Solid-phase immunofluorometric assay for quantification of CSF immunoglobulins

Journal of the Neurological Sciences, 1990, 96:229-240

229

Elsevier JNS 03319

Solid-phase immunofluorometric assay for quantification of CSF immunoglobulins Determination of normal reference values Tsuneyuki Takeoka 1, Yukito Shinohara 1, Kosuke Mori 2 and Koichi Furumi3 1Department of Neurology, Tokai University School of Medicine, Kanagawaken (Japan), 2Department of Neurology, Tachikawa-KyosaiHospital, Tokyo (Japan), and 3Department of Hygiene, Kyorin University School of Medicine, Tokyo (Japan) (Received 10 November, 1989) (Revised, received 27 December, 1989) (Accepted 28 December, 1989)

SUMMARY

Because a highly sensitive method is required to quantify low concentrations of immunoglobulin (Ig) classes in cerebrospinal fluid (CSF), there have been a few papers reporting normal values of CSF IgG, IgA and IgM determined in the same samples. Enzyme immunoassay (EIA) is most frequently used, but has such drawbacks as susceptibility of enzyme to inhibition and denaturation and the requirement for additional incubation with a substrate. Therefore, solid-phase immunofluorometric assay was evaluated for quantification ofCSF IgG, IgA and IgM in the nanogram range. We found this to be rapid and reproducible. The mean (SD) values of normal CSF samples obtained from 22 subjects with tension headache were 23.9 (7.6) #g/ml for IgG, 2.00 (0.90) #g/ml for IgA and 197 (87) ng/ml for IgM. The normal mean (SD) values of indexes were 0.51 (0.10) for IgG, 0.25 (0.05) for IgA and 0.044 (0.017) for IgM. These values agreed quite well with those determined by EIA. The values of CSF albumin correlated significantly with those of CSF IgG or IgA, but did not with those of CSF IgM. Levels of each of the three Ig classes in CSF and serum were significantly correlated. When CSF/serum ratio was introduced, a significant correlation between the albumin ratio and each Ig ratio was found. These results suggest that the Ig content of Correspondenceto: Dr. Tsuneyuki Takeoka, Department of Neurology, Tokai University School of Medicine, Iseharashi, Kanagawaken 259-11, Japan. 0022-510X/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

230 normal CSF may depend upon that of serum and upon the characteristics of the Ig molecule.

Key words: Solid-phase immunofluorometric assay; CSF IgG; CS F IgA; CSF IgM; Immunoglobulin index; Albumin ratio; Immunoglobulin ratio

INTRODUCTION In contrast to many reports concerning immunoglobulin (Ig) G of cerebrospinal fluid (CSF), only a few have simultaneously presented the concentrations of all three classes of immunoglobulins (IgG, IgA and IgM) of normal CSF samples. A highly sensitive and rather laborious technique is required to quantify CSF IgM, which normally occurs in low concentrations. Determinations of IgG, IgA and IgM in the same normal CSF samples have been performed with such methods as radioimmunoassay (RIA) (Nerenberg and Prasad 1975; Mingioli et al. 1978) and enzyme-immunoassay (EIA) (Kobatake et al. 1980; Hirohat~/et al. 1984; Lolli et al. 1989). In addition, particle counting immunoassay (PACIA) has been used to determine CSF IgM and IgA concentrations by Sindic et al. (1982, 1984). RIA has several drawbacks, such as a health hazard due to radiation, the need for special care in handling of reagents, expensive instrumentation, and the limited useful lifetime of a kit because of the half-life of the isotope. Among the alternative labels developed to substitute for radiotracers are enzymes and fluorescent probes. Fluoroimmunoassay (FIA), based on labelling immunoreactants with fluorescent probes, has superior sensitivity and is rapidly gaining wider applications (Hemmil/t 1985). We report here the methodology of quantification of CSF IgG, IgA and IgM by means of a solid-phase immunofluorometric assay (IFMA) using a commercially available kit. The other purpose of this study was to determine normal reference values (concentrations and indexes) of those immunoglobulins in CSF.

