- Email: [email protected]
The effect of reaction temperature for nephelometric assays for rheumatoid factor
Clinica Chimica Acta 292 (2000) 117–125 www.elsevier.com / locate / clinchim
The effect of reaction temperature for nephelometric assays for rheumatoid factor Hachiro Yamanishi*, Shigeru Iyama, Yoshihisa Yamaguchi, Takehiko Yanagihara The Central Laboratory for Clinical Investigation, Osaka University Hospital, 2 -15, Yamadaoka, Suita, Osaka, 565 -0871, Japan Received 8 June 1999; received in revised form 5 October 1999; accepted 11 October 1999
Abstract Even using the same assay parameter, reagent and calibrator (N-latex RF kit II), the results of the assay for serum rheumatoid factors (RFs) with the Behring Nephelometer Analyzer (BNA) were higher than those with the Behring Nephelometer II Analyzer (BNII) ([BNII] 5 0.76 [BNA] 2 5.7 kIU / l, r 5 0.997, Sy /x 5 60.73, n 5 99). The mean bias (BNA minus BNII)6S.D. was 52.7685.5 using the Bland and Altman plot method, and the bias was not constant. The only difference in the assay condition with the two methods was the reaction temperature with the BNA being performed at room temperature (25618C) and the BNII being performed at 378C. The ratio of the results with the BNII to the BNA (BNII / BNA) ranged from 0.23 to 1.18. A significant difference was observed in the BNII / BNA ratio in patients with high levels of C-reactive protein (CRP) over 2.0 mg / l (mean BNII / BNA ratio; 0.78) in comparison to patients with normal CRP levels under 2.0 mg / l (mean BNII / BNA ratio; 0.65) (P , 0.01). The RF concentrations with the BNA were reduced by addition of urea, which has been used as a mild protein-denaturing agent, and there was a significant correlation between the values calculated as (1-value treated with urea / original value without urea) 3 100 and the BNII / BNA ratio (r 5 0.652, P , 0.01). These data suggested that the bias between the RF values obtained by the BNA and BNII might be caused by the variation in the reactivity of autoantibodies, which might be decreased in some inflammatory diseases. 2000 Elsevier Science B.V. All rights reserved.
*Corresponding author. Fax: 1 81-66-879-6635. E-mail address: [email protected] (H. Yamanishi) 0009-8981 / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0009-8981( 99 )00207-7
118
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
1. Introduction Rheumatoid factors (RFs) are autoantibodies directed to the Fc portion of IgG [1], which are the most characteristic autoantibodies found in patients with rheumatoid arthritis (RA). The presence of serum RFs is one of the criteria used for the diagnosis of RA [2]. The biological significance has been studied in normal and diseased states [3], including the relation between synthesis of RFs and infection by bacteria or virus [3–6]. RFs are detected in serum of patients with RA and other diseases [7]. RFs are composed of different classes of immunoglobulins including IgG, IgA, IgM, IgD and IgE. Serum RFs consist of many autoantibodies that can bind to various epitopes of IgG molecules or other immunoglobulins, but the IgM class has been measured for clinical purposes in many instances [3]. For measurement of RFs, a radioimmunoassay [8–10], enzyme immunoassay [11,12] and nephelometric assay have been developed for routine analysis [13–16]. Recently, a fully automated nephelometric assay based on the reaction of IgG or gamma globulin-coated latex particles with RFs has been widely used. In our laboratory, 12 serum analytes including RFs have been measured by a nephelometric assay using the Behring Nephelometer Analyzer (BNA) (Dade Behring Marburg Gmbh). However, the BNA has been replaced with the Behring Nephelometer II Analyzer (BNII) (Dade Behring Marburg Gmbh) for routine assay, and we have found that the RF values obtained with the BNII were lower as compared with those obtained with the BNA, even though we used the same assay parameter, reagent and calibrator in all measurements. The only difference in the assay condition between the two methods was the reaction temperature, as the assay with the BNA being carried out at room temperature (25618C) and that with the BNII at 378C. Here, we describe a plausible explanation of the difference in RF values obtained with the BNA and BNII, including the presence of low avidity autoantibodies [17], whose reactivities are inhibited by addition of urea, a mild protein-denaturing agent. [18–21].
2. Materials and methods
2.1. Serum samples For the RF assay, serum samples from 99 patients (69 patients with RA and 30 patients with other diseases) examined in our hospital were analyzed. Whole blood samples were centrifuged after clotting at room temperature. The separated sera were assayed immediately without storage at 48C.
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
119
2.2. Methods for specimen analysis Serum RFs were measured by the BNA and BNII according to the manufacture’s instruction using ‘N-latex RF kit II’ (Dade Behring Marburg Gmbh). This commercial kit consists of two reagents, one is latex particles coated with human IgG-sheep anti-human IgG complex, and another is a supplementary reagent containing polyethyleneglycol and NaCl. In a modified BNA assay for RFs, the ‘N-Diluent’ (dilution buffer for the BNA and BNII) containing urea (0.06–8 mol / l) was used as the dilution buffer. The final dilution of urea added to the dilution buffer was 2.6-fold in the reaction mixture. The CRP concentration was measured by a routine turbidimetric assay using goat anti-CPR antibody on the Hitachi-747-400 (Roche).
