Comparison of Manual, Automated, and "Kit" Methods for the Measurement of Glucose. and Urea in Plasma

Comparison of Manual, Automated, and "Kit" Methods for the Measurement of Glucose and Urea in Plasma ROBERT C. ELSER, B.S., AND JOHN SAVORY, PH.D. I'ulhology Department, University of Florida College of Medicine, Gainesville, Florida 32601 ABSTRACT Elser, Robert C, and Savory, John: Comparison of manual, automated, and "kit" methods for the measurement of glucose and urea in plasma. Amer. J. Clin. Path. 54: 820-827, 1970. An evaluation of a commercial kit for determination of plasma urea nitrogen and glucose is presented. Close correlations were found among manual, automated, and kit urea methods. The kit and manual glucose methods also correlated closely, and both gave values significantly lower than the values given by the automated glucose procedure. The precision of every method is better than 6%. THE DEVELOPMENT of simplified methods for the measurement of blood constituents has led to the commercial marketing of kits to perform these determinations. The aim of the kit manufacturers has been to aid technical personnel in the small laboratory or physician's office by offering convenient means for performing common clinical laboratory tests. The service provided by the manufacturers has been to eliminate preparation of reagents and standards and to provide the utmost convenience and simplicity for performing the test. However, sound analytical principles must not be discarded in favor of this convenience and simplicity. The present study was undertaken to evaluate a kit which, in the authors' opinion does not violate sound analytical practice. The kit can be used to measure blood Received August 1, 1969; accepted for publication April 6, 1970. Doctor Elser's present address is: Department of Chemistry, 111 Leigh Hall, University of Florida, Gainesville, Florida 32601. Supported by National Institute of Health Training Grant GM-01142-05. 820 glucose, urea nitrogen, cholesterol, total protein, uric acid, bilirubin, and hemoglobin. For the purposes of this initial evaluation, glucose and urea nitrogen were measured and the results were compared with the results of frequently-used clinical laboratory procedures. Many methods for quantitative determination of glucose and urea in plasma have been proposed and reviewed. 11 Glucose may be determined by several methods, 3 ' 4-7, s, i2, i4, io, is, io anci there are three general methods 5 > B - 15 ' 17 > 19 for the measurement of urea. These methods can be divided into enzymatic 3 > 15 > 1B and nonenzymatic. 4 " 8-12 > 14 Enzymatic methods generally are specific, but they often have disadvantages such as high cost, 13 instability of reagents, nonspecific interferences in coupled reactions, 1-10 and impure enzyme preparations. 2 Colorimetric technics, on the other hand, may not be as specific but are usually less expensive and less time consuming than enzymatic methods. Consequently, colorimetric methods often are chosen in preference to enzymatic technics for routine use in

December 1970 MEASUREMENT OF GLUCOSE AND UREA IN PLASMA 821 busy laboratories. In this study the most frequently used nonenzymatic colorimetric methods were compared with the commercial kit. Three methodologies each for measuring glucose and urea in plasma have been studied and include continuous How automated methods, 17 " 19 manual methods, 6 ' 8 and commercial kit methods. Materials Apparatus Reagents for the kit procedures are packaged in screw-cap tubes by the manufacturer (Diagnostest, Dow Chemical Company, Indianapolis, Ind.). The tubes serve as both reaction vessels and cuvettes. A heating block supplied with the kit accommodates 12 tubes. In our laboratory, absorbance measurements are made with a Coleman Junior II