MATERIALSAND METHODS

Reference subjects It is generally not possible to collect CSF from healthy individuals. So, we regarded as "normal reference CSF", samples obtained from 22 patients with tension headache diagnosed during the period between 1982 and 1987. In addition to a standard neurological examination they had laboratory tests for liver function, serum electrolytes, peripheral blood figures and urinalysis, all of which were proved to be normal. They were 14 males and 8 females. The range of age was from 14 to 55 years with a mean (SD) of 32.0 (11.3). Informed consent was obtained.

231

Samples of CSF and serum Lumbar puncture was performed atraumaticaily between 9 : 00 and 10 : 00 am in a fasting state. Blood was drawn for paired analysis after the tap had been finished. A volume of 6-9 ml of CSF was obtained. After routine examinations for total protein (TP) and albumin concentration, glucose and chloride content, and the cell count, the sample was centrifuged at 400 g for 10 min. The supernatant fluid was divided into several aliquots and stored at - 80 ° C. The sample was analysed within 2-4 weeks after the spinal tap. It was thawed once for analysis. The following CSF mean (SD) values were all within the normal ranges: cell count, 2 (1)per mm3; glucose, 64.3 (6.5) mg/dl; chloride, 125 (2)mEq/1; CSF-albumin/serum-albumin ratio, 3.26 (0.67) x 10- 3. In 4 out of 22 subjects CSF albumin level could not be measured because insufficient volume of the sample had been obtained. Determination of protein concentration The TP content of CSF was determined by the Coomassie brilliant blue dyebinding method of Bradford (Johnson and Lott 1978). Serum IgG, IgA, IgM and CSF albumin were measured by laser nephelometry (Behring, GFR). Serum total protein and albumin were determined by an automatic clinical anaiyser (Hitachi Co., Japan). Determination of the optimum conditions for immunoassay The method was solid-phase IFMA which is a non-competitive, heterogeneous, "sandwich" technique. Immuno-Fluor kits (Bio-Rad Laboratories, U.S.A.) for measuring each Ig class were employed. The following are comprized in a kit. (a) One bottle containing 200 ml of standard buffer concentrate which makes 2liters of buffer, pH 7.5, containing 0.15mol/l NaC1 and 0.01 mol/1 phosphate (PB S). (b) One vial containing lyophilized Immunobead reagent which consists of a rabbit anti-human antibody preparation specific for IgG, IgA or IgM, covalently bonded to derivatized polyacrylamide beads. (c)One vial containing lyophilized fluorescein isothiocyanate (FITC)-conjugated rabbit anti-human antibody, specific for IgG, IgA or IgM. (d) One vial containing lyophilized immunoglobulin standard prepared from clarified human serum, with assayed values for IgG, IgA or IgM. All the above reagents contain NaN 3 as a preservative. A vial of Immunobead reagent was mixed with 50 ml of PBS, and FITC-conjugated antibody was reconstituted with 2.5 ml PBS at least 1 h before use. The fluorescent reagent was then centrifuged at 1700 x g for 10 rain. The supernatant was decanted into a clean container. The antibody reagents were stored at 5 oC when not in use. Serum Ig standard was reconstituted with 2.0 ml distilled water, divided into small aliquots, and kept at - 80 ° C until thawed for use. Dilution of the standard serum was performed with PBS containing lYo bovine serum albumin (Sigma, U.S.A.) (PBS-BSA). The procedure of determinationof the optimal conditionsfor immunoassaywas carried out essentiallyas described by Burgett et al. (1977). (1) Determination of the optimal amount of immobilized antibody (Immuno-