3. Results As shown in Fig. 1, the correlation between RF values measured with the BNII ( y) and BNA (x) was y 5 0.76x 2 5.7 kIU / l (r 5 0.997, slope 95% confidence limit 5 0.75 to 0.77, intercept 95% confidence limit 5 2 27.6 to 16.3). The mean RF values measured with the BNA and BNII were 291.7 kIU / l and 215.5 kIU / l, respectively. The mean bias (BNA minus BNII) by using the Bland and Altman method was 52.7 (S.D.; 85.5, 95% confidence limits; 2 7.6 to 322.6). This bias was not constant throughout the range as shown by the slope of 0.76. This problem was related specifically to the RF assay, as there were no biases for the other nephelometric assays (transferrin, ceruloplasmin, haptoglobin, etc.) between these two analyzers (data not shown). To examine the influence of complement, ethylenediaminetetraacetate (EDTA) was added into serum specimens, and serum RF values were measured with the BNA and BNII. There was no significant change in the measured values with either method, irrespective of whether EDTA was added or not. The ratio of the RF value obtained with the BNII to that obtained with BNA (BNII / BNA) was calculated for each sample. As shown in Table 1, the mean BNII / BNA ratio from patients with RA was 0.752 and that from control subjects was 0.665. However, the difference was not statistically significant by the Mann–Whitney test. By using the dilution buffer containing urea (modified BNA assay), serum RF values were measured with the BNA in five samples, where the BNII / BNA ratio ranged from 0.42 to 1.07 (Table 2). At the concentration of 8 mol / l, the measured results were all under the detection limit (11.0 kIU / l). At the concentration of 2 mol / l, positive values were detected only with two samples (sample A,B), where the BNII / BNA ratio was 1.07 and 0.95, respectively. The
120
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
Fig. 1. Comparison of RF concentrations measured by the BNA and BNII. (A) The correlation between values measured by the BNA and BNII. Regression equation: y 5 0.76x 2 5.7, r 5 0.997, Sy /x 5 60.7, n 5 99. Dotted line indicates y 5 x. (B) The Bland and Altman plot obtained from two analyzers (overall mean difference6S.D.; 52.7685.5).
values were still , 11.0 kIU / l at the concentration of 1 mol / l with the samples having the BNII / BNA ratio of 0.52 and 0.42 (sample D,E). At the concentration of 0.25 mol / l or less, all results were . 11.0 kIU / l, and the mean ratio of the
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
121
Table 1 Concentration of RFs measured by the BNA and BNII and the BNII / BNA ratio in various diseases Mean RF conc. (kIU / ml)
BNII / BNA
n
BNA
BNAII
Mean
S.D.
Range
Rheumatoid arthritis
69
356.2
272.7
0.752
0.202
0.23–1.18
Control subjects Systemic lupus erythematosus Sjogren’s syndrome Progressive systemic sclerosis Myeloma Liver disease Other Total
6 3 4 5 5 7 30
118.4 210.8 142.9 106.3 204.3 118.9 143.3
72.3 78.6 78.8 78.3 105.2 89.9 84.0
0.663 0.480 0.683 0.784 0.572 0.732 0.665
0.213 0.151 0.295 0.196 0.127 0.230 0.216
0.36–0.97 0.34–0.64 0.49–1.12 0.56–0.99 0.46–0.78 0.34–1.03 0.34–1.12
detected values to the initial values was 91.2% at the concentration of 0.06 mol / l. Fig. 2 shows the relation between the BNII / BNA ratio and the rate of decrease in the RF values (%) as measured by the modified BNA assay in the presence of urea at the concentration of 0.5 mol / l. A significant relation was observed between the BNII / BNA ratio and the rate of decrease (r 5 2 0.652, 95% confidence limits; 0.314–0.848, P , 0.01). There was no significant difference between the BNII / BNA ratio from patients with RA and control subjects (Table 1). As shown in Fig. 3, however, a significant difference was observed (P , 0.01), when they were divided into two groups according to the serum CRP concentration above and below 2 mg / l. The mean BNII / BNA ratio was 0.780 (S.D., 0.202) in patients with CRP conTable 2 Changes in RF values (kIU / l) by addition of various concentrations of urea to the assay with the BNA Sample Initial RF conc.
