Original Article Ann Clin Biochem 1999; 36: 365-371 Detecting anti-ssa and anti-ssb antibodies in routine analysis: a comparison between double immunodiffusion and immunoblotting Martin Yagol, Miguel A Belmonte", Maria J Olmos", Juan Beltran-, Carlos Teruel' and Marian Segarra 3 From the ILaboratory 01" Biochemistry, the 2Rheumatology Unit and the 'Service 01" Internal Medicine, Hospital General, Avda. de Benicasim sin, 12004 Castellon, Spain The aim of this study was to assess the performance of a commercially SUMMARY. available procedure for detecting anti-sjogren's syndrome A (anti-ssa) and anti Sjogren's syndrome B (anti-ssb) antibodies by immunoblotting (I B) and compare it with double immunodiffusion (DID). We also studied the clinical significance of these profiles in a series or unselccted anti-ssa positive patients. Serum samples from 534 patients that were positive on an immunofluorescent screening test using HEp-2 cells were analysed for anti-ssa and anti-ssb antibodies by DID and IB (Biolab Anablot System II), and the results on anti-ssa antibodies were confirmed by an enzyme-linked immunosorbent assay (ELISA). Fifty-five serum samples were found to be positive for anti-ssa antibodies. Among these, 24 were anti-ssa negative by IB but positive by DID and ELISA ('non-blotter sera'), whereas only three serum samples were anti-ssa negative by DID but positive by lb and ELISA. Of the 18 anti-ssb positive serum samples, eight were negative by DID. All the serum samples that were anti-ssb positive by DID were also positive by lb. Anti-SSB antibodies showed a significant association with eye dryness and leucopenia. Anti-52 kda SSA antibodies were associated with anti-ssb antibodies but showed no significant association with sicca symptoms, while anti-60 kda SSA antibodies were associated with lower rates of leucopenia. The 'non-blotter' profile showed no significant association with any clinical parameter. lb is less sensitive than DID for detecting anti-ssa antibodies but more sensitive than DID for detecting anti-ssb antibodies. The determination of anti-ssa immunoblotting profiles in patients positive for anti SSA antibodies by DID does not significantly improve the clinical usefulness of this test. As expected, anti-ssb antibodies were associated with clinical features of Sjogren's disease. Non-blotting (probably conformational) anti-ssa antibodies did not show any further association with clinical parameters and seem to have no clinical relevance. Additional key phrases: autoantibodies; connective tissue diseases; systemic lupus erythematosus; Sjogren's syndrome Anti-Sjogren's syndrome A (anti-ssa, anti-ro) and anti-sjogrcns syndrome B (anti-ssb, anti La) antibodies are frequently found in the serum of patients with systemic lupus erythematosus (SLE), Sjogren's syndrome (SS) and, more rarely, in patients with other connective tissue diseases. I 4 These antibodies have traditionally been identified in the clinical laboratory by double immunodiffusion in agarose gel (010).5 Correspondence: Dr Martin Yago. More recently, the introduction of the immunoblotting method (lb) has allowed laboratories to obtain more detailed information about their antigenic specificity. Anti-SSA antibodies react with either a 52 kda or a 60 kda polypeptide in immunoblots, whereas anti-ssb antibodies recognize a 48 kda polypeptide." Several anti-ssa antibodies specifically recognize the native SSA antigen (i.e. in DID or immunoprecipitation assays) but fail to bind the corresponding denatured polypeptides in immunoblots.? 365