232

bead). This assay system requires that the immobilized antibody be maintained in excess so that each Ig present in the samples to be tested will be bound completely. Increasing amounts of the immobilized antibody solution (20-950 #1) were added to 50 #1 of the corresponding diluted standard serum that contained 204 ng IgG, 202 ng IgA or 208 ng IgM. Blank samples containing the immobilized antibody, but no Ig, were run concurrently. The final volume of each reaction tube was brought to 1.0 ml with PBS. After incubation of each reaction tube at 37 °C for 1.5 h in a water bath equipped with a shaker (Taitec, Japan), 2.0 ml PBS was added. The samples were mixed and centrifuged at 1700 x g for 10 min. The supematant fluid was discarded from each sample and the pellet was resuspended in 1.0 ml PBS. A large excess of FITC-labeled antibody (50 #I) was then added to each test tube, and the contents were then mixed and incubated for 1 h at 37 °C in a water bath. Next, the reaction complex was washed twice to thoroughly remove unreacted FITC-labeled antibody, i.e., 2.0 ml PBS was added and mixed, and the tube was centrifuged at 1700 x gfor 10 min. The supernatant was discarded from each tube and the pellet was resuspended in 3.0 ml of PBS. The mixture and centrifugation were done once more. The washed immobilized antibody complex in each tube was resuspended in 2.5 ml PBS. The fluorescence of the sample was determined by a spectrofluorophotometer (Shimadzu RF510, Japan), with excitation at 485 nm and emission at 525 nm. The incubation time was decided according to our preliminary experiments and the manufacturer's specifications. Disposable glass tubes of 12 x 75 mm were used as reaction tubes. (2) Determination of the optimal amount of FITC-labeled antibody. It is necessary that abundant labeled antibody be present in the assay system in order to react completely with each Ig class in the test sample. To determine the optimal amount of the labeled antibody each assay tube was set up so that it contained the prior quantity of each Ig and a constant volume of immobilized antibody reagent (500 #1 for IgG or IgA or 400 #1 for IgM). Blank tubes containing only the immobilized antibody were run simultaneously. The final volume of each tube was brought to 1.0 ml with PBS. The tubes were incubated at 37 °C for 1.5 h in a water bath, then 2.0 ml PBS was added to each sample and mixed. The tubes were centrifuged at 1700 x g for 10 min. The supernatant fluid was discarded and each pellet was resuspended in 1.0 ml PBS. Subsequently, increasing amounts of FITC-labeled antibody solution (2-230 #1) were added to the samples, with an adjustment for the final volume with PBS. The tubes were incubated at 37 ° C for 1 h in a water bath. The antigen-antibody compound was washed twice in the same way as described above. The fluorescence of each sample was measured by a spectrofluorophotometer.

Examination of specificity of immobilized antibody (Immunobead) Specificity of anti-IgG immobilized antibody was examined as follows. A constant volume (500 #1) of anti-IgG immobilized antibody was added to 50 #1 PBS (blank), and to 50 #1 of the diluted standard serum that contained 25, 49, 98, 147, 172, or 196 ng IgG. The final volume of each reaction tube was brought to 1.0 ml with PBS. After that, the procedure was carried out in the same way as described above. But,

233 instead of anti-IgG FITC-labeled antibody 25 #1 anti-IgA or anti-IgM FITC-labeled antibody was added to each reaction tube. The fluorescence of each sample was corrected by subtracting the blank value in order to detect cross-reaction between anti-IgG immobilized antibody and IgA or IgM. The same procedure was performed to examine the specificity of anti-IgA or anti-IgM immobilized antibody. A constant volume (500 #1) of anti-IgA immobilized antibody was added to 50 #1 PBS (blank), and to 50 #1 of the diluted standard serum that contained 24, 49, 98, 146, 171 or 195 ng oflgA. Anti-IgG or anti-IgM FITC-labeled antibody (25 #1) was used so that cross-reaction b~tween anti-IgA immobilized antibody and IgG or IgM could be detected. In the test of specificity of anti-IgM immobilized antibody, 400 #1 anti-IgM immobilized antibody was mixed with 50 #1 PB S (blank), and with 50 #1 of the diluted standard serum that contained 26, 51, 103, 154, 179 or 205 ng IgM. Anti-IgG or anti-IgA FITC-labeled antibody (25 #1) was used in order to discover cross-reaction between anti-IgM immobilized antibody and IgG or IgA.

