Automated albumin method underestimates pharmaceutical-grade albumin in vivo.
If purified albumin is also underestimated after infusion into the patient, this could have serious consequences both medically and economically. The present study was performed to determine whether this was the case. In addition, a gel filtration experiment was performed to see if the underestimation was caused by the presence of albumin polymers in the purified and heat-treated product, which might interfere with sample diffusion in the Vitros slide.
Serum samples from 17 postoperative and intensive care patients were used. The patients had received substantial-amounts (40-600 g; median, 180 g) of albumin by infusion of a 200 g/L solution up to 72 h prior to the assay. For comparison, serum samples from 22 patients, covering a wide range of albumin values, were obtained from the routine laboratory. These patients were verified not to have received albumin infusions. All sera were kept at -20 [degrees]C until analysis.
Serum albumin was measured with a Vitros 950 analyzer, using ALB slides (lot no. 0911-0157-6785) and with a manual BCG method performed according to the method described in the Tietz Textbook of Clinical Chemistry (1). To minimize differences caused by standardization, the Vitros Chemistry Calibrator Kit 4 (lot no. 0426) and its primary assay values for albumin were used in calibrating the manual assay. These primary assay values were the albumin values found when the calibrators were measured by Ortho-Clinical Diagnostics with a conventional BCG method on a Cobas analyzer (2). The serum albumin was also quantified by serum protein electrophoresis (SPE; Beckman Paragon SPE System; Beckman Instruments), using densitometry and the total serum protein value measured by the Vitros 950 analyzer in calculations. For comparison with a fundamentally different method, albumin was also measured immunonephelometrically on a Beckman Immage Immunochemistry System (Beckman Instruments).
The relationships obtained by linear regression analysis of the albumin results obtained with the color-binding assays for the two patient populations are shown in Fig. 1, A and B. Fig. 1A demonstrates some calibration difference but reasonable correlations between the Vitros and manual BCG methods within each patient group: for the transfused patients, Vitros = 0.76(manual BCG) - 0.72 g/L; r = 0.93; for the nontransfused patients, Vitros = 0.77(manual BCG) + 5.27 g/L; r = 0.98. However, the Vitros analyzer significantly underestimated albumin concentrations (compared with the manual BCG method) in our transfused patients, as indicated by the difference in the y-intercepts of the two regression lines; the mean difference (vertical distance between regression lines) was 6 g/L over the measured albumin range. Comparison of the Vitros and Immage results (not shown) revealed significant discrepancies at low albumin concentrations, as described earlier (3). In this comparison, the Vitros assay also underestimated the transfused patients by a mean of 6 g/L, as in the comparison with the manual BCG method. In contrast, Fig. 1B demonstrates the similarity of results in the two patient populations when albumin values from the manual BCG method are compared with the SPE results: for transfused patients, SPE = 0.79(manual BCG) + 4.5 g/L; r = 0.97; for the nontransfused patients, SPE = 0.86(manual BCG) + 3.2 g/L; r = 0.99.
The bias of the Vitros results in the transfused patients depended on the amounts of albumin given. Compared with the manual BCG values, the 7 patients receiving up to 100 g of albumin (i.e., less than the albumin content in a normal plasma volume) had significantly lower bias than the remaining 10 patients, who had received 160 g or more (bias, 5.7 and 9.1 g/L, respectively; P = 0.0009; Student two-sided t-test). These bias values include both the population difference described above and the calibration difference between methods shown by the slope of the regression formula.
It had been shown previously that highly purified commercial albumin preparations for laboratory use may contain significant amounts of albumin polymers (4, 5). The presence of such polymers in the pharmaceutical-grade preparation could possibly explain the lower values measured with the Vitros assay because they could retard transport and color development in the dry chemistry slide. To investigate this possibility, we performed a gel chromatography experiment with the purified albumin preparation on a Superdex 200 column (conditions described in legend to Fig. 1C). To ensure albumin concentrations within the working ranges of the Vitros and Immage assays in the most important fractions, we overloaded the column with material, causing a broadening of the albumin monomer peak. Nevertheless, only negligible amounts of dieters and polymers were seen (Fig. 1C). The absorbance values at 280 net (not shown) closely followed these albumin values, and polyacrylamide gel electrophoresis of the starting material confirmed the low concentration of dieters or polymers (not shown). Notably, the results obtained with the Vitros assay were much lower than those obtained with the manual BCG and Immage methods in all fractions of the albumin monomer peak that were within the working range. The ratio between the Vitros method and the two other methods seemed to change over these fractions, which we assume is related to linearity problems at these very low albumin concentrations in a nonphysiological matrix. Thus, the conclusion is that the aberrant albumin results apparently are not caused by the presence of polymers, but by a change in the albumin monomer itself.