BNII/BNA
Urea added (mol/l)
(kIU/l) 0.06
A
0.125
BNA
BNII
99.7
107
1.07
103 (103.3)a
149 (102.1) 140 (95.9)
B
146
139
0.95
C
106
68
0.64
98.2 (92.6)
91.7 (92.0) 85.1 (80.3)
0.25
82.8 (83.0) 119 (81.5) 76.7 (72.3)
0.5
68.2 (68.4) 102 (69.9) 56.0 (52.8)
D
36.6
19
0.52
28.1 (76.8)
29.0 (79.2)
21.5 (58.7) , 11.0
E
47.2
20
0.42
38.4 (81.4)
34.3 (72.7)
25.4 (53.8)
a
1
2
8
44.8 (44.9) 23.2 (23.3) 74.6 (51.1)
37.7 (25.8)
30.0 (28.3) , 11.0
, 11.0 , 11.0 , 11.0
, 11.0
, 11.0
, 11.0
12.6 (26.7) , 11.0
, 11.0
, 11.0
Percentage of the initial concentration measured with the BNA is shown in the parenthesis.
122
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
Fig. 2. Relation between the BNII / BNA ratio and the rate of decrease of RF values before and after addition of urea. Rate of decrease,%: (1-value treated with urea / original value) 3 100.
centrations of . 2.0 mg / l and 0.658 (S.D., 0.199) in subjects whose CRP concentrations were below the cut off value of 2 mg / l.
4. Discussion Although many studies have been reported regarding the reactivity of human RFs, the mechanism or specificity of the reaction is not entirely clear [3]. In the measurement of serum RFs for clinical purposes, the measured values sometimes have significant bias based on the reagents or calibrators used for the assay. Even using the same reagent, same calibrator and same assay parameter in the present study, different values were obtained between measurement with the BNA and BNII. The reduction of values by the BNII was not caused by the complement reaction, because there was no significant change in RF values measured either with the BNA or BNII before and after the activator was chelated with EDTA. All RF measurements with the BNA and BNII were carried out without storage at 48C, which might lead to an increase in the RF titers [17]. The slope of the regression line between the BNA assay and the BNII assay
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
123
Fig. 3. Comparison of BNII / BNA ratio in patients with normal CRP levels and above the cut-off value. Group A: Mean values for BNII / BNA ratio6S.D. 5 0.65860.199 in patients with CRP concentration # 2.0 mg / l. Group B: Mean values for BNII / BNA ratio6S.D. 5 0.78060.202 in patients with CRP concentration . 2.0 mg / l. The significant differences are shown in both groups (P , 0.01) by the Mann–Whitney test.
was 0.76, but the bias (BNA minus BNII) and BNA / BNII ratio were not constant (Fig. 1). Therefore, the difference appears to be due to the reactivity characteristic of RFs. These data suggest the presence of autoantibodies which were measured as RFs and whose reactivities could be changed by some factors even when the same reagent, calibrator and assay parameters were used. The notable discrepancies between the BNA and the BNII assay were observed not only in patients with RA, but also in patients with other diseases (Table 1). Anti-nuclear and anti-nucleolar antibodies are the autoantibodies found in a high proportion of patients with systemic lupus erythematosus, Sjogren’s syndrome and progressive systemic sclerosis. It was not clear whether these autoantibodies were measured as RF with the BNA or not. However, it was considered that the discrepancies between both methods might be caused by the variation in the reactivity of these autoantibodies including RFs. Therefore we tried to suppress the reaction caused by low avidity autoantibodies by addition of urea. Guanidine, urea and chaotropic ions (SCN 2 , I 2) which cause reversible denaturation of protein structures, are known to inhibit the formation of immune complexes [21]. It has been reported that urea dissociates a low avidity antibody
124
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
from an antigen as shown in the enzyme immunoassay for detection of anti-hepatitis C virus antibodies [18–20], and that guanidine hydrochloride is useful as a protein-denaturing agent for estimation of the avidity of antibodies in the enzyme-linked immunosorbent assay [21]. Although the initial concentration of RFs measured with the BNA for sample C (BNII / BNA 5 0.64) was close to that for sample A (BNII / BNA 5 1.07) (Table 2), a value above the cut-off level was observed with sample A but not with sample C at 2 mol / l urea. This suggests that the amount of RFs in serum inhibited by urea did not depend on the initial concentration of RFs. The result shown in Fig. 2 suggested that the effect of inhibition by urea was more notable in samples with low BNII / BNA ratios. This indicates that these samples might have a relatively large amount of RFs to be inhibited by urea. On the other hand, there was a good correlation between the values measured in the presence of urea by the modified BNA assay ( y) and the initial values on BNII (x) with y 5 0.97x 2 37.9 (r 5 0.999) (graph not shown). However, it is difficult to conclude that only the RFs which could not be detected by BNII were inhibited by the addition of urea with the modified BNA assay, because the mean ratio of decrease was 32.0% even in the samples with the high BNII / BNA ratio . 0.90 (mean;0.96) as shown in Fig. 2. Based on these results, we conclude that RF concentrations measured with the BNA were higher than those with the BNII, in which the low avidity RFs inhibited by urea might have been measured by the BNA. These antibodies may be temperature dependent, as this was the only difference in the assay condition between the two instruments. The reaction of antibodies and antigens is faster at 378C than at room temperature. On the other hand, immune complexes of low avidity or non-specific antibodies and antigens may be more stable at room temperature. The BNII / BNA ratio for measurement of RFs was significantly lower in patients with elevated CRP concentration ( . 2.0 mg / l), suggesting that the presence of these antibodies in serum may be dependent on the degree of inflammation, where synthesis of these autoantibodies to be suppressed by urea may be suppressed in the active inflammatory state. For clinical testing, these problems must be resolved in order to establish an accurate assay method for RFs. We believe that RF values measured at 378C are more useful for clinical purpose, especially for the diagnosis of RA.
