Anal. Chem. 1994,66, 4265-4271
Determination of Urinary Albumin Using High-Performance lmmunoaffinity Chromatography and Flow injection Analysis Peggy F. Ruhn, Jay D. Taylor, and David S. Hage* Department of Chemistry, University of Nebraska, Lincoln, Nebraska 68588- 0304
The detection of low levels of albumin in urine can be an important aid in the early diagnosis of kidney disease. In this study, an automated system for urine analysis was developed that could simultaneously measure both albumin and creatinine, an indicator of urine output and volume. Albumin was determinedby use of a column that contained immobilized anti-albumin antibodies. An online flow injection analysis system was used to measure creatinine as this solute eluted nonretained from the antibody column, The total analysis time of the system was 5 min per injection. Limits of detection for albumin and creatinine in a 20 p L sample (at S/N = 2) were 3 and 5 mg/L, respectively. The response for these compounds covered the range of clinical interest. Within-run precision for both analytes was f.1-2% at normal urine concentrations. The results of this method showed good agreement with those obtained by reference techniques. This approach can be adapted for the detection of other urine components by changing the type of antibody column used in the system. High-performance immunoafbity chromatography (HPIAC) is a separation technique that uses immobilized antibodies to form a selective stationary phase for sample analysis.' In HPIAC, antibodies with the ability to bind the compound of interest are immobilized or adsorbed onto a rigid, high-performance support and placed into a column. When samples are applied to this column, the analyte of interest binds to the immobilized antibodies while other components elute nonretained. By changing the column conditions, the analyte can later be eluted for quantitation or collected for further use. In previous work, it has been shown that separations based on this approach generally take only a few minutes to perform and are easy to a ~ t o m a t e . ~The - ~ specificity of this method makes it appealing for use with a variety of complex biological samples, including sen1m,3 5 mg/L > 15 mg/L > 20 mg/L > 60 g/L > 30 mg/L
so.1 g/L >0.4 g/L
>2.5g/L >0.5 g/L >o.2 g/L > 88 mg/L
All values are based on a 1L output of urine per day (range, 0.425 L per day) .18 The reference ranges for acetoacetate and pyruvate are from ref 18, the normal range for albumin is from ref 9, and all other normal ranges are from ref 32. (1
the graph. The limit of detection for albumin was determined to be 3 mg/L at a signal-to-noiseratio of two (Le,, S/N = 2). The linear range (i.e., those concentrations giving a result within f 5 % of the best-fit line) extended up to approximately 1.0 g/L HSA. The limit of detection and linear region of this curve covered the entire range of interest for the detection of microalbuminuria, or from 5 to 300 mg/L.g The upper end of the linear range for the HPIAC column occurred at approximately a 1:l load of albumin versus the available antibody binding sites. However, because there is a decrease in binding efficiency at high sample loads (see eq l), even samples in excess of the column binding capacity could produce a change in response when injected. With this particular column, kinetic effects allowed the dynamic range to extend to approximately 3 g/L HSA. A similar extension of the dynamic range beyond the known column binding capacity has been reported previously for other HPIAC system^.^,^^ A typical calibration curve obtained for creatinine on the FIA system is shown in Figure 5. A nearly linear response was seen over the entire concentration range studied. The best-fit line shown in Figure 5 has a slope of 12.8 (10.3) Wg, an intercept of 0.03 (~t0.031,and a correlation coefficient of 0.9988 for the lower six points in the graph. The limit of detection for creatinine was determined to be 5 mg/L (S/N = 2), and the dynamic range was found to extend up to at least 10 g/L. This response brackets the levels of creatinine seen in normal urine samples, which are generally said to occur between 0.5 and 2.1 g/L.32 Within-day precision of the HPIAC system was determined by making 20 sequential injections of albumin standards containing either 15 or 35 mg/L HSA The relative standard deviations determined for these samples were f0.7%and f O . 8 6 , respectively. Withiiday precision of the FIA system was determined in a similar manner by making 24 sequential injections of a normal urine control containing 0.82 g/L creatinine. The relative standard deviation obtained for this sample was f1.7%. A commercial urine control containing known levels of albumin and creatinine was used to determine the accuracy of the HPIAC/ FIA system. The results obtained by this system for three measurements of the control sample were 25 i2 mg/L albumin (31) Thomas, D.H.; Beck-Westemeyer, M. S.; Hage, D.S. Anal. Chem., in press. (32) Tietz, N. W., Ed. Clinical Guide to Laboratory Tests,2nd ed.; W. B.Saunders Co.: Philadelphia, 1990.