366 Yago et al. Anti-60 kda SSA antibodies are mainly directed against the native protein whereas anti-52 kda SSA antibodies are mainly directed against the denatured protein." Serum containing antibodies against the denatured 52 kda protein also contains antibodies against the native 60 kda protein because antibodies against the denatured 52 kda protein are apparently a subpopulation of antibodies to the native 60 kda protein." This cross-reactivity, however, has not been confirmed using recombinant 60 kda and 52 kda proteins. 10 Several published studies on the clinical significance of the different patterns of anti SSA antibodies suggest the existence of a correspondence between different antibody patterns and different clinical syndromes. Anti 52 kda antibodies are found more frequently in patients with SS than in patients with SLE or rheumatoid arthritis (RA), whereas anti-60 kda antibodies are more common in patients with SLE than in patients with primary SS or RA.11 The incidence of anti-52 kda antibodies without anti-60kda antibodies is lower in patients with SLE than in patients with SS, whereas anti 60 kda antibodies without anti-52 kda antibodies are more common in SLE patients than in SS patients. 12.1J In order to determine the possible advantages of immunoblotting in routine laboratory work, we evaluated the analytical sensitivity and specificity of this procedure for detecting anti SSA and anti-ssb antibodies in serum samples of unselected patients, and compared it with the sensitivity and specificity of DI D. We also attempted to establish associations between the patterns of anti-ssa antibodies as determined by IB and the clinical syndromes or disease manifestations. MATERIALS AND METHODS Patients Between August 1992 and July 1996, serum samples from about 4800 patients were received in our laboratory, for routine determination of antinuclear antibodies. The laboratory is in a 600-bed hospital and analyses samples for autoantibodies submitted from different specialized units of the hospital (including rheumatology, nephrology, dermatology, haematology, cardiology, allergy and hepatology), several primary care centres (about 40'1., of samples) and three smaller hospitals located in the same district. The serum samples were analysed by indirect immunofluorescence (IIF) using HEp-2 cells, and screened for the presence ofnuclear or cytoplasmic fluorescence. Serum samples from 534 patients showed nuclear (462 serum samples) or cytoplasmic staining (72 serum samples), and were further evaluated by DID and IB for autoantibodies to SSA and SSB antigens. Of these 534 patients, 55 were positive for anti-ssa antibodies by DID and/or IB and the positive results were confirmed by enzyme-linked immunosorbent assay (ELISA). These 55 patients were initially selected for the clinical studies. In order to assess the disorders associated with the different autoantibody profiles, clinical and biochemical data on the patients with anti SSA antibodies were collected by reviewing their clinical records. In addition to the eleven American Rheumatism Association criteria for SLE,14 other clinical symptoms including alopecia, Raynauds phenomenon, morning stiffness, subcutaneous noduli, erosions, mouth dryness, eye dryness, scleroderma, myositis and oesophageal dysmotility were recorded. Symptoms present at the time when the analysis of autoantibodies was requested and in the past were taken into account. Forty-one patients were included in the study; 14 out of the 55 patients who were positive for anti-ssa antibodies were excluded because the clinical records were unavailable (4 patients) or incomplete (10 patients). Analytical methods Indirect immunofiuorescence The serum samples were diluted 1/40 in phosphate-buffered saline (PBS) (10 mrnol/l sodium phosphate, 150 mrnol/l NaCI, ph 7-4) and incubated on HEp-2 cells (Inova Diagnostics, San Diego, CA, USA) for 30 min at room temperature. After being washed (twice for 5 min each) with PBS the slides were incubated with fluorescein-labelled rabbit anti-human immunoglobulin G (IgG) (Inova Diagnostics) for 30 min at room temperature. The slides were washed with PBS again and read usmg a fl uorescencc microscope. Double immunodiffusion Precipitating antibodies against extractable nuclear antigens were detected by double immunodiffusion in Ouchtcrlony plates (Nova Gel plates, Inova Diagnostics) using rabbit thymus extract (MarDx, Carlsbad, CA, USA) for SSB antigen and primate spleen extract (MarDx) for SSA antigen. Results for serum samples showing Ann elin Biochem 1999: 36