References [1] Dorner RW, Alexander RL, Moore TL. Rheumatoid factor. Clin Chim Acta 1987;167:1–21. [2] Arnett FC, Edworthy SM, Bloch DA, Mcshane DJ, Fries JF, Cooper NS et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24.
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
125
[3] Moore TL, Dorner RW. Rheumatoid factor. Clin Biochem 1993;26:75–84. [4] Slaughter L, Carson DA, Jenson FC, Holbrook TL, Vaughan JH. In vitro effects of Epstein-Barr virus on peripheral blood mononuclear cells from patients with rheumatoid arthritis and normal subjects. J Exp Med 1978;148:1429–33. [5] Hirohata S, Inoue T, Miyamoto T. Frequency analysis of human peripheral blood B cells producing IgM-rheumatoid factor. Differential effects of stimulation with monoclonal antibodies to CD3 and Staphylococcus aureus. J Immunol 1990;145:1681–6. [6] Karsh J, Goldstein R, Lazarovits AT. In vitro IgM and IgM rheumatoid factor production in response to Staphylococcus aureus Cowan I and pokeweed mitogen: The contribution of CD5 1 (Leu 1) B cells. Clin Exp Immunol 1989;77:179–83. [7] Shmerling RH, Delbanco TL. The rheumatoid factor: An analysis of clinical utility. Am J Med 1991;91:528–34. [8] Nineham LJ, Hay FC, Roitt IM. Laboratory diagnosis of rheumatoid arthritis: a solid phase radioassay for IgG and IgM antiglobulins. J Clin Pathol 1976;29:1121–6. [9] Knez V, Reimer CB. An indirect solid-phase microradioimmunoassay for human IgM-antiIgG (rheumatoid factor). J Immunol Meth 1977;18:105–21. [10] Wernick R, Lospalluto JJ, Fink CW, Ziff M. Serum IgG and IgM rheumatoid factor by solid phase radioimmunoassay. Arthritis Rheum 1981;24:1501–11. [11] Maiolini R, Ferrua B, Quaranta JF, Pinoteau A, Euller L, Ziegler G et al. A sandwich method of enzyme-immunoassay II. Quantitation of rheumatoid factor. J Immunol Method 1978;20:25–34. [12] Teitsson I, Valdimarsson H. Use of monoclonal antibodies and F(ab9) 2 enzyme conjugates in ELISA for IgM. IgA and IgG rheumatoid factors. J Immunol Method 1984;71:149–61. [13] Deaton CD, Maxwell KW, Smith RS, Creveling RL. Use of laser nephelometry in the measurement of serum proteins. Clin Chem 1976;22:1465–71. [14] Virella G, Waller M, Fudenberg HH. Nephelometric method for determination of rheumatoid factor. J Immunol Method 1978;22:247–51. [15] Finley PR, Hicks MJ, Williams RJ, Hinlicky J, Lichti DA. Rate nephelometric measurement of rheumatoid factor in serum. Clin Chem 1979;25:1909–14. [16] Painter PC, Lyon JM, Evans JH, Powers WW, Whittaker RL, Decker MJ. Performance of a new rate-nephelometric assay for rheumatoid factor, and its correlation with tube-titer results for human sera and synovial fluid. Clin Chem 1982;28:2214–8. [17] Ueda K, Nakagawa T, Shimizu A. Rheumatoid factor in the serum of hepatitis C virusinfected patients: an increase in the titre during cold storage. Ann Clin Biochem 1996;33:438–42. [18] Goeser T, Blazek M, Gmelin K, Kommerell B, Theilmann L. Washing with 8 mol / l urea to correct false-positive anti-HCV results [Letter]. Lancet 1990;335:878. [19] Wreghitt TG, Gray JJ, Aloyisus S, Contreras M, Barbara JAJ. Antibody avidity test for recent infection with hepatitis C virus [Letter]. Lancet 1990;335:789. [20] Gray JJ, Wreghitt TG, Friend PJ, Wight DGD, Sundaresan V, Calne RY. Differentiation between specific and non-specific hepatitis C antibodies in chronic liver disease [Letter]. Lancet 1990;335:609–10. [21] Inoue S, Hasegawa A, Matsuno S, Katow S. Changes in antibody avidity after virus infections: Detection by an immunosorbent assay in which a mild protein-denaturing agent is employed. J Clin Microbiol 1984;20:525–9.
