Quantification of albumin in cerebrospinal fluid

ANALYTICAL

BIOCHEMISTRY

175,300-304

Quantification

( 1988)

of Albumin in Cerebrospinal

Fluid

PETERL. BONATE’ 1615 Bleasner Dr. SE #40, Pullman, Washington 99163 Received January 2 I,1988 A standard spectrophotometric method for the quantitation of urinary albumin using bromphenol blue is evaluated to determine whether this method could be used to quantify cerebrospinal fluid albumin. In the modified procedure, 200 ~1 of sample is added to 3 ml of bromphenol blue solution and the absorbance is read spectrophotometrically at 6 10 nm. Using standards and controls, the results were compared with known values and found to be both precise and accurate. The bromphenol blue method was compared with an immunoturbidometric method and found to be more precise and accurate, easier to perform, and cost effective.When compared to other dye-binding methods the bromphenol blue method is unique in its extremely low linear range and limit of detection. A minor disadvantage was the increased sample size necessary to obtain the increased precision. 0 1988 Academic Fwss, Inc. KEY WORDS: albumin; cerebrospinal fluid; bromphenol blue; immunoturbidometry; spectrophotometry; blood-brain barrier permeability.

Electrophoresis of cerebrospinal fluid (CSF) shows that the major band seen is albumin. Albumin is the second fastest moving protein (the fastest being prealbumin) with a molecular weight of 69,000 and comprises 50-70s of the CSF total protein (1). Serum concentrations range from 3.5 to 5.0 g/dl; however, due to the presence of the blood-brain barrier (BBB),’ the concentration of albumin in CSF is only 10 to 30 mg/dl(2). Current methodologies for the quantitation of CSF albumin include nephelometry/immunoturbidometry (ITM), radial-immunodiffusion (RID), and electroimmunodiffusion. Some researchers are of the opinion that it is within the state of the art to adapt many of the known protein determination methods to CSF analysis. While this may be true for antigen-antibody methods, this is not the case for dye-binding ’ Present address: Department of Pharmacology/Toxicology, Washington State University, Pullman, Washington 99163. ’ Abbreviations used: BBB, blood-brain barrier; ITM, immunoturbidometry; RID, radial-immunodiffusion; BCG, bromcresol green; BCP, bromcresol purple; BPB, bromphenol blue. 0003-2697/88 $3.00 Copyright 0 1988 by Academic Prw, Inc. All rights of reproduction in any form resawd.

methods. Bromcresol green (BCG) is the method for serum albumin determination recommended by the American Association for Clinical Chemistry (3). Kaplan and Pesce (4) recommend the bromcresol purple (BCP) method. However, the lower limit of linearity of the BCG and BCP methods are 1000 and 500 mg/dl, respectively (4,5). These methods are obviously inadequate for CSF albumin determination. Use of a dye-binding method for the selective determination of CSF albumin is unique among today’s methodologies. This procedure was originally outlined by Schosinsky et al. (6) for the determination of urinary albumin. The BBB protects the internal environment of the brain. The BBB is formed by the continuous merging of capillary cells in the brain creating what are called “tight junctions.” Surrounding the capillary cells are astrocyte foot processes which, when combined with tight junctions, form a barrier between the systemic circulation and the brain (7). The BBB should be viewed as more of a concept than a discrete anatomic feature. Compounds that (a) have a molecular weight of 300

SPECTROPHOTOMETRIC

QUANTITATION

greater than 60,000, including albumin, (b) are lipophobic, and (c) are ionized at physiological pH tend to remain in the general circulation and do not cross the BBB (8). Tourtellote estimated that only one albumin molecule out of 230 crosses the BBB (9). However, the concentration of albumin in CSF varies with serum concentrations and various disease states. Since albumin is synthesized in the liver and not to any significant extent in the central nervous system, the ratio of CSF albumin (in mg/dl)/serum albumin (in g/dl), coined the albumin index, produces a useful measure for the integrity of the BBB. Any increase in the index indicates increased BBB permeability. An intact BBB has an index value of less than 9. Moderate impairment is interpreted as having an index of 14-30, while any value greater than 30 is interpreted as severe impairment ( 10). Demonstration of increased BBB permeability in itself is not a complete diagnosis but information which must be combined with other clinical data to produce a diagnosis. Clinical conditions associated with increased bloodbrain barrier permeability include multiple sclerosis, encephalopathy, head trauma, and Guillan-Barre syndrome (11). Accurate and rapid determination of CSF albumin is necessary to detect any subtle changes in BBB integrity. MATERIALS

