow injection analysis using the Jaffe´ reaction: chemometric study P. Campins Falco´ *, L.A. Tortajada Genaro, S. Meseger Lloret, F. Blasco Gomez, A. Sevillano Cabeza, C. Molins Legua Departament de Quı´mica Analı´tica, Facultad de Quı´mica Uni6ersitat de Valencia, C/Dr. Moliner 50, E46100 -Burjassot, 46100 Valencia, Spain Received 10 April 2001; received in revised form 21 June 2001; accepted 4 July 2001

Abstract The classical Jaffe´ reaction for the determination of creatinine in urine samples is tested. A comparative study of the main analytical characteristics focussed to minimize the bias error and improve the precision, for the batchwise and flow injection (FI) methods is realized. Also, the effect of the albumin concentration in the determination of creatinine has been studied. Different analytical signals were studied. Absorbance increments at different times permit to estimate the creatinine concentration free from bias error in urine by the batchwise method using the calibration graph obtained with creatinine standards and no measurement of the blank solution is needed. The lineal interval was 0.92– 50 mg l − 1 and seven samples can be processed per hour by an operator. No previous treatment of the urine sample is necessary. The FI method provides also good results. The lineal interval was 30 – 100 mg l − 1 and the sample rate was around 20 samples per hour. If increased albumin levels are detected in the urine, standard addition method or the calibration graphs with standards in presence of albumin are needed in order to obtain accurate results when FI method is employed. The obtained accuracy of the both methods allows its application as diagnostic tool to establish the urinary creatinine levels. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Creatinine; Urine; Flow and batch mode; Bias error

1. Introduction The creatinine is a nitrogen compound formed by the nonenzymatic cyclization of creatine in muscle tissue [1]. Its level in biological fluids * Corresponding author. Tel.: +34-96-3983002; fax: + 3496-3864406. E-mail address: [email protected] (P. Campins Falco´).

(serum and urine) is generally accepted as an indication of the presence or absence of renal failure, muscular and thyroid functions. Since the creatinine concentration in urine is decreased in muscular atrophy and myotonic dystrophy, urinary creatinine in newborn babies and infants is monitored. The concentration range of creatinine is 1000 –2500 mg per 24 h and its production and release occurs at a relatively constant rate.

0039-9140/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 0 1 ) 0 0 5 2 2 - 7

1080

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

A problem that is commonly associated with the quantitative measurement of any compound of clinical interest in urine, is that some correction must be made for the daily variations that occur in urine output and volume. One way this problem can be minimized is by collecting samples over long time periods (24 h). For short-term samples, an alternative approach is to divide the measured concentration of an analyte by the concentration of some reference compound present in the same matrix. Generally, in clinical laboratories, the creatinine is used in this sense [2– 4]. Its concentration is also used as a correction factor for fluctuations in urine volume [2]. Although several methods have been proposed for the determination of creatinine, for instance, capillary electrophoresis (CE) [5,6], liquid chromatography [7,8] or enzymatic methods [9], the classical Jaffe´ reaction continues being the method of choice [10]. Table 1 shows a selection of the main procedures described in literature corresponding to creatinine determination in urine samples. As can be seen, the sample treatment, the lineal interval or R.S.D. (%) reached for the different procedures are similar. However, analytical properties such as analytical cost or time analysis are improved by the procedures that make use of Jaffe´ reaction. In these methods, creatinine is combined with an alkaline picrate solution to form an orange– red complex. Several studies have emphasized the use of the kinetic phase of the reaction in order to reduce the effects of interferences [11– 15] and optimized batchwise procedures have been described for biological fluids. Because of a rapid procedure for urinary creatinine is needed, Flow Injection Analysis (FI)based Jaffe´ reaction has been also proposed [16,17]. As can be seen in Table 1, there are not big differences between the procedures described, being the cost and time analysis the most relevant. The aim of this paper is to study the analytical characteristics and the presence or absence of bias error, for the batchwise and FI methods in the determination of creatinine in urine samples. The two methodologies based on the kinetic phase of the Jaffe´ reaction are chemometrically tested.