Double immunodiffusion and immunoblotting [or anti-ssa and anti-ssb antibodies 367 multiple reactivities or overlapping precipitation lines were confirmed using partially purified antigens (Inova Diagnostics). Imrnunoblotting We used the Anahlot System II kit (Cat. No. ZAB02, Gull, Lirnal, Belgium) and followed the instructions of the manufacturer exactly. Membrane strips containing the nuclear antigens and five precoloured calibration bands were incubated for 15min with ()'5mL of a solution of protein to block the non-specific protein binding sites. The blocked strips were incubated for 15min with I ml of a IjI 00 dilution of patient's serum or control. After being washed four times the strips were incubated for 15min with a solution of peroxidase-labelled rabbit antihuman immunoglobulins. The strips were washed again to eliminate the excess of antiserum, incubated for 15min with a peroxidase chromogen solution and washed with distilled water to stop the reaction. All incubation and washing steps were performed at room temperature on a shaking platform. The presence of bound autoantibodies makes blue-green bands appear on the membrane strips. The location of the bands depends on the molecular weight of the antigen involved and allows autoantibody identification. Serum samples containing anti-ssa antibodies recognize either a 57 5 kda or a 48 3 kda antigen on the membrane strips (corresponding to the described 60 kda and 52 kda polypeptides), whereas anti SSB serum samples produce a characteristic profile composed of one band at 44 5 kda, one triplet at about 42 kda and one doublet at about 30 kda, representing breakdown products of a 48 kda antigen (completely converted)." The results were read visually by aligning the five violet calibration bands of the test strips with those of the control strips and then carefully comparing any blue-green band present on the test strips with those on the control strips. ELISA To further confirm the results obtained by double immunodiffusion and immunoblotting, all the serum samples from patients that were found positive for anti-ssa antibodies by any of these procedures were also analysed by ELISA, using commercially available reagents (Quanta Lite SS-A, Inova Diagnostics) and following the manufacturer's instructions. Serum samples found positive by at least two of the three procedures were considered true positives. Because of the very characteristic profile produced by anti-ssb antibodies when analysed by immunoblotting, an ELISA for these antibodies was not done to confirm the results. Statistics Differences in antibody responses between groups of patients with different clinical diagnoses or defined clinical symptoms were assessed using the chi-squared and Fisher exact test. No correction for multiple comparisons was performed. RESULTS Of the 534 patients whose serum samples were found to produce a nuclear or cytoplasmic stain of HEp-2 cells on indirect immunofluorescence, 55 were positive for anti-ssa antibodies by at least two out the three procedures utilized. Anti-SSA and anti-ssb antibodies detected by double immunodiffusion Anti-SSA and anti-ssb antibodies were detected by DID in the serum samples of 52 and 10, patients respectively (see Table I). All the serum samples containing anti-ssb antibodies were also found to contain anti-ssa antibodies. All the serum samples positive for anti-ssa antibodies by DID were also positive by ELISA. The serum samples of three patients were negative for anti-ssa antibodies by DID and positive by IB and ELISA, but they showed a weak reaction. Eight out of the 18 serum samples that were positive for anti-ssb by IB were negative by DID (see Table I), while all the serum samples that were anti-ssb positive by DID were also found positive by lb. TABLE I. Anti Sjogren's syndrome A (SSA) and anti Sjagren'«syndrome B (SSB) antibodies detected by immunohlotting (IB) and double immunodiffusion (DID) in 534 serum samples positive hy indirect immunofluorescence Antibody specificity IB(+) IB(-) SSA DlD(+) 28 24 DlD(- ) 8* 474 SSB DlD(+) 10 0 DlD(-) 8 516 *Five serum samples were found to be negative by enzyme-linked immunosorbent assay. Ann cu«biochem 1999: 36