AND METHODS

Equipment. A Beckman DU-40 spectraphotometer was used for all absorbance readings. Oxford precision pipets were used for pipetting. All samples were pipetted and mixed in 12 X 75-mm disposable glass test tubes (Scientific Products T 1290-3). Reagents. Glycine buffer (230 mmol/liter, pH 3.0): In a 1-liter volumetric flask dissolve 17.26 g glycine in about 800 ml distilled water. Add 100 mg sodium azide as a preservative. Adjust pH to 3.0 using 6.0 N hydrochloric acid. Dilute to 1 liter using distilled water. Working bromphenol blue solution: In a l-liter volumetric flask dissolve 125.7 mg bromphenol blue (Aldrich 11439-l) in 3 ml

OF ALBUMIN

301

0.1 N sodium hydroxide. Dilute to about 150 ml using distilled water. Add 800 ml glycine buffer. Add 4 ml Brij 35 and dilute to 1 liter using glycine buffer. The reagent is stable for at least 6 months when stored at room temperature. Standards. Serum standards (Gilford Automated Reference Serum: No. 606-298-2 1, Lot No. 037602) and unassayed controls (Serachem Clinical Cbemistry controls, Unassayed Level I and II: No. 3 I 1O-34,3 1 l-34) were diluted 1: 100. These standards were compared by electrophoresis with actual cerebrospinal fluid and found to mimic the spinal fluid in total protein and globulin fractions. The mean values for the unassayed, undiluted controls were obtained over a period of months using a bromcresol green method performed on an SMA-20. Aqueous standards were obtained (Sigma microalbumin standards: No. 6 lo- 11) with concentrations of 15, 30, and 50 mg/dl. Method. Three milliliters of BPB solution was pipetted into a 12 X 75-mm test tube in triplicate. Two hundred microliters of sample was pipetted into each test tube and the absorbance was measured at 30 + 3 s at 6 10 nm. The concentration of the unknown was determined by the expression: Concentration,

mg/dl

mean ABS unknown = mean ABS standard X concentration

of standard.

The standard used in all cases was the Gilford Reference Serum with a known value of 36 mg/dl. The instrument blank was the BPB solution. RESULTS

Sample size.This procedure was modified slightly for two reasons. The original method calls for 100 ~1 of sample to be used because the concentration of urinary albumin in 100 ~1 of sample was found to have the longest linear range. As sample size is increased the

302

PETER L. BONATE TABLE 1 DIFFERENCEINVARIANCEOFABSORPTIONUNITSUSING~OOAND~~~~~OFSAMPLESIZE

Sample

Variance lOOpI

Level I Level II Standard

8.94 15.65 24.10

g

Variance 200 pl

cv (%)

Percentage difference

F value

3.24 6.25 12.30

1.70 1.30 1.90

63.80 60.10 49.00

2.75~ 2.50* 1.96*

5.96 5.20 6.50

* Significantly different at 10% critical level.

variance in absorbance units was found to decrease drastically (Table 1). A one-tailed F test performed on the variances of the absorption units indicated that the variance using 200 ~1 of sample is considerably less than the variance using 100 ~1 of sample at a critical level of 0.10. Results using aqueous standards. A comparison of the known values for all three standards and the calculated values show no significant difference between the mean (P < 0.001) (Table 2). Using the BPB method the 30 mg/dl standard had a calculated mean value of 30 mg/dl, a standard deviation of 0.8 mg/dl, and a coefficient of variation of 2.7% (n = 26). The 15 mg/dl aqueous standard had a calculated mean value of 16 mg/dl, a standard deviation of 0.9 mg/dl, and a coefficient of variation of 5.6% (n = 26). The 50 mg/dl standard had a calculated mean of 50 mg/dl,