Determination of CSF IgG, IgA or IgM by solid-phase IFMA Prior to assay, the CSF samples were diluted appropriately with PBS-BSA. The degree of dilution depended upon the CSF TP level. For example, when CSF TP concentration was within the normal range, the dilution was 1:10 for IgG but non-diluted specimen was used for the IgA or IgM assay. The immunoassay procedure was as follows. Numerals in parenthesis are for the IgM assay. (1) Pipet accurately 50 #1 (300 #1) of the diluted or non-diluted CSF specimens into each disposable glass tube. (2) Add 500 #1 (400 #1) Immunobead reagent to each assay tube and vortex. (3) Incubate all the test tubes at 37 °C for 1.5 h in a water bath. (4) Add 25 #1 fluorescent reagent to each tube and then vortex. (5) Incubate for 1 h in water bath. (6)Add 2.0 ml PBS to each tube and vortex. (7)Centrifuge at 1700 × g for 10 min. (8) Decant the supernatant carefully so as not to disturb the pellet at the bottom of the tube. (9) Add 3.0 ml PBS to each tube and vortex. Then, centrifuge again. (10) Decant the supernatant. Resuspend the pellet in 2.5 ml PBS, and vortex. (11) Measure fluorescence. A standard curve was constructed by assaying samples of diluted standard serum and their accompanying blanks containing no serum. Each sample was examined in duplicate. The fluorescence of each sample was corrected by subtracting the blank value. Usually 48 assay tubes were run at the same time under our experimental conditions. For comparison, the IgG values of CSF samples were also determined by laser nephelometry. The index for IgA or IgM was calculated according to the formula: (CSF Ig/serum Ig)/(CSF albumin/serum albumin) which was originally proposed as an IgG index (Tibbling et al. 1977).

234 RESULTS

Optimum conditionsfor the immunoassay The experimental results on the Ig binding capacity of the immobilized antibody are shown in Fig. 1. An estimation of the Ig binding capacity can be made by extrapolating the linear portion of the saturation curve to the maximal fluorescence (Burgett et al. 1977). Similar saturation curves were obtained by using different lots of Immunobeads (lots No. 24971 and 26103 for IgG; lots No. 25147 and 27514 for IgA; lots No. 24183 and 25665 for IgM). Therefore, 500 #1 of the immobilized antibody solution was enough for the immunoassay of 200 ng of IgG or IgA, and 400 #1 of the immobilized antibody was sufficient for the assay of 208 ng IgM. The saturation curves for FITC-labeled antibody are shown in Fig. 2. The optimal amount was calculated in the same way as for Immunobead. There was an initial steep rise with the use of up to 10 #1 labeled antibody and thereafter a slower increase of fluorescence occurred due to excess of unreacted antibody. This phenomenon was confirmed at each experiment by usage of different lots of the labeled antibody (lots No. 25173 and 26097 for IgG; lots No. 25045 and 25847 for IgA; lots No. 23787 and 25785 for IgM). It was concluded that 25 #1 FITC-labeled antibody satisfied the requirement for the assay of approx. 200 ng IgG, IgA and IgM.

50 40!

n-,

20

/

$

•~ ~ • ntligG-~An--tl "--'-- •..... "~ •A

::;?o; :;iiii

t,

I L I I I I L [ J 0 50 150 250 350 450 550 650 750 850 950

Immunobead Reagent

Fig. 1

.e

.i ~

20

l

• AntiIgG Antibody o• AntMgA Antibbody ody Anti-lgMAnti

i 10 0

0I0 30

J 70

i J h I10 150 190 230

FITC-Labeled Antibody

~e

Fig. 2

Fig. 1. Determinations of the optimal amount of rabbit anti-human immunoglobulin antibody covalently bonded to derivatized polyacrylamide beads (Immunobead). A fixed amount ofimmunoglobulin was used in each assay: 204 ng IgG, 202 ng IgA or 208 ng IgM. A constant volume of FITC-labeled antibody (50/d) was added for each experiment. Fig. 2. Saturation curves for FITC-labeled rabbit anti-human immunoglobulin antibody. A constant amount of immunoglobulin was used in each assay: 204 ng IgG, 202 ng IgA or 208 ng IgM. A previously determined optimal volume of Immunobead reagent was applied in each experiment.