2. Experimental

2.1. Reagents and apparatus All solutions were prepared in distilled water and all reagents were of analytical grade unless stated otherwise. Sodium hydroxide (Probus) stock solution, 5.0 mol l − 1 and picric acid (Merck) stock solution, 2.08× 10 − 2 mol l − 1 were used. Working alkaline picrate reagent was prepared by mixing 90 ml of picric acid and 10 ml of sodium hydroxide stock solutions. This reagent was prepared fresh daily. The creatinine (Fluka) stock solution was 1000 mg l − 1, working standards were prepared by diluting this stock solution. Albumin solution, 50 000 mg l − 1 was prepared from bovine albumin (Fluka, Fraction V). A detection system consisting of a HewlettPackard HP8453 UV-Visible spectrophotometer was used. The FI peaks were recorded every second from 0 to 60 s, and the spectra every nanometer from 200 to 800 nm. The spectrophotometer was interfaced to a Hewlett-Packard XM 5/90 personal computer, equipped with G1115AA software. The pH was measured with a Crison micropH 2000 pH-meter.

2.2. Batchwise procedure In a plastic cuvette were placed appropriate volumes of standard creatinine or a volume of urine from different volunteers (0.05–0.08 ml) plus distilled water up to 1.5 ml [13–15]. To the mixture was added 1 ml of alkaline picrate, so the final volume in the cuvette was always 2.5 ml and the reactant concentrations were as follows: picric acid, 6.6× 10 − 3 mol l − 1 NaOH, 0.2 mol l − 1 creatinine between 3.2 and 100 mg l − 1. The reaction was assumed to start after the last drop of the reaction mixture was added to a 1 cm pathlength quartz cell. Absorbance values at 485 nm and 25 °C were registered at 45 or 60 and 180 s. A reaction mixture blank was also prepared and measured. The number of replicates appears in the text.

Table 1 Principal analytical properties of the main procedures described in the literature about creatinine determination in urine samples Sample treatment

Calibration graph

R.S.D. (%)

CE CZE

20 ml diluted 25-fold Centrifuged and diluted (1:80)

2.98–6.8 2.9–8.4

HPLC Enzymic rate assay FIA-UV (Jaffe´ reaction)

Centrifuged and diluted (1:49) Diluted 25-fold Collected on a soft adsorbent cotton pad and wrung Diluted 25-fold Diluted 1:100

5–80 mg l−1 23.1 mM–0.2 mM 10–300 mM −500 mg l−1 5–200 mg l−1

FIA-UV (Jaffe´ reaction)

0.8–5.8 1.2–2.7 2.9

LD

3 mM 3 ppm

3–10

Kinetic spectrophotometric (Jaffe´ reaction)

0.05–0.45 ml directly mixed with reagents

0.92–100 mg l−1

FIA-UV (Jaffe´ reaction)

1 ml diluted 25-fold

30–60 mg l−1

2.7–3.7

0.3 ppm

3 ppm

Time analysis

Cost analysis

Reference

12 min \3.5 min

**** ****

[5] [6]

40 min \15 min Sample throughout h−1 Sample throughout h−1 Sample throughout h−1 Sample throughout h−1

*** ** ***

[7] [9] [16]

***

[17]

*

This work

**

This work

102

120

7

20

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

Technique

1081

1082

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

2.3. Flow-injection procedure (FI) The manifold of the FI system used is shown in Fig. 1. The reagent was alkaline picrate (picric acid 0.025 M and NaOH 0.2 M) and the carrier was distilled water. A peristaltic pump, Gilson Miniplus 3, worked with a flow rate of 1.75 ml min − 1. The loop has a 500 ml internal volume and the coil was 10 m long. A six-port multiposition valve (LEA130BP6P) electrically actuated (LEA 1301D) was used along with the fittings and tubing to complete the system. Tygon tubing (i.d.= 0.8 mm) was used with peristaltic pump and other tubing was made of PTFE with i.d.= 0.5 mm. The absorbance of the picrate-creatinine addition compound was monitored at 485 nm and 25 °C after 170 s from the injection of the creatinine in the FI system. A quartz microcell of 10 mm pathlength was used. The linear and dynamic ranges of the FI system were determined by making injections of 0– 60 mg l − 1 creatinine standards.

samples were employed, to which variable volumes of a creatinine standard were added giving concentrations between 0 and 50 mg l − 1, when the batchwise procedure was employed. For the FI analysis, the diluted urine samples were fortified with creatinine concentration between 0 and 60 mg l − 1. Then, the samples are processed as mentioned above, respectively.