spectrophotometer fitted with a cuvette adapter designed to accommodate the special kit cuvettes. A filter colorimeter is available from the kit manufacturer. A Technicon AutoAnalyzer (Technicon Company, Ardsley, N. Y.) is used for the automated methods. Reagents 1. Kit reagents for the measurement of plasma glucose and urea nitrogen are supplied by the manufacturer. The glucose reagent is a 9% solution of o-toluidine in glacial acetic acid, contained in the cuvette tube. Two reagents for the determination of urea are supplied. These reagents, packaged in a cuvette tube, consist of 1.6 N sulfuric acid, 2.9 N phosphoric acid, 0.005% (w/v) thiosemicarbazide, 0.026 mg. of urea per 100 ml. of solution, and a separate reagent bottle containing 2% (w/v) diacetylmonoxime. Standard solutions of glucose and urea are provided with the kit. 2. Reagents for the continuous flow automated procedures are those recommended by the Technicon Corporation for using the N-16b methodology for simultaneous measurement of plasma glucose and urea. 3. o-toluidine, 6.0% (v/v) in glacial acetic acid containing 0.15% (w/v) of thiourea as a preservative. 8 Reagent grade materials are used, and the final solution is stored in an amber bottle. 4. Stock diacetyl monoxime reagent." About 500 ml. of distilled water are added to a 1-liter volumetric flask; 10 Gm. of diacetyl monoxime and 0.3 Gm. of thiosemicarbazide are added and mixed until dissolved. Distilled water is then added to the mark. 5. Arsenic acid reagent. 6 Into a 2-liter Erlenmeyer flask are placed 200 ml. of distilled water, and 300 ml. of concentrated sulfuric acid are added slowly with mixing. Arsenic pentoxide, 20 Gm., is added, and the mixture is stirred on a hotplate until solution is complete. The solution is cooled, and distilled water is added to a volume of 2 liters. 6. Glucose standard solution. In 100 ml. of a saturated aqueous benzoic acid solution are dissolved 250 mg. of anhydrous reagent grade glucose. This stock standard is diluted further with the benzoic acid solution to obtain standards of 25, 50, 75, 100, 125, 150, 175, 200, 225, and 250 mg. of glucose per 100 ml. 7. Urea standard solution. In 100 ml. of 0.100 N hydrochloric acid, 100 mg. of urea (National Bureau of Standards standard reference material #912) are dissolved. This standard is diluted further with the hydrochloric acid solution to obtain standards of 5, 10, 15, 20, 25, 30, 40, 50, 60, and 70 mg. of urea per 100 ml. Methods Glucose 1. Kit method: Capillary micropipettes calibrated to contain 100 p\. are supplied with the kit and are used to measure plasma samples and standards. The micropipettes containing the samples are dropped into the glucose reagent solution, which is packaged in the screw-capped cuvette tube,

822 ELSER AND SAVORY A.J.C.P. Vol. 54 1.000.800 ".600.200 T 0 50 100 150 200 250 GLUCOSE (mg/looml) FIG. 1. Calibration curves for measurement of glucose. and mixing is accomplished by inverting the tube several times. The tubes are heated at 100 C. for 10 min., cooled, and the absorbance of the solution is measured at a wavelength of 595 nra. The micropipette is adhered by surface tension to the top of the cuvette tube during absorbance measurements and thus is removed from the light path. Concentrations of unknown samples are calculated from standards by UREA NITROGEN Imq/lODmll Fic. 2. Calibration curves for measurement of urea. relating absorbance after appropriate blank corrections. 2. Manual method: This procedure is a slight modification of the method described by Feteris. 8 In the present method, 50 pi. of sample are added to 4 ml. of 6% o-toluidine reagent instead of the 100 pi. added to 5 ml. described by Feteris. 