368 Yago et al. Anti-Sjogren '.I' syndrome A (SSA) profiles and double immunodiffusion (DID) results for 55 patients' serum samples positive!;)r anti-ssa anlihodies TABLE 2. DID(+) DID(-) Only Only 60 kda 60 kda 52 kda and 52 kda Non-blotter II I 5 2 Anti-SSA and anti-ssb antibodies detected by immunoblotting Anti-SSA and anti-ssb antibodies were detected by IB in the serum samples of 36 and 18 patients, respectively (see Table I). Twenty-four out of the 52 serum samples that were found to contain anti-ssa antibodies by DID and ELISA were negative by IB Cnonblotter sera'). On the other hand, five samples of the 534 IIF-positive serum samples analysed were classified as positive by IB, but these results were not confirmed by DID nor by ELISA. Of the 31 serum samples that were positive for anti-ssa antibodies by IB and ELISA, 12 reacted only with the 60 kda polypeptide, 7 reacted only with the 52 kda polypeptide and 12 reacted with both the 60 kda and 52 kda polypeptides (see Table 2). 24 o Clinical studies The anti-ssa antibody profiles as detected by IB and selected clinical features for the 41 patients 12 o included in the study are shown in Table 3. These patients were divided into three subgroups according to their immunoblotting profile: those reacting only to the 60 kda antigen (anti-60 kda profile, n = 10), those reacting to the 52 kda antigen [anti-52 kda profile, 11= 12, grouped according to whether they had (11 = 9) or did not have (n = 3) associated anti-60 kda reactivity], and those without immunoblotting reactivity (non-blotter profile). The anti-52 k Da profile showed no association with any clinical symptom, except a nonsignificant tendency to have Raynaud's phenomenon (P= 0'07), whereas the incidence of leucopenia was significantly lower in patients with the anti-60kda profile than in those with the 52 kda profile (P = 0 0 I). There was an association between the presence of eye dryness and anti-ssb antibodies, as detected either by DID (P=0'008) or IB (P = 0'007). Anti-SSB antibodies were also associated with leucopenia, as detected either by IB (P=0'04) or DID (P=0 006). A significant association was also found between anti-52 kda and anti-ssb antibodies (P< 0 00 I). Anti-SSB antibodies were detected by immunoblotting in 75% of patients with anti 52 kda antibodies but in only 18% of anti 52 kda negative patients. However, no association was found between the anti-52 kda profile and eye dryness (P= 0'13) or mouth dryness (P=0-46). TABLE 3. Association hetween the presence ofanti-sjogren's syndrome A (SSAj and anti-sjogren's syndrome B (SSB) antibodies as detected by immunoblotting in patients' serum samples anti-ssa positive hy enzyme-linked immunosorbent assay (ELISA) (n ~ 4/) and selected clinical features 60 kda profile 52 kda profile Non-blotter SSB Clinical symptom* (n= 10) (n~ 12) (n= 19) (11= 14) Malar rash 2 3 7 4 Photosensitivity 2 3 8 7 Oral ulcers I I 3 I Arthritis 6 6 10 7 Pleuritis 2 2 3 0 Nephritis 2 3 5 2 Anaemia 4 2 9 4 Leucopenia 0 6 6 7 Lymphopenia 3 4 4 3 Thrombocytopenia 0 3 5 4 Raynaud's phenomenon I 5 2 5 Morning stiffness 3 3 4 2 Eye dryness 3 8 9 II Mouth dryness 3 5 7 8 *Clinical symptoms present in a low number of patients (in parenthesis) are not tabulated: discoid lesions (2), pericarditis (4), central nervous system disease (2), neuritis (2), psychosis (I), myositis (I), oesophageal dysmotility (1), erosions (0), subcutaneous noduli (2), alopecia (4), scleroderma (4). Ann Clin Biochem 1999: 36