235

Examination of specificity of immobilized antibody (Immunobead) In each examination of immobilized antibodies, the fluorescence of every test tube was not higher than the blank value. So, there was no cross-reaction between specific immobilized antibody and either of the other Ig classes. These results also revealed indirect evidence of specificity of FITC-labeled antibodies.

Standard curves for each of the Ig classes A standard curve was created in every experiment. Typical results are shown in Fig. 3. When four points of Ig concentration values were analysed by the linear regression method, the standard curve for IgG assay was linear between 61 ng/ml (3 ng IgG per aliquot) and 3924 ng/ml (196 ng per aliquot). The standard curve for IgA assay was linear between 976 ng/ml (49 ng IgA per aliquot) and 3906 ng/ml (195 ng IgA per aliquot), but it deviated from linearity at 488 ng/ml. The standard curve for IgM assay was linear between 86 ng/ml (26 ng IgM per aliquot) and 684 ng/ml (205 ng IgM per aliquot), but it skewed below a level of 64 ng/ml. Each Ig concentration of CSF samples to be examined was calculated by applying the linear equation of the respective standard curve. However, for measurement of very low concentrations of CSF IgA or IgM, the sample volume applied was increased so that the level might fall within the linear portion of the standard curve.

Sensitivity, precision and comparison of CSF lgG values determined by two methods The detection limits oflgG, IgA and IgM were 30 ng/ml (1.5 ng/aliquot), 60 ng/ml (3 ng/aliquot) and 20 ng/ml (6 ng/aliquot), respectively. The intra-assay precision was studied by 10 repetitive determinations for each of two CSF samples, one with a normal TP concentration and the other with increased protein content. The coefficients of variation (CV) for each of the Ig classes were all less than 5 ~ . The inter-assay variance was estimated by 10 different analyses of the same pooled CSF. The CVs for IgG, IgA and IgM were 4.3~, 6.9~ and 5.7 ~o, respectively.

,/

/

,/

/.

/

/

/

® 3o

/

~"

20

3c

•

~ 20

®

20

/•

/

/o

/

Io

/

/

/'

,/

•/

IgG Concentration

.gml

IgA Concentration

~,ml

IoM Concentration

.g/ml

Fig. 3. Standard curves obtained bylinear regression analysis of 4 values (large circles)ofeachimmunoglobulin. For IgG measurement, the plot was linear between 61 ng/ml and 3924 ng/ml. For IgA determination, it was linear between 976 ng/ml and 3906 ng/ml, but it deviated at 488 ng/ml. For IgM assay, it was linear between 86 ng/ml and 684 ng/ml, but it skewed below a level of 64 ng/ml.

236

5O

/"

CSF IgG

J"

J j, ° j,~ 3O

:.~.." 2O /

/ •

"

n=20

// s

~

I

~

~o

,o

CSF IgG (~g/m{) Laser NeDhelometry

Fig. 4. Comparison of CSF IgG values determined by two methods, immunofluorometricassay (1FMA) and laser nephelometry. There was a good correlation (r = 0.94, P < 0.001).

Because laser nephelometry is not sensitive enough to quantify C S F IgM concentration, only C S F IgG values were compared between this i m m u n o a s s a y and laser nephelometry. The correlation coefficient was 0.94 (P < 0.001) (Fig. 4).

Normal reference values of CSF IgG, IgA and IgM As shown in Table 1 m e a n (SD) values were 28.8 (7.6) mg/dl for C S F T P and 15.1 (3.2) mg/dl for C S F albumin. M e a n (SD) values of IgG, IgA and IgM were 23.9 (7.6) #g/ml, 2.00 (0.90) #g/ml and 197 (87) ng/ml, respectively. The m e a n (SD) values of the indexes were 0.51 (0.10) for IgG, 0.25 (0.05) for IgA and 0.044 (0.017) for IgM.