The effect of reaction temperature for nephelometric assays for rheumatoid factor Hachiro Yamanishi*, Shigeru Iyama, Yoshihisa Yamaguchi, Takehiko Yanagihara The Central Laboratory for Clinical Investigation, Osaka University Hospital, 2 -15, Yamadaoka, Suita, Osaka, 565 -0871, Japan Received 8 June 1999; received in revised form 5 October 1999; accepted 11 October 1999
Abstract Even using the same assay parameter, reagent and calibrator (N-latex RF kit II), the results of the assay for serum rheumatoid factors (RFs) with the Behring Nephelometer Analyzer (BNA) were higher than those with the Behring Nephelometer II Analyzer (BNII) ([BNII] 5 0.76 [BNA] 2 5.7 kIU / l, r 5 0.997, Sy /x 5 60.73, n 5 99). The mean bias (BNA minus BNII)6S.D. was 52.7685.5 using the Bland and Altman plot method, and the bias was not constant. The only difference in the assay condition with the two methods was the reaction temperature with the BNA being performed at room temperature (25618C) and the BNII being performed at 378C. The ratio of the results with the BNII to the BNA (BNII / BNA) ranged from 0.23 to 1.18. A significant difference was observed in the BNII / BNA ratio in patients with high levels of C-reactive protein (CRP) over 2.0 mg / l (mean BNII / BNA ratio; 0.78) in comparison to patients with normal CRP levels under 2.0 mg / l (mean BNII / BNA ratio; 0.65) (P , 0.01). The RF concentrations with the BNA were reduced by addition of urea, which has been used as a mild protein-denaturing agent, and there was a significant correlation between the values calculated as (1-value treated with urea / original value without urea) 3 100 and the BNII / BNA ratio (r 5 0.652, P , 0.01). These data suggested that the bias between the RF values obtained by the BNA and BNII might be caused by the variation in the reactivity of autoantibodies, which might be decreased in some inflammatory diseases. 2000 Elsevier Science B.V. All rights reserved.
*Corresponding author. Fax: 1 81-66-879-6635. E-mail address: [email protected] (H. Yamanishi) 0009-8981 / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0009-8981( 99 )00207-7
118
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
1. Introduction Rheumatoid factors (RFs) are autoantibodies directed to the Fc portion of IgG [1], which are the most characteristic autoantibodies found in patients with rheumatoid arthritis (RA). The presence of serum RFs is one of the criteria used for the diagnosis of RA [2]. The biological significance has been studied in normal and diseased states [3], including the relation between synthesis of RFs and infection by bacteria or virus [3–6]. RFs are detected in serum of patients with RA and other diseases [7]. RFs are composed of different classes of immunoglobulins including IgG, IgA, IgM, IgD and IgE. Serum RFs consist of many autoantibodies that can bind to various epitopes of IgG molecules or other immunoglobulins, but the IgM class has been measured for clinical purposes in many instances [3]. For measurement of RFs, a radioimmunoassay [8–10], enzyme immunoassay [11,12] and nephelometric assay have been developed for routine analysis [13–16]. Recently, a fully automated nephelometric assay based on the reaction of IgG or gamma globulin-coated latex particles with RFs has been widely used. In our laboratory, 12 serum analytes including RFs have been measured by a nephelometric assay using the Behring Nephelometer Analyzer (BNA) (Dade Behring Marburg Gmbh). However, the BNA has been replaced with the Behring Nephelometer II Analyzer (BNII) (Dade Behring Marburg Gmbh) for routine assay, and we have found that the RF values obtained with the BNII were lower as compared with those obtained with the BNA, even though we used the same assay parameter, reagent and calibrator in all measurements. The only difference in the assay condition between the two methods was the reaction temperature, as the assay with the BNA being carried out at room temperature (25618C) and that with the BNII at 378C. Here, we describe a plausible explanation of the difference in RF values obtained with the BNA and BNII, including the presence of low avidity autoantibodies [17], whose reactivities are inhibited by addition of urea, a mild protein-denaturing agent. [18–21].
2. Materials and methods
2.1. Serum samples For the RF assay, serum samples from 99 patients (69 patients with RA and 30 patients with other diseases) examined in our hospital were analyzed. Whole blood samples were centrifuged after clotting at room temperature. The separated sera were assayed immediately without storage at 48C.
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
119
2.2. Methods for specimen analysis Serum RFs were measured by the BNA and BNII according to the manufacture’s instruction using ‘N-latex RF kit II’ (Dade Behring Marburg Gmbh). This commercial kit consists of two reagents, one is latex particles coated with human IgG-sheep anti-human IgG complex, and another is a supplementary reagent containing polyethyleneglycol and NaCl. In a modified BNA assay for RFs, the ‘N-Diluent’ (dilution buffer for the BNA and BNII) containing urea (0.06–8 mol / l) was used as the dilution buffer. The final dilution of urea added to the dilution buffer was 2.6-fold in the reaction mixture. The CRP concentration was measured by a routine turbidimetric assay using goat anti-CPR antibody on the Hitachi-747-400 (Roche).