3. Automated method: This automated procedure involves the oxidation of glucose by potassium ferricyanide and is described in the method file published by the manufacturer of the automated instrument (Technicon AutoAnalyzer). 19 Glucose and urea are determined simultaneously. Urea 1. Kit method: Using a calibrated reagent dropper, 0.5 ml. of the diacetylmonoxime reagent is added to each vial and mixed. Twenty microliters of sample or standard are added to the vials, which are heated at 100 C. for 12 min., cooled, and the absorbances measured at 520 nm. Concentrations of unknown samples are calculated by relating absorbances with standards after appropriate corrections for blanks. 2. Manual method: The method is as described by Sherrick and de la Huerga. 6 The color reagent is prepared immediately before the test by mixing one volume of diacetyl monoxime reagent with five volumes of arsenic acid reagent. Six ml. of color reagent are transferred to the test tubes and 10 pi. of sample or standard are added and mixed. The tubes are heated at 100 C. for 12 min., cooled, and the absorbances read at 520 nm. Concentrations of unknowns are calculated as described above for the kit method. 3. Automated method: This procedure involves the reaction between urea and diacetylmonoxime in the presence of acid and ferric ion, and is described in the Auto- Analyzer method file. 19

December 1970 MEASUREMENT OF GLUCOSE AND UREA IN PLASMA 823 Results Standard Curves The standard curves for the determination of glucose in plasma by the kit, manual o-toluidine, and automated procedures are shown in Figure 1. The manual and kit methods obeyed Beer's law to 250 mg. per 100 ml. The automated procedure, which is a color reduction method, was linear to 225 mg. per 100 ml. Standard curves for the determination of urea in plasma are shown in Figure 2. The automated procedure gave a concave curve, whereas the manual procedure was linear to a concentration of 70 mg. of urea nitrogen per 100 ml. The curve for the kit method was linear to 20 mg. per 100 ml., at which concentration the slope of the curve changed. The kit manufacturers acknowledge in their instructions this nonlinearity of the standard curve, and they recommend using two standard solutions of urea in the test. One standard solution has a concentration of approximately 20 mg. per 100 ml., and the other standard is equivalent to 50 mg. per 100 ml. The manufacturers claim that the maximum error incurred by the nonlinearity of the standard curve will be 5% low, provided the two standards are used as recommended. The manufacturers' instructions were observed during the present study and, as the Method Glucose Kit Manual Automated Urea Kit Manual Automated Table 1. Precision of Duplicate Analyses No. of Tests 48 48 48 50 50 50 Mean (mg-/ 100 ml.) 121 113 142 29 31 33 SD (mg./ 100 ml.) ±5.3 ±5.7 ±1.8 ±1.1 ±1.5 ±0.7 C.V. (%) ±4.4 ±5.0 ±1.3 ±3.8 ±4.8 ±2.1 correlation studies indicate, the error created by the nonlinear standard curve was minimal. Precision Precision of the methods was determined by performing duplicate analyses of plasma samples. The standard deviations calculated from the differences between duplicate values {*>- $) and coefficients of variation / SD X 100 \ Mean ) are summarized in Table 1, Table 2. Comparison of Methods for Measurement of Plasma Glucose and Urea Methods Compared Standard Error of Estimate (mg./loo ml.) Correlation Coefficient Student's /* for Comparison of Means F Valuef for Comparison of Precisions Glucose Kit vs. manual Kit vs. automated Automated vs. manual Urea Kit vs. manual Kit vs. automated Automated vs. manual 8.0 10.9 11.9 2.5 2.5 1.9 0.972 0.946 0.962 0.981 0.980 0.993 0.818 2.018 3.124 0.298 0.751 0.442 1.157 8.670 10.028 1.860 2.469 2.592 * Critical t for comparison = 2.015 for significance (95% probability), t Critical F for comparison = 1.610 for significance (95% probability).