2.6. Interferences A set of creatinine standards containing 300 mg l − 1 of albumin was prepared in order to study the possible modification of the analytical signal for creatinine in urine of patients with microalbuminuria. Calibration graphs of standard solutions and urine samples (fortified or not with albumin 300 mg l − 1) were performed.

3. Results and discussion

3.1. Selection of analytical signals 2.4. Urine sample preparation In batch procedure volumes of the urine samples ranged between 0.05– 0.45 ml were directly added to the mixture reagent. For FI the urine samples were diluted 25 times, and volumes of 500 ml were injected in the system.

2.5. Youden and standard addition methods Volumes of the human urine samples between 0.05 and 0.45 ml were used for the Youden method [18]. For the standard addition method [19,20] volumes of 0.05– 0.08 ml of the urine

Fig. 1. FI assembly employed for the determination of creatinine. C, carrier (distilled water); S, sample/distilled water; St, standard.

Previous studies showed that experimental conditions for the determination of creatinine must be chosen so that the kinetic behavior is the same in the dynamic concentration range, with a constant value of the rate constant [13–15]. One of these conditions are: 0.2 M sodium hydroxide; 6.6×10 − 3 M picric acid; 25 °C and 485 nm at which the batchwise procedure is carried out; and 45–180 s time interval. Similar experimental conditions are achieved in the FI manifolds [16,17]. Predilution of the urine samples is generally preferred instead of Jaffe´ procedures that work in less sensitive experimental conditions for avoiding this step. Bearing in mind the previous statements, the analytical signals tested are absorbances around 485 nm measured at 45, 60 and 180 s, and absorbance increments between 45 and 180 s and 60 and 180 s for the batchwise procedure. Fig. 2 shows the linear regressions obtained at 485 nm for different times and the corresponding absorbance increments versus concentration plots. Measurements at 470 and 490 nm have been also included, due to the features of the spectrum of

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

1083

Fig. 2. Calibration curves measured at u= 470, 485, 490 and at fixed time (a) A(t) vs. ccreatinine (b) and increment A(t2) −A(t1) vs. ccreatinine.

the complex (see Fig. 3). As can be seen in Fig. 2 Fig. 3, the blank signal is higher at 470 nm than measuring at 485 or 490 nm. The best slope values are obtained at 180 s at any tested wavelength. Similar sensibilities are obtained processing absorbance increments. Table 2 shows some analytical characteristics of the batchwise procedure at 485 nm in function of the analytical signal used, n indicates the number of replicates made in several months. As can be seen the best results are obtained when the absorbance increments are used because the precision of the method is improved. The best detection limit is achieved working with absorbance increment between 180 and 45 s as it is shown in Table 2. As can be seen high detection limit (LD) and quantification limit (LQ) are obtained by working at 45 or 60 s., because of the standard deviation (S.D.) of the blank solution is too high. These limits were estimated as 3 SB (SB = blank S.D.) or 10 SB divided by the slope, for LD and LQ, respectively. The ordinate values of the different calibration graphs, DA45, 180 versus creatinine concentration, are coincident with the analytical signals of the measured blanks. The least squares method gives the following equation: Aordinate =1.014 Ablank ( 90.27) + 0.0047

(9 0.01), r 2 = 0.9991 (Syx = 1.427×10 − 3), being the general equation A= b Ablank (9 Sb) + a(9 Sa), b=slope, Sb= slope S.D., a= ordinate, and Sa= ordinate S.D. The test t applied to the ordinate gave tcal = 3.7014, h=0.066(so the ordinate is statistically equal to cero). The 1 is included in confidential interval of the slope at 95% of probability. According to these results, it can conclude that no bias error is introduced by the blank in the calibration step.

Fig. 3. Absorption spectra of creatinine-Jaffe´ reaction. (1) Blank reagent at 60 s; (2) Blank reagent at 180 s; (3) Creatinine 25.6 mg l − 1 at 60 s and (4) Creatinine 25.6 mg l − 1 at 180 s. Conditions: Picric acid 6.6 ×10 − 3 mol l − 1, sodium hidroxide 0.2 mol l − 1 and 25 °C.