3. Results As shown in Fig. 1, the correlation between RF values measured with the BNII ( y) and BNA (x) was y 5 0.76x 2 5.7 kIU / l (r 5 0.997, slope 95% confidence limit 5 0.75 to 0.77, intercept 95% confidence limit 5 2 27.6 to 16.3). The mean RF values measured with the BNA and BNII were 291.7 kIU / l and 215.5 kIU / l, respectively. The mean bias (BNA minus BNII) by using the Bland and Altman method was 52.7 (S.D.; 85.5, 95% confidence limits; 2 7.6 to 322.6). This bias was not constant throughout the range as shown by the slope of 0.76. This problem was related specifically to the RF assay, as there were no biases for the other nephelometric assays (transferrin, ceruloplasmin, haptoglobin, etc.) between these two analyzers (data not shown). To examine the influence of complement, ethylenediaminetetraacetate (EDTA) was added into serum specimens, and serum RF values were measured with the BNA and BNII. There was no significant change in the measured values with either method, irrespective of whether EDTA was added or not. The ratio of the RF value obtained with the BNII to that obtained with BNA (BNII / BNA) was calculated for each sample. As shown in Table 1, the mean BNII / BNA ratio from patients with RA was 0.752 and that from control subjects was 0.665. However, the difference was not statistically significant by the Mann–Whitney test. By using the dilution buffer containing urea (modified BNA assay), serum RF values were measured with the BNA in five samples, where the BNII / BNA ratio ranged from 0.42 to 1.07 (Table 2). At the concentration of 8 mol / l, the measured results were all under the detection limit (11.0 kIU / l). At the concentration of 2 mol / l, positive values were detected only with two samples (sample A,B), where the BNII / BNA ratio was 1.07 and 0.95, respectively. The
120
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
Fig. 1. Comparison of RF concentrations measured by the BNA and BNII. (A) The correlation between values measured by the BNA and BNII. Regression equation: y 5 0.76x 2 5.7, r 5 0.997, Sy /x 5 60.7, n 5 99. Dotted line indicates y 5 x. (B) The Bland and Altman plot obtained from two analyzers (overall mean difference6S.D.; 52.7685.5).
values were still , 11.0 kIU / l at the concentration of 1 mol / l with the samples having the BNII / BNA ratio of 0.52 and 0.42 (sample D,E). At the concentration of 0.25 mol / l or less, all results were . 11.0 kIU / l, and the mean ratio of the
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
121
Table 1 Concentration of RFs measured by the BNA and BNII and the BNII / BNA ratio in various diseases Mean RF conc. (kIU / ml)
BNII / BNA
n
BNA
BNAII
Mean
S.D.
Range
Rheumatoid arthritis
69
356.2
272.7
0.752
0.202
0.23–1.18
Control subjects Systemic lupus erythematosus Sjogren’s syndrome Progressive systemic sclerosis Myeloma Liver disease Other Total
6 3 4 5 5 7 30
118.4 210.8 142.9 106.3 204.3 118.9 143.3
72.3 78.6 78.8 78.3 105.2 89.9 84.0
0.663 0.480 0.683 0.784 0.572 0.732 0.665
0.213 0.151 0.295 0.196 0.127 0.230 0.216
0.36–0.97 0.34–0.64 0.49–1.12 0.56–0.99 0.46–0.78 0.34–1.03 0.34–1.12
detected values to the initial values was 91.2% at the concentration of 0.06 mol / l. Fig. 2 shows the relation between the BNII / BNA ratio and the rate of decrease in the RF values (%) as measured by the modified BNA assay in the presence of urea at the concentration of 0.5 mol / l. A significant relation was observed between the BNII / BNA ratio and the rate of decrease (r 5 2 0.652, 95% confidence limits; 0.314–0.848, P , 0.01). There was no significant difference between the BNII / BNA ratio from patients with RA and control subjects (Table 1). As shown in Fig. 3, however, a significant difference was observed (P , 0.01), when they were divided into two groups according to the serum CRP concentration above and below 2 mg / l. The mean BNII / BNA ratio was 0.780 (S.D., 0.202) in patients with CRP conTable 2 Changes in RF values (kIU / l) by addition of various concentrations of urea to the assay with the BNA Sample Initial RF conc.