Double immunodiffusion and immunoblotting [or anti-ssa and anti-ssb antibodies 369 The 'non-blotter' profile for anti-ssa antibodies (DID-positive but IB-negative serum) showed no statistically significant association with any clinical parameter. DISCUSSION In IB, antibody specificity is established by determining the molecular weight of the denatured antigen or antigens recognized by the antibody. When non-purified cellular extracts are used as the antigen source, antibodies with different specificities may react with antigens of the same molecular weight, and this can lead to erroneous results. Although reliability increases in eases in which the antibody recognizes several polypeptides and gives characteristic patterns on immunoblots (i.e., anti-ssb antibodies in this study), the interpretation of the results always has an inductive rather than deductive character. On the other hand, it is known that anti-ssapositive serum samples often recognize exclusively the native SSA antigen and fail to bind to the denatured antigen in immunoblots.? These limitations of immunoblotting probably account for the low analytical sensitivity and specificity we have found for this procedure in the detection of anti-ssa antibodies. A high percentage of anti-ssa positive serum samples appear to recognize only conformational determinants of the SSA antigen, and give negative results by IB but positive by DID and ELISA. In contrast, some serum samples that were apparently positive for anti-ssa antibodies by IB were negative when analysed by DID and ELISA. These serum samples presumably contain antibodies that react with antigens other than the 52 kda and 60 kda SSA polypeptides but with about the same apparent molecular weight. On the other hand, we have found that a large proportion of anti-ssb-positive serum samples by IB were negative when tested by DID. Therefore, IB is more sensitive than DID for detecting anti-ssb antibodies. These results are in agreement with those published by several authors." IH However, Uyttenbroeck ct al.,19 in a study comparing detection of antibodies to extractable nuclear antigens by DID and IB, have reported that IB is more sensitive than DID in determining anti SSA antibodies in a series of 100 anti-nuclear antigen (AN A)-positive serum samples, and that most serum samples found negative by DID but positive by IB reacted only with the 60 kda polypeptide. In contrast to this study, and in spite of our using the same reagents as these authors for IB, we found only three serum samples anti-ssa-positive by IB and ELISA but negative by DID; and of these only one reacted with the 60 kda polypeptide. The reason for this discrepancy is not clear but we have observed that about 5% of the ANA-positive serum samples analysed produce bands at about 56 kda on immunoblots, and they could be classified erroneously as positive for anti-ssa antibodies if test and control strips are not carefully compared. These samples are invariably negative for anti-ssa antibodies when tested by DID and ELISA (data not shown). Unfortunately, Uyttenbroeck et al. 19 did not perform any independent confirmatory analysis on their samples. In early reports on the occurrence of autoantibodies to the 52kDa and 60kDa polypeptides in patients with SLE and SS, reactivity with the 60 kda polypeptide alone (without reaction with the 52 kda polypeptide) in immunoblots was found exclusively in SLE serum samples, whereas reactivity with the 52 kda polypeptide alone was detected only in SS serum samples.pr'? In our study, however, patients with a diagnosis of SLE were found with about the same frequency in each serological subgroup (see Table 4). These results partially agree with more recent studies 11 12, 16,2 1,22 in which anti-52 kda antibodies alone were also found in SLE serum samples, and anti-60 kda antibodies alone were also detected in the serum samples of some patients with rheumatoid arthritis and patients with undefined connective tissue disease. We have found that anti-52 kda antibodies were associated with anti-ssb antibodies but showed no significant association with sicca symptoms, in agreement with the results obtained by Zimmermann et al. 16 in SLE patients. In contrast to the findings of these authors, malar rash was not associated with the nonblotting subgroup in our series of patients. TABLE 4. Percentage ofpatientsfulfilling four or more American Rheumatism Association (ARA) criteria for systemic lupus erythematosus in each serological suhgroup Anti-SSA profile Anti-52kDa (n= 12) Anti-60 kda (n = 10) Non-blotter (n = 19) SSA = Sjogren's syndrome A. % of patients fulfilling four or more ARA criteria 33 40 32 Ann CUn Biochem 1999: 36