Our findings raise several important questions and problems. The first relates to the fact that the Vitros assay cannot be used reliably to monitor serum albumin concentrations in patients receiving albumin infusions. Such use of albumin is in itself questionable in most cases (6) or even harmful (7), and studies indicate that albumin infusions usually are prompted by low albumin values and not by clinical judgment of the patients' needs (8). In cases of uncorrected hypoalbuminemia, the Vitros albumin assay often gives falsely high values because of the low albumin/globulin ratio (3). Thus, in hospitals using the Vitros assay, the practical impact of a falsely low albumin value may be especially strong.
[FIGURE 1 OMITTED]
Our experiments demonstrated the problem to be associated with all albumin preparations tested, from several suppliers. All were made from blood donor plasma by modifications of the Cohn serum protein fractionation process, followed by further purification, addition of stabilizers (N-acetyltryptophan and sodium caprylate, both at 0.08 mmol/g albumin in the preparation used here), and heat treatment. It is possible that this introduces changes that affect BCG binding by the albumin, causing a shift in spectral band in the Vitros slide but not affecting the same process in the manual BCG assay used in this study. It should be noted that in the Vitros slide, the albumin concentration remains essentially unchanged in the 10-F.i,L serum sample during color development, whereas in the manual BCG method, the serum sample is diluted 1:250 with color reagent.
If a change in the BCG-binding properties of serum albumin is the explanation, it raises the possibility that in vivo transport functions also are altered in the purified albumin preparations. To our knowledge, this important question has not been investigated to any extent, and certainly deserves further investigations.
(1.) Silverman LM, Christenson RH. Amino acids and proteins. In: Burns CA, Ashwood ER, eds. Tietz textbook of clinical chemistry, 2nd ed. Philadelphia: WB Saunders, 1994:625-734.
(2.) Albumin test methodology. Publication no. MP2-17. Rochester, NY: Johnson & Johnson Clinical Diagnostics, 1996.
(3.) Leerink CB, Winckers EK. Multilayer-film bromcresol green method for albumin measurement significantly inaccurate when albumin/globulin ratio is less than 0.8 [Letter]. Clin Chem 1991;37:766-8.
(4.) Keyser JW. Standardization of dye-binding methods for the estimation of serum albumin. Clin Chim Acta 1965;11:477-9.
(5.) Blaaberg 0, Hyltoft Petersen P. Effect of aggregates on albumin standardization. Scand J Clin Lab Investig 1979;39:751-7.
(6.) Soni N. Wonderful albumin? [Editorial]. Br Med J 1995;310:887-8.
(7.) Cochrane Injuries Group Albumin Reviewers. Human albumin administration in critically ill patients: systematic review of randomised controlled trials. Br Med J 1998;317:235-40.
(8.) Yim JM, Vermeulen LC, Erstad BL, Matuszewski KA, Burnett DA, Vlasses PH. Albumin and nonprotein colloid solution use in US academic health centers. Arch Intern Med 1995;155:2450-5.
Ole P. Bormer,  Lise Marit Amlie,  Elisabeth Paus,  and Ulf Kongsgard 
( Central Laboratory and  Department of Anesthesiology, Norwegian Radium Hospital, N-0310 Oslo, Norway; * author for correspondence: fax 47 2273 0725, e-mail firstname.lastname@example.org)
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Technical Briefs|
|Author:||Bormer, Ole P.; Amlie, Lise Marit; Paus, Elisabeth; Kongsgard, Ulf|
|Date:||Jul 1, 1999|
|Previous Article:||The use of infrared spectrophotometry for measuring body water spaces.|
|Next Article:||Thiols as a measure of plasma redox status in healthy subj ects and in patients with renal or liver failure.|