BNII/BNA
Urea added (mol/l)
(kIU/l) 0.06
A
0.125
BNA
BNII
99.7
107
1.07
103 (103.3)a
149 (102.1) 140 (95.9)
B
146
139
0.95
C
106
68
0.64
98.2 (92.6)
91.7 (92.0) 85.1 (80.3)
0.25
82.8 (83.0) 119 (81.5) 76.7 (72.3)
0.5
68.2 (68.4) 102 (69.9) 56.0 (52.8)
D
36.6
19
0.52
28.1 (76.8)
29.0 (79.2)
21.5 (58.7) , 11.0
E
47.2
20
0.42
38.4 (81.4)
34.3 (72.7)
25.4 (53.8)
a
1
2
8
44.8 (44.9) 23.2 (23.3) 74.6 (51.1)
37.7 (25.8)
30.0 (28.3) , 11.0
, 11.0 , 11.0 , 11.0
, 11.0
, 11.0
, 11.0
12.6 (26.7) , 11.0
, 11.0
, 11.0
Percentage of the initial concentration measured with the BNA is shown in the parenthesis.
122
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
Fig. 2. Relation between the BNII / BNA ratio and the rate of decrease of RF values before and after addition of urea. Rate of decrease,%: (1-value treated with urea / original value) 3 100.
centrations of . 2.0 mg / l and 0.658 (S.D., 0.199) in subjects whose CRP concentrations were below the cut off value of 2 mg / l.
4. Discussion Although many studies have been reported regarding the reactivity of human RFs, the mechanism or specificity of the reaction is not entirely clear [3]. In the measurement of serum RFs for clinical purposes, the measured values sometimes have significant bias based on the reagents or calibrators used for the assay. Even using the same reagent, same calibrator and same assay parameter in the present study, different values were obtained between measurement with the BNA and BNII. The reduction of values by the BNII was not caused by the complement reaction, because there was no significant change in RF values measured either with the BNA or BNII before and after the activator was chelated with EDTA. All RF measurements with the BNA and BNII were carried out without storage at 48C, which might lead to an increase in the RF titers [17]. The slope of the regression line between the BNA assay and the BNII assay
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
123
Fig. 3. Comparison of BNII / BNA ratio in patients with normal CRP levels and above the cut-off value. Group A: Mean values for BNII / BNA ratio6S.D. 5 0.65860.199 in patients with CRP concentration # 2.0 mg / l. Group B: Mean values for BNII / BNA ratio6S.D. 5 0.78060.202 in patients with CRP concentration . 2.0 mg / l. The significant differences are shown in both groups (P , 0.01) by the Mann–Whitney test.
was 0.76, but the bias (BNA minus BNII) and BNA / BNII ratio were not constant (Fig. 1). Therefore, the difference appears to be due to the reactivity characteristic of RFs. These data suggest the presence of autoantibodies which were measured as RFs and whose reactivities could be changed by some factors even when the same reagent, calibrator and assay parameters were used. The notable discrepancies between the BNA and the BNII assay were observed not only in patients with RA, but also in patients with other diseases (Table 1). Anti-nuclear and anti-nucleolar antibodies are the autoantibodies found in a high proportion of patients with systemic lupus erythematosus, Sjogren’s syndrome and progressive systemic sclerosis. It was not clear whether these autoantibodies were measured as RF with the BNA or not. However, it was considered that the discrepancies between both methods might be caused by the variation in the reactivity of these autoantibodies including RFs. Therefore we tried to suppress the reaction caused by low avidity autoantibodies by addition of urea. Guanidine, urea and chaotropic ions (SCN 2 , I 2) which cause reversible denaturation of protein structures, are known to inhibit the formation of immune complexes [21]. It has been reported that urea dissociates a low avidity antibody
124
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
from an antigen as shown in the enzyme immunoassay for detection of anti-hepatitis C virus antibodies [18–20], and that guanidine hydrochloride is useful as a protein-denaturing agent for estimation of the avidity of antibodies in the enzyme-linked immunosorbent assay [21]. Although the initial concentration of RFs measured with the BNA for sample C (BNII / BNA 5 0.64) was close to that for sample A (BNII / BNA 5 1.07) (Table 2), a value above the cut-off level was observed with sample A but not with sample C at 2 mol / l urea. This suggests that the amount of RFs in serum inhibited by urea did not depend on the initial concentration of RFs. The result shown in Fig. 2 suggested that the effect of inhibition by urea was more notable in samples with low BNII / BNA ratios. This indicates that these samples might have a relatively large amount of RFs to be inhibited by urea. On the other hand, there was a good correlation between the values measured in the presence of urea by the modified BNA assay ( y) and the initial values on BNII (x) with y 5 0.97x 2 37.9 (r 5 0.999) (graph not shown). However, it is difficult to conclude that only the RFs which could not be detected by BNII were inhibited by the addition of urea with the modified BNA assay, because the mean ratio of decrease was 32.0% even in the samples with the high BNII / BNA ratio . 0.90 (mean;0.96) as shown in Fig. 2. Based on these results, we conclude that RF concentrations measured with the BNA were higher than those with the BNII, in which the low avidity RFs inhibited by urea might have been measured by the BNA. These antibodies may be temperature dependent, as this was the only difference in the assay condition between the two instruments. The reaction of antibodies and antigens is faster at 378C than at room temperature. On the other hand, immune complexes of low avidity or non-specific antibodies and antigens may be more stable at room temperature. The BNII / BNA ratio for measurement of RFs was significantly lower in patients with elevated CRP concentration ( . 2.0 mg / l), suggesting that the presence of these antibodies in serum may be dependent on the degree of inflammation, where synthesis of these autoantibodies to be suppressed by urea may be suppressed in the active inflammatory state. For clinical testing, these problems must be resolved in order to establish an accurate assay method for RFs. We believe that RF values measured at 378C are more useful for clinical purpose, especially for the diagnosis of RA.