370 Yago et al. In most of the studies cited above the association between antibody patterns and clinical syndromes or disease manifestations has been investigated by using clinically predefined groups of patients - i.e., the patients were selected on the basis of their clinical diagnosis or clinical symptoms, and the frequencies of the different antibody profiles in the various patient groups were compared. However, from the laboratory point of view, it would be more useful to know whether the finding of a particular antibody profile in a serum sample routinely analysed for antinuclear antibodies can be related to the presence of a defined clinical syndrome or specific disease symptom. The fact that a large proportion of patients presenting the same disease or clinical symptom may have the same autoantibody profile does not mean that the converse is also true. Because ofthe poor disease specificity ofanti SSA antibodies and their widespread distribution, the composition of a group of patients selected on the basis of the presence of anti-ssa antibodies in their serum tends to be clinically heterogeneous, depending on the prevalence of the different anti-ssa associated diseases in the patient population. In addition, the number of disease symptoms present can be influenced by the duration of the disease. Therefore, it is not surprising that some associations established using clinically predefined groups of patients are often not confirmed when the patients are selected at random solely on the basis of the presence in their serum of a particular antibody. For example, Swaak et al. 23 observed no correlation between the presence of anti-ssa antibodies, proposed by others" as a diagnostic criterion for Sjogren's syndrome, and specific clinical symptoms, including sicca symptoms, in a series of 339 patients selected on the presence of antinuclear antibodies in their serum samples. We also have found a strong association between anti-ssb antibodies and the presence of eye dryness, which confirms previous results." In the group of patients studied, anti-ssb antibodies were detected by immunoblotting in five patients anti-ssb-negative by DID. Of these, two fulfil more than three ARA criteria for SLE and do not have symptomatic sicca syndrome, two have a diagnosis of primary Sjogren's syndrome and one has rheumatoid arthritis and sicca symptoms. These results sustain the observation that IB is more sensitive than DID for detecting anti-ssb antibodies in SLE and SS serum." CONCLUSION Immunoblotting should not be used as a screening procedure for detecting anti-ssa antibodies because of its high false-negative rate. Although some associations were found between the clinical features and immunoblotting profiles of anti-ssa-positive patients, they did not seem very relevant for clinical practice. In particular, nephritis was not associated with any anti-ssa specificity. As expected, anti-ssb antibodies were associated with clinical features of Sjogren's disease. Immunoblotting is more sensitive than DID for detecting anti-ssb antibodies but it is a laborious procedure, difficult to automate, and therefore does not seem appropriate for routine laboratory work. However, it may have a role as a complementary assay for further characterization of selected serum samples previously analysed by a more practicable routine procedure, such as gel diffusion or ELISA techniques. REFERENCES Scopelitis E, Perez M, 8iundo JJ. Anti-SSA(Ro) antibody: a connective tissue disease marker. J Rheumatol1985; 12: 1105-8 2 Harley 18, Scofield RH, Reichlin M. Anti-Ro in Sjogren's syndrome and systemic lupus erythematosus. Rheum Dis Clin North Am 1992' 18: 337 58 ' 3 St Clair EW. Anti-La antibodies. Rheum Dis C/in North Am 1992: 18: 359-76 4 Tan EM. Antinuclear antibodies: Diagnostic markers for autoimmune diseases and probes for cell biology. Adv Immunol1989; 44: 93-151 5 Alspaugh M, Maddison P. Resolution of the identity of certain antigen-antibody systems in systemic