824 ELSER AND SAVORY A.J.C.P. Vol. 54 compared favorably with the coefficients o variation for the methods listed in Table I. The F-ratio test for comparing the precision of different sets of data is often used. 11 F values calculated from the precisions of the blind duplicate experiments are summarized in Table 2. It may be seen that the degrees of precision of the manual and kit glucose methodologies were nearly identical, whereas both were significantly less accurate than the automated technic. The kit method for determining urea nitrogen was significantly better than the manual method, although again both were inferior to the automated procedure. 25 50 75 100 125 150 175 GLUCOSE (mg/loomll o - TOLUIDINE MANUAL PROCEDURE Fic. 3. Comparison of manual and kit glucose procedures. A commercially available control serum was employed over a period of 20 days in the estimate of daily variation. The day-today precision of 4.5% for glucose and 5.3% for urea nitrogen (68% confidence limits) Recovery Values for recovery of glucose added to five normal serum samples in concentrations of 50 and 100 mg. per 100 ml. were 95.2 ± 2.7% and 97.2 ± 5.7%, respectively. Recoveries of 20 and 40 mg. per 100 ml. of urea nitrogen added to the same samples were 95.6 ± 3.5% and 96.4 ± 3.7%, respectively. FIG. 4. Comparison o automated and kit glucose procedures. GLUCOSE ImgTIOOmt) FERRICYANI DC AUTOMATED PROCEDURE

December 1970 MEASUREMENT OF GLUCOSE AND UREA IN PLASMA 825 FIG. 5. Comparison of manual and 5 125 automated glucose procedures. 5 50 75 100 125 150 175 200 GLUCOSE lmg/100ml) 0 - TOLUIDINE MANUAL PROCEDURE Comparison of Methods A comparison was made of the results of the kit, manual, and automated methods for determining plasma glucose and urea. Samples were collected at random from among freshly drawn (same day) blood specimens which had been sent to the laboratory for either glucose or urea nitrogen determinations. These specimens were preserved and anticoagulated with sodium fluoride and mixed oxalate. No identification by patient number was made prior to analysis to avoid bias from knowledge of previous results. Each sample was divided into two parts and coded so that the analyst did not know which code represented duplicate samples. All samples were treated as individual unknown samples and each was analyzed by the kit, manual, and automated methods with a time period not exceeding two hours. Each of the three methods for measuring glucose and urea was compared with the other two methods. The correlations are illustrated graphically in Figures 3 to 8. The broken line indicates the calculated regression line and the solid line represents the theoretical relationship of x = y. The correlation coefficient, standard error of estimate, and Student's t value for comparison of means are given in Table 2. 10 20 30 40 50 UREA NITROGEN Img/lOOml) DIACETYLMONOXIME MANUAL PROCEDURE FIG. 6. Comparison of manual and kit urea procedures.

826 ELSER AND SAVORY A.J.CP. Vol. 54 ee I I I I I I = 10 20 30 40 50 60 UREA NITROGEN ImgMOOml) OIACETYLMONOXIME AUTOMATED PROCEDURE FIG. 7. Comparison of automated and kit urea procedures. Discussion Plasma glucose and urea are frequently determined by the clinical chemistry laboratory. Requests for these tests both on a daily routine basis and in emergency situations require many laboratories to employ UREA NITROGEN ImgMOOmll DIACETYLMONOXIME MANUAL PROCEDURE Fic. 8. Comparison of manual and automated urea procedures. two procedures for each test. One procedure, for the routine tests, is generally automated. The other procedure, available for the emergency tests, should fulfill the same criteria of accuracy and precision as the routine analysis while still being rapid, easily performed, and yielding comparable results. Commercial kits for the more common tests offer packaged reagents, and occasionally offer pipettes, colorimeters, and standards. The manufacturers aim their appeal to the physician's office or small laboratory where technical personnel do not possess the expertise in analytical biochemistry necessary for preparing and maintaining high quality reagents and standards. Often large laboratories have difficulty in recruiting capable