a standard deviation of 1.5, and a coefficient of variation of 3.0% (n = 10). Results using controls. The Serachem controls are multiple analyte controls. The results obtained using these controls might be influenced by other components present. However, the results obtained using the Serathem controls do not differ significantly from the true means (Table 2). The range of the Level I control was 19-24 mg/dl. The calculated value was 2 1 mg/dl with a standard deviation of 0.4 mg/dl. The difference between the known value and the calculated value was not significant (P < 0.01). The range of the Level II control was 35-4 1 mg/dl. The calculated result was 37 mg/dl with a standard deviation of 0.5 mg/dl. Again the differences were not significant. The coefficients of variation (CV) for the Level I and Level II controls were 1.6 and 1.3%, respectively. When the se-

TABLE 2

Observed Sample Serachem Level I Sexachem Level II Sigma 50 mg/dl 30 mg/dl 15 m&U

Range or true mean (mtidl)

Mean N

SD hw/dl)

95% Confidence interval hw/~)

P value

SPECI-ROPHOTOMETRIC

QUANTITATION

303

OF ALBUMIN

TABLE 3 COMPARISON

OF RESULTS BETWEEN THE BPB AND ITM

Bromphenol blue

Immunoturbidometric

Calculated Aqueous standard 50 30 15

Mean b-&W

N

50 30 15

10 26 26

Mean bg/dl) 50 30 16

METHOD

Calculated SD N 1.5 0.8 0.9

rum standards are compared to the aqueous standards, the former are more precise.

12 28 18

Mean h/dU 51 29 14

SD 4.4 3.1 1.3

CV (B)

P value

7.7 9.8 9.9


DISCUSSION

For increased precision 200 ~1 is the rec-

Comparison of results between the BPB ommended sample size when sufficient samand ITM method using aqueousstandards. A ple is available. A sample size of greater than comparison of the results between the ITM method and BPB method using aqueous standards are shown (Table 3). The ITM method is significantly less accurate than the BPB method. The results using the 50-mg/dl standard show the greatest discrepancy. The other major difference is the precision using the ITM method. The ITM method demonstrated a CV of 7.7,9.8, and 9.9% for the 50-, 30-, and 15-mg/dl standard, respectively. The average CV using the BPB method was 3.8%, a difference of approximately 240%.

Linear regression of the calibration curve. By plotting the known vs calculated values a calibration curve was obtained. Statistical analysis of the curve using least-squares linear regression shows that the BPB method follows the equation Y = 0.9758X + 0.6468. A correlation coefficient of 0.9988 was also obtained from the curve. The 95% confidence interval for slope and intercept was (0.8873, 1.064) and (-2.295, 1.064), respectively. The standard error of the slope and intercept was 0.0278 mg/dl and 0.9252 mg/dl, respectively. Based on the equation of the line and the deviation of the calculated values given the observed values the 99.7% (3 SD’s) limit of detection for the BPB method was found to be 2 mg/dl.

200 ~1 was considered untenable for routine use. The drawbacks of the 200~~1 aliquot are that more sample is needed, and the linear range is decreased. The maximum absorbance unit using 200 ~1 of sample decreases from 0.900 to 0.600. Schosinsky et al. (6) calculate a linear range of 0.000 to 0.600 absorbance units using 200~~1 sample with 3.0 ml BPB. The Sigma aqueous standards are single analyte standards and as such any change in absorbance is due entirely to the quantity of albumin present. This type of standard was used to check the validity of the BPB method. From the results using aqueous standards, the BPB method can be assumed to be accurate throughout the entire range of normal up to an elevated level of 50 mg/dl. The BPB method also has an average CV of less than 5%, which represents a significant step toward improved precision in the clinical laboratory. The data presented using serum controls suggest that other analytes present, i.e., globulins and lipoproteins, do not interfere with the BPB method. The disparity in the mean values obtained using the BPB method vs the ITM method can be explained by either of two assumptions: (1) the error lies with the ITM method