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

1084

Table 2 Parameters of linear regression for different analytical signal measured at u =485 nm Analytical signal

A45

A60 A180

DA180,45

DA180,60 A170a

b 9 Sb

0.007 9 0.002 (n= 8)

0.006 90.001 (n =3) 0.017 9 0.003 (n = 11)

0.011 9 0.001 (n= 8)

0.008 90.001 (n =3) 0.0129 90.0001 (n = 3)

% CV of b

28

LD (mg l−1)

19.5

LQ (mg l−1)

65

17

4.4

14.6

18

4.7

15.7

13

15 1

0.3

0.6 3

0.92

2.0 30

Interday

Intraday

C (mg l−1) 9s

(E%)

C (mg l−1) 9s

(E%)

10.8 9 7.7 (n =) 25.12 911.4 (n = 3)

−15.6

13.10 98.2 (n = 3)

+2.3

2.98 93.13 (n = 3)

+1.40

2.7 9 0.8 (n = 3)

−0.8

12.0 93.4 (n =6) 27.5 96.0 (n = 6) 13.4 9 3.7 (n =3) 30.8 97.2 (n =3)

−21.5

−6.4 +14.15 +4.3 −3.9

16.07 9 0.09 (n =3) 32.4 9 2.0 (n =3)

0.43 1.3

b, slope of the calibration graph; Sb, slope S.D.; CV%, variation coefficient; n, number of calibration curves made; c, concentration found in standard solutions; s, S.D.; E%, relative error. a FI method.

No better variation coefficients are obtained processing multivariate calibration as multiple linear regression (MLR) employing analytical signals at different times at the same wavelength or different wavelengths at the same time. Also multicolinearity problems arise. In order to avoid these problem Partial Least Squares (PLS) and Principal Component Regression (PCR) algorithm were used. For this purpose the concentrations of the standards were used as the Y-block in combination with the X-block data (absorbances between 440– 600 nm at 180 s). The PLS model gave a percentage of explained variance (%EV) of 98% and 2% for the first and second latent variable (LV), respectively, for the X matrix, and 98%, 1% for the Y matrix. Similar results of %EV were obtained for X matrix by employing the PCR algorithm. In Fig. 4 the scores plot is shown, it can be seen that the concentration is related with the first PC or VL. The

second PC or VL is due to other variation source, probably the blank reaction signal. Table 3 includes the root mean squared RMSE, considered as a measure of the prediction ability of these calibration models, being RMSE= (1/n(yˆ i − yi))1/2, where n indicates the number of tested samples, yˆ i are the predicted values by interpolation in the model, and yi are the real values. Good RMSE values have been obtained, these values are near to the detection limits obtained in the batch procedure given in Table 2. Predicted versus measured equations are also included for standards. As can be seen, the unit was included in the confidence interval of the slope and the zero was included in the confidence interval of the intercept. From these data it can concluded that the models obtained are good and they could other option for calibration, although more computer time is needed than that required by the DA45, 180 method.

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

1085

ported value corresponds to repeatability instead of reproducibility as it is given for the batchwise method. The sensitivity achieved by the FI method is the 75% of the value obtained measuring in batch mode. The detection limit was similar; this value was obtained considering a signal/noise ratio of three. The accuracy and precision of both procedures are also showed in Table 2, and acceptable results are obtained.

3.2. Determination of creatinine in urine samples

Fig. 4. Scores plots obtained by applying PLS and PCR algorithm. LV, latent variable, PC, principal component. The points one upto eight correspond to creatinine concentrations (10, 20, 30, 40, 50, 70, 80 and 100 ppm).

The optimization of the FI conditions provides the best analytical signal at 170 s, 0.2 M sodium hydroxide, 0.025M picric acid and 485 nm. The proposed FI procedures [16,17] measure the absorbance at smaller times but use higher temperatures (35, 50 °C). The literature option makes the manifold more expensive than that proposed in this paper. For FI method using the signal at FI peak at 485 nm, the precision is better than that achieved by the batchwise method (Table 2). Although this fact is inherent to automated methods, the re-