technologists for night and weekend emergency coverage. Thus, a kit manufacturer can offer a valuable contribution to clinical chemistry provided his product utilizes accurate and precise methods and contains high quality components. We have evaluated such a kit and compared results for plasma glucose and urea with those of automated routine and manual emergency procedures which have provided reliable results in our laboratory for several years. The precision of the kit procedures was similar to that of our manual procedures; however, the automated methods were much more precise. Standard curves for the kit and manual methods followed Beer's law up to high concentrations of glucose. There was a slight deviation from linearity with the kit urea method in contrast to the linear standard curve obtained using the manual urea procedure. A close correlation between the three urea methods, all of which use the same chemical reaction for color development, was found. The manual and kit methods for glucose measurement correlated well with each other, but both gave significantly lower values than the automated procedure. The o-toluidine reaction used in the

December 1970 MEASUREMENT OF GLUCOSE AND UREA IN PLASMA 827 kit and manual methods has been shown 8 to give results closely corresponding to those of the Nelson-Somogyi method, long recognized as specific for glucose. Thus, the ferricyanide reduction method apparently measures other non-glucose reducing substances; this observation confirms the findings of Fingerhut." We conclude that for plasma glucose and urea the kit tested provides reliable results which correlate closely with results of well established procedures. References 1. Aberg, B.: Interference of light on the determinations of low glucose concentrations with glucose oxidase. Acta Physiol. Scand. 71: 186-193, 1967. 2. Barton, R. R.: A specific method for quantitative determination of glucose. Anal. Biochem. 14: 258-260, 1966. 3. Beach, E. F., and Turner, J. J.: An enzymatic method for glucose determination in body fluids. Clin. Chcm. 4: 462-475, 1958. 4. Brown, M. E.: Ultra-micro sugar determinations using 2,9-dimethyl-l,10-phenanthroline hydrochloride (neocuproine). Diabetes 10: 60-62, 1961. 5. Chaney, A. L., and Marbach, E. P.: Modified reagents for determination of blood urea nitrogen, with special reference to automatic colorimetric analysis. Clin. Chem. 5: 617-620, 1959. 6. de la Huerga, J., Sherrick, J. C, and Petrus, E. A.: Measurements of urea in serum and urine. In Sunderman, F. W., and Sunderman, F. W., Jr. (eds.): Laboratory Diagnosis of Kidney Disease. St. Louis, W. H. Green Co. (in press). 7. Dubowski, K. M.: An o-toluidine method for body-fluid glucose determination. Clin. Chem. 8: 215-235, 1962. 8. Feteris, W. A.: A serum glucose method without protein precipitation. Amer. J. Med. Techn. 31: 17-21, 1965. 9. Fingerhut, B.: Automated serum glucose levels in uremia. Amer. J. Clin. Path. 51: 157-160, 1969. 10. Free, A. H.: Enzymatic determinations of glucose. In Sobotka, H., and Stewart, L. P. (eds.): Advances in Clinical Chemistry. Vol. 6. New York, Academic Press, 1963, pp. 67-96. 11. Henry, R. J.: Clinical Chemistry Principles and Technics. New York, Harper and Row, 1965, pp. 137, 625-654. 12. Johnson, J.: Blood glucose, protein free filtrate or dialysate. Amer. J. Med. Techn. 24: 271-280, 1958. 13. Mitchell, T., and Rydalch, V.: An evaluation of a true glucose procedure utilizing the enzyme hexokinase. Amer. J. Clin. Path. 50: 401^02, 1968. 14. Nelson, N.: A photometric adaptation of the Somogyi method for the determination of glucose. J. Biol. Chem. 153: 375-380, 1944. 15. Richter, H. J., and Lapointe, Y. S.: A simple method for the determination of blood urea nitrogen, with special reference to automatic colorimetric analysis. Clin. Chem. 5: 617-620, 1959. 16. Schmidt, F. H.: Die enzymatische bestimmung von Glucose und Fructose nebeneinander. Klin. Wschr. 39: 1244-1247, 1961. 17. Technicon Method File N-la. Ardsley, N. Y., Technicon Instruments Corporation, 1963. 18. Technicon Method File N-2b, Ardsley, N. Y., Technicon Instruments Corporation, 1965. 19. Technicon Method File N-16b. Ardsley, N. Y., Technicon Instruments Corporation, 1967.