TABLE 1 NORMAL REFERENCE VALUES OF CSF IMMUNOGLOBULINS AND INDEXES Subjects Age (yr)

Mean SD Range

32.0 11.3 1455

TP mg/dl

28.8 7.6 1843

Albumin mg/dl

15.1 3.2 1021

IgG

IgA

IgM

#g/ml

index

Fg/ml

index

ng/ml

index

23.9 7.6 14.347.1

0.51 0.10 0.330.68

2.00 0.90 0.743.82

0.25 0.05 0.170.33

197 87 48358

0.044 0.017 0.0170.080

CSF samples were obtained from 22 subjects (14 males and 8 females) with tension headache. Techniques of protein determination were described in Methods. TP = total protein. The mean (SD) value of CSF-albumin/serum-albuminratio was 3.26 (0.67) x 10- 3 Immunoglobulinindex was calculated according to the formula (CSF-Ig/serum-Ig)/(CSF-albumin/serum-albumin).

237 /~g/ml 40

,ug/ml

./ ./ r=O n=1873

3.(

300

ro--O;0

n=18 P
~0~

/,...

(...)

ng/ml 400

r=O 58

i/~

P<0 01

4.(

.

.

NS

U.. 2.0

EL 20C

I.O

I00

==

=

* o

i

=

5

I

I

o

i

10 15 20 CSF Albumin

25

~

l

k

10

mo/di

~

15

i

2

CSF Albumin

o

25

I

5

mg/dl

I

I

10

I

15

20

CSF Albumin

25 rag/dr

Fig. 5. Relationships between CSF immunoglobulin values and CSF albumin concentrations. A significant correlation was observed in the case of IgG or IgA. The correlation for IgM was not significant (NS).

Relationships between CSF immunoglobulins and other proteins A significant correlation was found between CSF TP and CSF IgG (r = 0.61, P < 0.01), but CSF TP values did not significantly correlate with the values for CSF IgA (r = 0.17) or IgM (r = 0.35). As shown in Fig. 5, the values of CSF albumin correlated significantly with the values for IgG (r = 0.73, P < 0.01) or IgA (r = 0.58, P < 0.02). However, CSF IgM values did not correlate significantly with the values of CSF albumin (r = 0.40). Fig. 6 is largely self-explanatory. Significant correlations were observed between CSF IgG and serum IgG (r = 0.45, P < 0.05), between CSF IgA and serum IgA (r = 0.81, P < 0.01) and between CSF IgM and serum IgM (r - 0.53, P < 0.05). There were also significant correlations at a level of P < 0.01 between CSF/serum ratio for

,ug/ml

#g/ml 4.0

50

ng/ml

400

r=o81 n=22 * p
t= O 45

n=22

p
30

3O

* *

*. . ~

/ 300

<

--~2,

~a~200

O

•

•

*

/~. ."

I

I

1500

Serum IgG

I

2000"

rag/d(

I

I00

r=o53 n=22 p<0 o5

•

/"

°°:

I000

100

*

I.

I

200

Serum

I

300

IgA

I

400

mg/dl

0

L

I ZOO

I

I 200

Serum IgM

I

i

300

rng/dl

Fig. 6. Relationships between immunoglobulin levels in CSF and those in the serum. A significant correlation was observed for each of the immunoglobulin classes.

238 xlO

×~o

'

•

/

2

, =o 62

o ~

~ "

30

/

, =0 6B

/

r,=~B P<0 01

P

.

,

~

25

•/

< ""

J

•

o

2O

~

15

•

p
/ /

0

0 I

AJbumlr~ Ratio

5!0

xl0'

2,o

~'~

,'o

Albumm Ratio

~!0 Xl8 '

~Io

~!o

?o

Albumin Ratio

¢o x 10 '

Fig. 7. Relationships between CSF/serum ratio for each immunoglobulin and CSF/serum ratio for albumin. A significant correlation was observed in each case.

albumin and CSF/serum ratio for Ig; r = 0.69 for IgG; r = 0.62 for IgA; r = 0.68 for IgM (Fig. 7).