3.2.1. Testing bias error The presence or absence of constant and/or proportional bias error due to the matrix of the sample is tested by employing the Youden and the standard addition methods (MOSA), respectively. Fig. 5 shows the results of Youden method. The signals for different urine volumes are presented using different times or intervals. The presence a constant error is evident when the single absorbance signal is employed (Fig. 5a). The use of absorbance increments (Fig. 5b) permits correct the blank error, and the total Youden blank (TYB) can been considered equal to zero (TYB= Ordinate= 7.605×10 − 2 9 0.49, and the Student’s t tcal = 1.552, h= 0.196). The results of standard addition method are presented in Fig. 6. As can be seen in both figures (a and b), the different At1 or the different DAt1-t2 versus cadded lines intersect at the same point, that should correspond to the unknown concentration of creatinine in the urine samples. However, by using DAt1 − t2 the absence of constant errors is achieved, what would indicate that the time evolution of the matrix would be a straight line. The comparison of calibration and standard addition regression confirms of the presence or

Table 3 Parameters of multivariate regression by applying PLS and PCR algorithms Model

PLS PCR

RMSE

1.056 1.058

Predicted vs. measured Intercept 9 Sa

Slope 9 Sb

R2

Syx

0.0849 0.925 0.084 9 0.927

0.998 9 0.019 0.998 9 0.019

0.9990 0.9990

1.5412 1.5443

1086

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

Fig. 5. Youden calibration curve measured at u= 485 nm and at fixed time (a) A(t) versus 6urine and (b) increment A(t2) −A(t1) versus 6urine, for sample s-5.

absence of proportional errors. The slope obtained by using DAt1-t2 (b9 Sb) for both regressions are comparable: bstandard addition/bcalibration = 1.059 0.05. The application of two sample t test for the slopes (n=10), confirms that both results are statistically equal (tcal =0.382, h = 0.713). The percentage recoveries obtained for fortified urine samples were 9493 (n = 20), 11794 (n = 20), 10996 (n =20) 1149 10 (n =5) and 1149 11 (n=5) for the A45s, A180s, DA180-45s, PCR and the PLS models, respectively. Based on these results, the method based on DA180-45s signal provided the best recoveries and was selected for further calculations. For FI method, the obtained standard addition equation for urine sample was: A485 nm =0.0137 ccreatinine (90.0001)+ 0.5638 (90.09), r 2 =0.998 (Syx =1.718× 10 − 2) using the FI peak signal. In these conditions, we can observe the absence of matrix effect because the slope value is statistically equal to that obtained with standards (see Table 2).

3.2.2. Interferences In serum or plasma there are several non creatinine solutes that can give a response in the Jaffe´

reaction. Urine usually has fairly low concentrations for most of these compounds and in the studies proposed in the literature [16,17,21] the concentrations of the potential interferences at which deviations in the creatinine response begin to occur were much higher than those seen in typical urine samples. T. Sakai et al. [17] reported that organic compounds such as urea, glucose, lactose uric acid, acetone and creatine had no significant effect even at a concentration of 10–200 times that of creatinine in this determination. Ammonia or the common inorganic salts also did not interfere. Results in the literature suggest that albumin is the most serious interferent [12]. Our previous studies [13–15] demonstrated that its effect is greatest during the early part of the reaction and to obtain unbiased results for creatinine in serum samples the standard addition method should be applied. The effect of the albumin concentration in the determination of creatinine in urine has been studied. Normal levels of urinary albumin occur at approximately 5–30 mg l − 1 for a 1-l daily urine output. For the same urine volume, the presence of increased albumin levels (i.e. microal-

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

buminuria) is generally said to occur when concentrations of 30– 300 mg l − 1 are detected. The concentration of added albumin is 300 ppm to standards and to samples, this concentration is equivalent to 7.5 g l − 1 in urine. A similar behavior to that previously reported [14,15] for standards containing albumin at a concentration of 7.5 g l − 1; no variation in the slope of the calibration graphs is produced with respect to that obtained with creatinine alone. The regression of spiked and non-spiked absorbance increments DA45, 180 has a slope (b 9 Sb) of (1.04 90.05) and an intercept (a 9Sa) of (0.0059 0.005) with r 2 =0.998. The obtained slopes and intercepts for both regressions are comparable, therefore, the absence of bias error is corroborate for the batchwise procedure, even for intense proteinurie (albumin concentration \3.5 g 24 h − 1). When the FI method is applied, the slope (1.046× 10 − 2 90.0001) corresponding to the standards fortified with albumin concentration constant decreases in a percentage of 20% with respect to that obtained with standards without albumin Table 2. The regression of data corre-

1087

sponding to albumin spiked or non-spiked has a slope ranged from 0.797 to 0.823 (95% probability), being the 1 not included in this interval. However, the slope obtained applying the standard addition method (1.087× 10 − 2 9 0.0001) was similar to that obtained with the standards containing albumin. Thus, the bias error can be eliminated applying this method as previously established in previous works for serum samples [14,15].