References [1] Dorner RW, Alexander RL, Moore TL. Rheumatoid factor. Clin Chim Acta 1987;167:1–21. [2] Arnett FC, Edworthy SM, Bloch DA, Mcshane DJ, Fries JF, Cooper NS et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24.
H. Yamanishi et al. / Clinica Chimica Acta 292 (2000) 117 – 125
125
[3] Moore TL, Dorner RW. Rheumatoid factor. Clin Biochem 1993;26:75–84. [4] Slaughter L, Carson DA, Jenson FC, Holbrook TL, Vaughan JH. In vitro effects of Epstein-Barr virus on peripheral blood mononuclear cells from patients with rheumatoid arthritis and normal subjects. J Exp Med 1978;148:1429–33. [5] Hirohata S, Inoue T, Miyamoto T. Frequency analysis of human peripheral blood B cells producing IgM-rheumatoid factor. Differential effects of stimulation with monoclonal antibodies to CD3 and Staphylococcus aureus. J Immunol 1990;145:1681–6. [6] Karsh J, Goldstein R, Lazarovits AT. In vitro IgM and IgM rheumatoid factor production in response to Staphylococcus aureus Cowan I and pokeweed mitogen: The contribution of CD5 1 (Leu 1) B cells. Clin Exp Immunol 1989;77:179–83. [7] Shmerling RH, Delbanco TL. The rheumatoid factor: An analysis of clinical utility. Am J Med 1991;91:528–34. [8] Nineham LJ, Hay FC, Roitt IM. Laboratory diagnosis of rheumatoid arthritis: a solid phase radioassay for IgG and IgM antiglobulins. J Clin Pathol 1976;29:1121–6. [9] Knez V, Reimer CB. An indirect solid-phase microradioimmunoassay for human IgM-antiIgG (rheumatoid factor). J Immunol Meth 1977;18:105–21. [10] Wernick R, Lospalluto JJ, Fink CW, Ziff M. Serum IgG and IgM rheumatoid factor by solid phase radioimmunoassay. Arthritis Rheum 1981;24:1501–11. [11] Maiolini R, Ferrua B, Quaranta JF, Pinoteau A, Euller L, Ziegler G et al. A sandwich method of enzyme-immunoassay II. Quantitation of rheumatoid factor. J Immunol Method 1978;20:25–34. [12] Teitsson I, Valdimarsson H. Use of monoclonal antibodies and F(ab9) 2 enzyme conjugates in ELISA for IgM. IgA and IgG rheumatoid factors. J Immunol Method 1984;71:149–61. [13] Deaton CD, Maxwell KW, Smith RS, Creveling RL. Use of laser nephelometry in the measurement of serum proteins. Clin Chem 1976;22:1465–71. [14] Virella G, Waller M, Fudenberg HH. Nephelometric method for determination of rheumatoid factor. J Immunol Method 1978;22:247–51. [15] Finley PR, Hicks MJ, Williams RJ, Hinlicky J, Lichti DA. Rate nephelometric measurement of rheumatoid factor in serum. Clin Chem 1979;25:1909–14. [16] Painter PC, Lyon JM, Evans JH, Powers WW, Whittaker RL, Decker MJ. Performance of a new rate-nephelometric assay for rheumatoid factor, and its correlation with tube-titer results for human sera and synovial fluid. Clin Chem 1982;28:2214–8. [17] Ueda K, Nakagawa T, Shimizu A. Rheumatoid factor in the serum of hepatitis C virusinfected patients: an increase in the titre during cold storage. Ann Clin Biochem 1996;33:438–42. [18] Goeser T, Blazek M, Gmelin K, Kommerell B, Theilmann L. Washing with 8 mol / l urea to correct false-positive anti-HCV results [Letter]. Lancet 1990;335:878. [19] Wreghitt TG, Gray JJ, Aloyisus S, Contreras M, Barbara JAJ. Antibody avidity test for recent infection with hepatitis C virus [Letter]. Lancet 1990;335:789. [20] Gray JJ, Wreghitt TG, Friend PJ, Wight DGD, Sundaresan V, Calne RY. Differentiation between specific and non-specific hepatitis C antibodies in chronic liver disease [Letter]. Lancet 1990;335:609–10. [21] Inoue S, Hasegawa A, Matsuno S, Katow S. Changes in antibody avidity after virus infections: Detection by an immunosorbent assay in which a mild protein-denaturing agent is employed. J Clin Microbiol 1984;20:525–9.