lupus erythematosus and Sjogren's syndrome: an interlaboratory collaboration. Arthritis Rheum 1979: 22: 796-8 6 8en-Chetrit E, Chan EKL. Sullivan KF, Tan EM. A 52 kd protein is a novel component or the SS AIRo antigenic particle. J Exp Med 1988; 167: 1560-71 7 Boire G, Lopez-Longo Fl, Lapointe S, Menard HA. Sera from patients with autoimmune disease recognize conformational determinants on the 60 kd RolSSA protein. Arthritis Rheum 1991: 34: 722-30 8 lloh Y, Reichlin M. Autoantibodies to the Ro/ SSA antigen are conformation-dependent. I. Anti 60 kd antibodies are mainly directed to the active protein; anti-52 kd antibodies are mainly directed to the denatured protein. Autoimmunity 1992; 14: 57--65 Ann eli" Biochem 1999: 36
Double immunodiffusion and immunoblotting for anti-ssa and anti-sse antibodies 371 9 Itoh Y, Itoh K, Frank MB, Reichlin M. Autoantihodies to the Ro/SSA antigen are conformation dependent. II. Antibodies to the denatured form of 52 kd Ro/SSA are a cross reacting subset of antibodies to the native 60 kd Ro/SSA molecule. Autoimmunity 1992; 14: 89-95 10 St Clair EW, Burch JA, Saitta M. Specificity of autoantibodies for recombinant 60 kd and 52 kd Ro autoantigens. Arthritis Rheum 1994; 37: 1373-9 II Lopez-Longo FJ, Rodriguez-Mahou M, Escalona M, Perez T, Monteagudo I, Sanchez-Atrio A, et at. Heterogeneity of the anti-ro(ssa) response in rheumatic diseases. J Rheumato/1994; 21: 1450-56 12 Tsuzaka K, Fujii T, Akizuki M, Mimori T, Tojo T, Fujii H, et al. Clinical significance of antibodies to native or denatured 60 kd or 52 kd Ro/SS-A proteins in Sjogren's syndrome. Arthritis Rheum 1994; 37: 88-92 13 Ben-Chetrit E, Fox RI, Tan EM. Dissociation of immune responses to the SS-A (Ro) 52 kd and 60 kd polypeptides in systemic lupus erythematosus and Sjogren's syndrome. Arthritis Rheum 1990; 33: 349-55 14 Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, et at. The 1982 revised criteria for the classification of SLE. Arthritis Rheum 1982; 25: 1271-7 15 Gull Lahoratories. Anablot system Il data sheet. Salt Lake City: Gull Laboratories, 1991 16 Zimmermann C, Smolen J, Graninger W, Petera P, Fabini G, Hassfeld W, et al. Fine specificity of anti Ro(SSA) autoantibodies and clinical manifestations in patients with systemic lupus erythematosus. J Rheumatol 1996; 23: 1897-903 17 Mei!of JF, Bantjes I, van Dam AP, Smeenk RJ. The detection of anti-ro/sfi-a and anti-la/ss-b antibodies. A comparison of counterimmunoelectrophoresis with irnmunoblot, ELISA and RNA-precipitation assays. J Immunol Methods 1990; 133: 215-26 18 Williams DG, Stocks MR, Charles PJ, Maini RN. Antibodies to La, Jo-I, nrnp and Sm detected by multi-track immunoblotting using a novel filter holder: a comparative study with counterimmunoelectrophoresis and immunodiffusion using sera from patients with systemic lupus erythematosus and Sjogren's syndrome. J Immunol Methods 1986; 91: 65-73 19 Uyttenbroeck W, Cooreman W, Scharpe S. The presence of antibodies against extractable nuclear antigens in serum: a comparison of immunoblotting versus radial immunodiffusion. Ann Clin Biochem 1993; 30: 83--{j 20 Slobbe RL, Pruijn GJM, Damen WGM, van der Kemp JW, van Venrooij WJ. Detection and occurrence of the 60 and 52 kd Ro (SS-A) antigens and of autoantibodies against these proteins. Clin Exp lmmunol 1991; 86: 99-105 21 Bozic B, Pruijn GJM, Rozman B, van Yenrooij WJ. Sera from patients with autoimmune diseases recognize different epitope regions on the 52-kD Ro/SS-A protein. Clin Exp Immuno/1993; 94: 227 35 22 Sibilia J, Goetz J, Berthel C, Neyrolles N, Javier RM, Imler M, et at. Anti Ro/SS-A antibodies: characterization and follow-up of a cohort of 127 patients [Abstract]. Arthritis Rheum 1997; 40 (Suppl): S82 23 Swaak AJG, Huysen Y, Smeenk RJT. Antinuclear antibodies in routine analysis: the relevance of putative clinical associations. Ann Rheum Dis 1993; 52: IID-4 24 Fox RI, Robinson CA, Curd JG, Kozin F, Howell FY. Sjogren's syndrome: proposed criteria for classification. Arthritis Rheum 1986; 29: 577-85 Acceptedfor publication 30 October 1998 Ann cu«biochem 1999: 36