3.2.3. Determination in urine samples The estimated concentration of creatinine in urine samples using different analytical signals is presented in Table 4. These concentrations have been calculated by applying the MOSA and the calibration graphs with standards. In general, by using the batch procedure when the analytical signal used is DA180-45s, the results obtained by using the calibration graphs are comparable to that obtained by using standard addition method, hence providing yet another possibility to the determination of creatinine in urine samples. These results confirm that in these conditions any systematic errors are present. The precision of

Fig. 6. Standard addition calibration measured at u= 485 nm and at fixed time (a) A(t) versus cadded A(t2) − A(t1) versus cadded creatinine for sample s-6.

creatinine

and increment (b)

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

1088

Table 4 Found creatinine concentration (mg l−1) in different human urine samples using different analytical signals: (A) Batch procedure: Absorbance at u=485 nm at different times (180 s and increment in the interval 45–180 s); (B) FI procedure: absorbance at u= 485 nm at 170 s Sample

Volume urine (ml)

1-A

0.05

2-A

0.05

3-A

0.05

4-A

0.05

5-A

0.05

5-A

0.05–0.45

6-A

0.1

6-A

0.1–0.4

7-B 8-B 8-B

0.1 0.1** 0.1**

Signal

A180 A180-45 A180 A180-45 A180 A180-45 A180 A180-45 A180 A180-45 A180 A180–45 A180 A180-45 A180 A180-45 A170 A170 A170

(t)

Found concentration (mg l−1) MOSA

Calibration graph

Eq. (1)

Eq. (2)

753 805 577 649 660 592 415 345 507 638

1032 9 152 (n =9) 894 990 (n =9) 710 994 (n =6) 735 985 (n =6) 710 999 (n =4) 593 996 (n =4) 626 9128 (n= 6) 485 9108 (n= 6) 983 9330 (n= 6) 564 929 (n =6) 620 983 (n =5) 565 9100 (n =5) 1779 9 84 (n= 5) 1589 9 89 (n= 5) 1331 9182 (n = 4) 1300 9175 (n = 4) 966 931 (n =16) 1255 9 43 (n= 16)

933 9 41 (n = 9) 823 9 33 (n = 9) 650 9 33 (n =6) 633 9 24 (n = 6) 808 9 75 (n =4) 581 9 75 (n = 4) 560 9 22 (n =6) 334 9 16 (n = 6) 960 9 289 (n = 6) 646 9 33 (n = 6)

969 9 74 (n =9) 823 9 36(n= 9) 726 9 22 (n = 6) 645 9 29 (n =6) 761 9 80 (n = 4) 541 9 51 (n =4) 503 9 20 (n = 6) 250 9 12 (n =6) 702 9 126 (n = 6) 643 9 32 (n =6)

1640 922 (n = 5) 1579 974 (n = 5)

1620 930 (n = 5) 1432 9 19 (n = 5)

992 9 36 (n = 16) 1279 934 (n = 16)

992 9 36 (n =16) 1279 934 (n =16)

1645 1484

966 1191

**, fortified urine samples with 300 mg l−1 albumin.

these results is also better than that obtained by using a single absorbance. In relation to FI procedure similar creatinine concentration is obtained by using the calibration graph with standard and the standard addition method. For urine samples with high albumin concentration (microalbuminurie), MOSA or the calibration graph obtains comparable results with standards fortified with constant albumin concentration. Different urine volumes of samples 5 and 6 were processed. As can be seen in Table 4 the results obtained by using the calibration graphs with standards is similar to that obtained with MOSA. In order to improve the results precision, no higher urine volumes of 0.25 ml processed were included in the concentration estimation. Taking into account the absence of proportional and constant errors, the creatinine concen-

tration (Cm) in the urine sample can be calculated by using the following equations: Cm =

Sm × Cadded Sr − Sm

(1)

Sm × Cr S*r

(2)

or Cm =

(Sm) is the absorbance value of the samples with an unknown creatinine concentration (Cm) and (Sr) the absorbance value of a reference solution with a concentration (Cr) that contains the same volume of urine as the sample and a known addition of creatinine standard solution (Cadded). The analytical signal for this solutions is Sr = Sm + Sadded. The Eq. (2), only requires to known the sample analytical signal (Sm) and the concentration (Cr) of an standard and its signal (S*). r

P. Campins Falco´ et al. / Talanta 55 (2001) 1079–1089

In Table 4 are shown the results obtained using the Eq. (1) and Eq. (2) and it can be seen that they are in agreement with those obtained using the standard additions method and the calibration graphs with standards.