4270 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
(A1 SD) and 0.82 10.08 g/L creatinine. These results showed good agreement with the values provided by the manufacturer (Le., 20-30 mg/L albumin and 0.89 f 0.09 g/L creatinine). The HPIAC column described in this study was used for over 650 sample and standard injections before any signs of apparent degradation were noted (e.g., a sudden increase in column back pressure or a significant decrease in the size of retained peaks). This type of stability has been reported in earlier work with other immobilized antibody columns operated under similar elution condition^.^^^,^^ When stored in pH 7.0 phosphate buffer at 4 "C, the unpacked anti-HSA silica had no apparent loss of activity over the course of at least 6 months. This result also agrees with observations made in previous studies using silica-based immobilized antibody support^.^^^^^^ Interference Studies. Besides albumin, a number of other proteins can be found in urine. Table 1 lists some common examples of these proteins and gives their normal urine levels. The effect of these proteins and other urine components on the response of the HPIAC system was examined by injecting a normal urine sample spiked with various amounts of these solutes. The original concentration of albumin in the urine sample was 17 mg/L. The levels of each test solute that was added to this sample are summarized in the Experimental Section. The height of the retained peak obtained for each spiked sample was compared to that obtained with the same urine mixed with only pH 7.0 application buffer. The results showed that there was less than a 7%change in peak size for all of the compounds and concentrations studied. This indicated that none of these solutes had any signscant effects on the HPIAC system under the conditions tested. The results of this study agreed with SDS-PAGE experiments that examined the protein content of the retained HPIAC fraction for normal control urine. Only one protein band was observed in this retained fraction. This band comigrated with known albumin standards, thus confirming the identity of this species. A number of potential interferences were also examined for the FIA system. In serum or plasma there are several noncreatinine solutes that can give a response in the Jaffe reaction,18 but urine usually has fairly low concentrations for most of these
compound^.^*^^^ Table 1lists a number of these compounds and gives their normal urine levels. The effects of these solutes on the FIA method were examined by spiking each compound into a control urine sample which contained an initial creatinine concentration of 0.82 g/L. The levels of the test solutes that were added are given in the Experimental Section. Most of the compounds tested did not cause any detectable change in the creatinine response, but greater than 10%deviations were noted for acetoacetate, oxaloacetate,and a-ketoglutarate at levels above the cutoffs listed in Table 1. However, in all cases the concentrations of the potential interferencesat which these deviations began to occur were much higher than those seen in typical urine samples.
techniques is that the same antibody population can be reused for multiple analyses. In this particular experiment, the same antibody column was used for over 650 sample and standard injections. This factor, along with the high precision of most HPLC systems, is probably what helps to produce the small withinrun variation noted for the HPIAC method. One advantage of the HPIAC/FIA system is its ability to provide a direct measurement of albumin in samples. This type of measurement does not require the use of any labeled compounds for detection, such as those used in radioimmunoassays or enzyme immunoassays. Besides reducing the number of required reagents, this approach helps to decrease the analysis time because fewer steps are needed for reagent addition, sample washing, and signal production. Problems associated with label stability and handling, especially when radioisotopes are used, are CONCLUSION In this study, we examined the use of high-performance also eliminated by this approach. immunoaftinity chromatographyand flow injection analysis for the The ability to use HPIAC along with an FIA assay for creatinine determination of albumin and creatinine in urine. The final system simplifies the measurement of urinary albumin because it allows had an analysis time of 5 min for both analytes and required only the albumin results to be automatically normalized for changes 20 pL of sample per injection. This sample size is comparable to in urine volume. This means that short-term urine samples, as that used in many previous methods for urinary a l b ~ m i n . ~ ~ J ~ J well ~ J ~as 12 or 24 h specimens, can be processed and used to obtain However, the overall analysis time is much shorter than the quantitative results. By using different types of HPIAC columns, periods of 1-16 h reported for earlier manual method^'^-'^ and this same approach could be adapted for use in the measurement is slightly shorter than the time required by automated techniques of other solutes in urine. Possible applications include the based on immunoturbidimetry.lOJ1 determination of other urinary proteins (e.g., hemoglobin and Calibration curves obtained on the HPIAC/FIA system cover transferrin) or the measurement of excreted drugs and their the range of clinical interest for both albumin and creatinine. The metabolites. within-day precision of this method for albumin in normal urine samples (f0.7-0.8% at 15-35 mg/L HSA) is smaller than that ACKNOWLEDGMENT reported for either manual methods (5-9%)14J5 or previous This work was supported by the National Institutes of Health automated techniques for this analyte (2.3-4.8%).loJ1 In addition, under Grant GM44931. P.F.R was supportedby a fellowship from the results of the HPIAC/FIA technique show no major interferthe University of Nebraska Center for Biotechnology. The authors ences from many common solutes and proteins found in urine thank John Swart and Teresa Urlacher for their assistance in the samples. SDS-PAGE studies. Like HPIAC, many current techniques for urinary albumin require the use of antibodies for the selective determination of Received for review July 18, 1994. Accepted September HSA in samples. Examples include methods based on immuno2, 1994. turbidimetry,1°-12 enzyme immunoassays,13J4and radioimmunoa ~ s a y s . ~One ~ J ~general advantage of HPIAC over these other @Abstractpublished in Advance ACS Abstracts, October 1, 1994.
Analytical Chemistty, Vol. 66, No. 23, December 1, 1994
4271