DISCUS SION

Since the first introduction of EIA for determination of CSF IgG, IgA and IgM by our department (Kobatake et al. 1980) the technique of EIA, in place of RIA, has gained increasing acceptance for measurement of low concentrations of CSF IgM. However, in his review of FIA and IFMA Hemmil~t (1985) pointed out that the problems in EIA are caused by the bulky, labile label (enzyme), its susceptibility to inhibition and denaturation, and the additional incubation with a substrate for monitoring the enzyme's activity. In the present study we have described the methodology of solid-phase IFMA for quantification of CSF immunoglobulins in the nanogram range. It was rapid and reproducible. The problem of the background activity of the fluorometric measurement could be solved by the solid-phase separation system. We confirmed that the solid-phase IFMA is a suitable method for determination of low concentrations of CSF Ig. By utilizing the same kind of kit for FIA as that in the present study, CSF IgG was measured by Menonna et al. (1977), and the three Ig classes of CSF were quantified by Delisi et al. (1981). However, the optimal assay conditions were not described and seemed not to have been studied in those two reports. So it is difficult to compare their results with ours. Hische et al. (1979) briefly examined the experimental conditions of a similar IFMA for estimation of CSF IgM, and Wolters et al. (1988) employed their method to determine IgM values. We have now established the optimal conditions for these commercially obtainable antibodies for solid-phase IFMA for determination of CSF Ig. Our mean concentration of IgG of normal CSF was in excellent agreement with those determined by EIA, but was lower than those quantified by RIA. As regards normal reference concentrations of CSF IgA and IgM the values differ slightly among the various reporters, including us. This may be partly because of the variety of"normal"

239 subjects and partly because of the diversity of methodologies; RIA (Nerenberg and Prasad 1975; Mingioli et al. 1978), EIA (Kobatake et al. 1980; Hirohata et al. 1984; Lolli et al. 1989) and PACIA (Sindic et al. 1982, 1984). The value of IgM index of this report agreed quite well with the values proposed by Sindic et al. (1982), Forsberg et al. (1984), and Wolters et al. (1988), but it was slightly higher than that reported by Lolli et al. (1989). Concerning the level of IgA index PACIA (Sindic et al. 1984) offered a similar result to ours, but Lolli et al. (1989) gave a lower value. There was good agreement between the value of IgG index determined by EIA and that found by the present method. The CSF levels of IgG concentration and index have been examined satisfactorily in various neurological disorders by many researchers. In contrast, those of IgA and especially IgM remain to be evaluated to establish their clinical significance. Our result that values of CSF albumin correlated significantly with those of CSF IgG, but not with those ofCSF IgM is in accordance with that of Forsberg et al. (1984) using EIA. The correlations between the levels of each of the three Ig classes in CSF and serum were significant in our reference group, in agreement with the results of Lolli et al. (1989). A significant correlation of IgM between CSF and serum was also identified by Forsberg et al. (1984). The absence of a significant correlation between CSF albumin and CSF IgM is of interest, because CSF albumin is generally considered to reflect the relationship between the two compartments, serum and CSF, better than CSF TP. However, when CSF/serum ratio was calculated for each protein, significant correlations were found not only between the albumin ratio and the IgG (or IgA) ratio but also between the albumin ration and the IgM ratio. Sindic et al. (1982) and Lolli et al. (1989)obtained similar results. It is generally agreed that in normal CSF, albumin originates exclusively from serum, and that the CSF/serum ratio for albumin is an indicator of the blood-brain barrier integrity (Tourtellotte et al. 1985). Hence, on the basis that the molecule oflgM is approx. 5 times larger than that oflgG or IgA, our results may be explained as follows. In normal CSF the concentrations of all three classes of immunoglobulins (IgG, IgA and IgM) may depend upon those in serum, but this natural leakage of each Ig from serum to CSF may be influenced by molecular weight, size and shape. However, further analysis of CSF samples on a large scale is necessary to confh'm these results.