4. Conclusions This paper has shown the possibility to determinate creatinine in normal human urine sample by the Jaffe´ reaction free of any systematic error. The analytical signal used was DA180-45s as analytical signal in the batch procedure and absorbance at mixed time 170 s for the FI procedure. The application of Youden’s and MOSA methods, show that the proposed methods do not exhibit any constant or proportional systematic error. So the creatinine concentration can be calculated using the calibration graphs with standards or analytical signal of the reference standard solution. In patients with microalbuminurie, the presence of high albumin concentration exhibits a proportional systematic error in the FI procedure. However, the creatinine concentration can be calculated by using the MOSA method or the calibration graphs with standards fortified with constant albumin concentration.

Acknowledgements The authors are grateful to de DGICYT (Project N° PB97-1387) for the financial support. L.A. Tortajada-Genero acknowledges a grant from the Ministerio de Educacio´ n y Ciencia (Spanish Ministry) for carrying out PhD studies.

1089

References [1] N.W. Tietz (Ed.), Textbook of Clinical Chemistry, Saunders, Philadelphia, 1986. [2] J. Lemann Jr, B.T. Doumas, Clin. Chem. 33 (1987) 297. [3] Z.K. Shihabi, J.C. Konen, M.L. O’Connor, Clin. Chem. 37 (1991) 621. [4] E.M. Vestergaard, H. Wolf, T.H. Orntoft, Clin. Chem. 44 (1998) 197. [5] M.G. Cheng, T. Liang, K.Y. Zhou, G.X. Ling, SEPU 16 (2) (1998) 149. [6] R. Gatti, V. Lazzarotto, C.B. de Palo, E. Cappellin, P. Spinella, Electrophoresis 20 (14) (1999) 2917. [7] K.J. Shingfield, N.W. Offer, J. Chromatogr. B 723 (1-2) (1999) 81. [8] J.A Resines, M.J Arin, M.T Diez, P. Garcia del Moral, Liquid Chromatogr. Relat. Technol. 22 (16) (1999) 2503. [9] T. Fujita, S. Takata, Y. Sunhara, Clin. Chem. 39 (10) (1993) 2130. [10] J. Vasilades, J. Clin. Chem. 22 (1976) 1664. [11] M.H. Kroll, R. Chesler, C. Hagengruber, D.W. Hestner, M. Rawe, Clin. Chem. 32 (1986) 446. [12] H.L. Pardue, B.L. Bacon, M.G. Nevius, J.W. Skoug, Clin. Chem. 33 (1987) 278. [13] M. Llobat Estelle´ s, A. Sevillano Cabeza, P. Campı´ns Falco´ , Analyst 114 (1989) 597. [14] P. Campı´ns Falco´ , A. Sevillano Cabeza, M. Llobat Estelle´ s, Analyst 114 (1989) 603. [15] F. Bosch Reig, P. Campı´ns Falco´ , A. Sevillano Cabeza, R. Herra´ ez Herna´ ndez, C. Molins Legua, Anal. Chem. 63 (1991) 2424. [16] J.F. van Staden, Z. Fresenius, Anal. Chem. 315 (1983) 141. [17] T. Sakai, H. Ohta, N. Ohno, J. Imai, Anal. Chim. Acta 308 (1995) 446. [18] W.J. Youden, Anal. Chem. 19 (1947) 946. [19] J.S. Foster, G.O. Langstroth, D.R. McRae, Proc. R. Soc. London, Ser. A 153 (1935) 141. [20] J.S. Foster, C.A. Horton, Proc. R. Soc. London, Ser. B 123 (1937) 422. [21] P.F. Ruhn, J.D. Taylor, D.S. Hage, Anal. Chem. 66 (1994) 4265.