304

PETER L. BONATE

or (2) the error lies with the BPB method. Assuming the ITM method is to blame, one plausible explanation is the ITM method requires several dilutions to obtain a calibration curve while there is only one dilution for the CSF sample. The discrepancy in dilutions is then compensated by division of the result by some factor. In addition, each dilution represents an error which will increase the inaccuracy of the method, i.e., the ITM method is prone to propagation of errors. Another plausible explanation is the assumption that the calibration curve created by the ITM method is always linear. This is not always the case. If the calibration curve is not linear, then the accuracy of the results will be off. If, however, we assume the BPB method is to blame then the BPB method must have an interferent present which constructively interferes with the absorption spectrum of the BPB complex. This question was answered through standard additions. In an independent experiment, a random sample was analyzed using the BPB method. The sample was diluted 1: 1 with 15, 30, and 50 mg/dl standard. The dilutions were then analyzed using the BPB method. The change in absorbance was then back calculated and found to match the original result. Thus, it is safe to assume the error lies with the ITM method. A major disadvantage of the BPB method is that of sample size. In practice, the clinical laboratory is asked to perform multiple assays using a limited quantity of CSF. The ITM method requires about 70 ~1 of sample to run in duplicate while RID methods require about 35 ~1 to run in duplicate. The BPB method requires about 400 ~1 of sample to run in duplicate. While this may seem to be a great amount, it is not a prohibitive one. If the total protein of the sample is suspected to be grossly elevated, as in a CNS disorder such as Guillan-Barre Syndrome, the sample volume can be decreased and still achieve sufficient absorbance readings without a significant loss of precision.

A major advantage of the BPB method is its ruggedness. The BPB method is easy to perform. Other methods, especially ITM, are difficult for the average technician to perform as evidenced by the higher CV and variance. Another advantage of the BPB method is the time required for analysis. ITM method requires about 1 h to complete whereas RID requires about 8-24 h. Using the BPB method the actual run time is decreased to a matter of minutes, saving time and money. Reagent costs are decreased since the BPB method does not require the expensive monoclonal antibodies the ITM and RID methods do. The BPB working solution is also very inexpensive. In summary, the advantages of the BPB method merit its adoption by clinical laboratories. ACKNOWLEDGMENT This research was made possible by a grant from the University of Nevada, Las Vegas, Graduate School Association.

REFERENCES 1. Killingsworth, L. M. (1983) High Resolution Protein Electrophoresis: A Clinical Overview with Renal and Neurological Case Studies, p. 9, Helena Laboratories, Beaumont, TX. 2. Davidsohn, I., and Henry, J. B., Eds. (1974) ToddSanford: Clinical Diagnosis by Laboratory Methods, p. 1392, Saunders, Philadelphia. 3. Doumas, B. T., and Biggs, H. G. (1972) Standard Methods of Clinical Chemistry, Vol. 7, pp. 175188, Amer. Assoc. Clin. Chem. 4. Pesce, A., and Kaplan, L. (1987) Methods in Clinical Chemistry, p. 107 1, Mosby, St. Louis. 5. Tietz, N., Ed. (1986) Textbook of Clinical Chemistry, p. 599, Saunders, Philadelphia. 6. Schosinsky, K. H., Vargas, M., Esquivel, A. L., and Chavarra, M. A. (1987) Clin. Chem. 33,223-226. 7. Goldstein, G. W., and Lorris Betz, A. (1986) Sci. Amer. 255( lo), 74-83. 8. Craig, C. R., and Stitzel, R. E., Ed. (1982) Modem Pharmacology, pp. 444-445, Little, Brown, Boston/Toronto. 9. Tourtellote, W. W. (1970) J. Neural. Sci. 10, 279304. 10. Tie& N., Ed. (1986) Textbook of Clinical Chemistry, p. 609, Saunders, Philadelphia. 1 I. Pesce, A., and Kaplan, L. (I 987) Methods in Clinical Chemistry, p. 729, Mosby, St. Louis.