REFERENCES Burgett, M.W., S.J. Fairfield and J.M. Monthony (1977) A solid phase fluorescent immunoassay for the quantitation of the C3 component of human complement. Clin. Chim. Acta, 78: 277-284. Delisi, L.E., D.R. Weinberger, S. Potkin, L.M. Neckers, D.J. Shiling and R.J. Wyatt (1981) Quantitative determination of immunoglobulins in CSF and plasma of chronic schizophrenic patients. Brit. J. Psychiat., 139: 513-518. Forsberg, P., A. Henriksson, H. Link and S. Ohman (1984) Reference values for CSF-IgM, CSF-IgM/S-IgM ratio and IgM index, and its application to patients with multiple sclerosis and aseptic meningoencephalitis. Scand. J. Clin. Lab. Invest., 44: 7-12. Hemmil~i, I (1985) Fluoroimmunoassays and immunofiuorometric assays. Clin. Chem., 31: 359-370. Hirohata, S., T. Inoue, A. Yamada, S. Hirose and T. Miyamoto (1984) Quantitation of IgG, IgA and IgM

240 in the cerebrospinal fluid by a solid-phase enzyme-immunoassay. Establishment of normal control values. J. Neurol. Sci., 63:101-110. Hische, E.A.H., H.J. van der Helm and T. Out (1979) Estimation of IgM in cerebrospinal fluid by fluoroimmunoassay. Clin. Chim. Acta, 97: 93-95. Johnson, J.A. and J, A. Lott (1978) Standardization of the Coomassie Blue method for cerebrospinal fluid proteins. Clin. Chem., 24: 1931-1933. Kobatake, K., Y., Shinohara and S. Yoshimura (1980) Immunoglobulins in cerebrospinal fluid. Enzymeimmunoassay. J. Neurol. Sci., 47: 273-283. Lolli, F., I. Halawa and H. Link (I 989) Intrathecal synthesis of IgG, IgA, IgM and IgD in untreated multiple sclerosis and controls. Acta Neurol. Scand., 80: 238-247. Menonna, J., D. Galantowicz, P. Dowling and S. Cook (1977) Rapid fluorometric assay for cerebrospinal fluid immunoglobulin G. Neurology, 27: 481-483. Mingioli, E. S., W. Strober, W.W. Tourtellotte, J.N. Whitaker and D.E. McFarlin (1978) Quantitation of IgG, IgA and IgM in the CSF by radioimmunoassay. Neurology, 28: 991-995. Nerenberg, S.T. and R. Prasad (1975) Radioimmunoassays for Ig classes G, A, M, D, and E in spinal fluid: normal values of different age groups. J. Lab. Clin. Med., 86: 887-898. Sindic, C. J. M., C. L. Cambiaso, A. Depr6, E. C. Laterre and P. L. Masson (1982) The concentration of IgM in the cerebrospinal fluid of neurological patients. J. Neurol. Sci., 55: 339-350. Sindic, C.J.M., D.L. Delacroix, J.P. Vaerman, E.C. Laterre and P.L. Masson (1984) Study oflgA in the cerebrospinal fluid of neurological patients with special reference to size, subclass and local production. J. Neuroimmunol., 7: 65-75. Tibbling, G., H. Link and S. (Jhman (1977) Principles of albumin and IgG analyses in neurological disorders. Part I. Establishment of reference values. Scand. J. Clin. Lab. Invest., 37: 385-390. TourteUotte, W.W., S.M. Staugaltis, M.J. Walsh, P. Shapshak, R.W. Baumhefner, A.R. Potvin and K. Syndulko (i 985) The basis of intra-blood-brain-barrier IgG synthesis. Ann. Neurol., 17:21-27. Wolters, E.C., E.A.H. Hische, J.A. Tutuarima, L. van Trotsenburg, R.V.W. van Eijk, J.D. Bos, T.M. Starink, L.J. Emsbroek and H.J. van der Helm (1988) Central nervous system involvement in early and late syphilis: the problem of asymptomatic neurosyphilis. J. Neurol. Sci., 88: 229-239.