This study characterized 52 consecutive patients fulfilling 4 or more of the American Rheumatism Association criteria for systemic lupus erythematosus in order to provide, for the first time, a homogeneous sample for statistical comparison of antinuclear antibody (ANA)-positive and ANA-negative groups.
Ten patients (19%) were seronegative. There was no significant difference in age, disease activity, organ system involvement, erythrocyte sedimentation rate, immune complex levels, or C3 levels. The ANA-negative group showed a higher incidence of involvement for whites and men. Leukopenia, lower levels of antibody to DNA, and higher C4 levels were also characteristic of the ANA-negative group.
The presence or absence of antinuclear antibodies (ANA) is considered by many to be an important diagnostic tool in the differential diagnosis of systemic lupus erythematosus1-3. The American Rheumatism Association (ARA) study which set forth the criteria currently utilized for diagnosis of systemic lupus erythematosus (SLE)4 has been revised to allow for the replacement of the LE cell preparation by antinuclear antibodies as a serologic marker5.
Using these criteria, various frequencies of ANA-negative systemic lupus erythematosus have been observed (Table 1). Although it has been reported that active SLE is invariably ANA-positive6,7, Donadio et al8 observed ANA-positivity in only 22 of 28 patients (78%) with lupus membranous nephropathy. Since a number of these patients had a positive ANA only when undiluted serum was assessed, this maybe an overestimate of positivity because up to 30% of healthy individuals may have positive test results with undiluted serum2.
Urowitz9 emphasized the possible value of dividing SLE into subsets for better determination of prognosis. This has been done on the basis of clustering according to:
Since SLE is considered by many to be an immune complex disorder and since ANA may participate in the formation of such complexes, we sought to determine if ANA-negativity defines a subset of SLE patients differing from those who are ANA positive. We studied only those patients who were clinically diagnosed as having SLE and who met at least 4 of the ARA criteria4. Because of the consecutive (unbiased) nature of patient selection, this is the first study of ANA-negative SLE which is amenable to statistical evaluation.
Patients. The current study evaluated 52 consecutive patients, seen in university-affiliated rheumatology clinics (Memphis, Tennessee), who fulfilled 4 or more ARA criteria for the diagnosis of SLE4. The patients involved were quite representative of the community, and included outpatients and inpatients from a large city hospital, from the largest private hospital in the country, from the faculty private practice plan, and from the largest rheumatology private practice group in the city. One-fourth of patients represented new referrals.
Activity indices. The global index graded disease activity on the basis of severity of symptoms and organ manifestations, according to the technique of Maury et al10. Disease activity was divided into the following categories: none, questionable, mild, moderate, severe, and life-threatening. The clinical activity of SLE was also assessed by the clinical activity index developed by Dillon and Jones11. The index divides into 4 groups the major features influencing morbidity and mortality in SLE. Each group is ranked on a logarithmic scale to prevent inappropriate weighting of factors in cumulative patient scores. For convenience, the logarithm of each patient's cumulative score is used as a final index which ranges from 0 to 4 with increasing activity and severity of SLE.
Renal disease. Renal disease was considered to be present if the serum creatinine was > 1.4 mg/dl in the absence of pre-renal azotemia, if red cell casts were present in the urine, or if urinary protein excretion was > 1.5 gm/24 hours.
Plasma samples. Blood was collected by clean atraumatic venipuncture without a tourniquet. Nine parts of blood were transferred into 1 part of 3.8% (weight/volume) sodium citrate, and plasma was obtained by centrifugation for 25 minutes at 1,600g. Plastic syringes, centrifuge tubes, and storage containers were used. Plasma samples were stored at -70°C until studied.
Serum samples. Serum was obtained from 2.5 ml of blood clotted in a glass tube containing 9 mg Amicar and 50 NIH units of bovine thrombin (topical), and incubated for 2 hours at 37°C until studied.
Antinuclear antibody. Antinuclear antibody was assayed with a fresh frozen mouse kidney substrate12 (Kallestad Laboratories, Chaska, MN) at a dilution of 1:20 and 1:100, using fluorescein-conjugated goat antiserum to human total immunoglobulin which had been titered using the World Health Organization (WHO) primary standard (66/233)13. The IgG, IgM, kappa. and lambda specificities of this conjugate (lot #153NO15) were documented by evaluating reactivity with known IgG and IgM class specific ANAs in chessboard titrations. The endpoint titers were equivalent to monospecific IgG and IgM antibody activity ranges of 30-60 µg/ml as specified by the Centers for Disease Control. Sera from 50 normal blood donors assayed by this technique showed 84% of samples negative at 1:20 and 92% negative at 1:100. Of the ANA-positive blood donor sera, one showed a homogeneous pattern while the remainder were detected as having a speckled pattern. Thus, 1:20 was established as the normal range for this ANA technique.
Rheumatoid factor. Rheumatoid factor was assayed by the latex fixation technique utilizing plasma fraction II (Hyland Laboratories, Costa Mesa, CA) coated to latex12.
Erythrocyte sedimentation rate. The erythrocyte sedimentation rate (ESR) was assessed by the Westergren technique14.
Immune complexes. Immune complexes were measured by methods described in detail by Robinson et al15. In brief, aggregated IgG was prepared by heating and purified by column chromatography. The solid phase C1q binding assay was modified15 from that described by Hay et al16. The results were expressed as microgram equivalents of aggregated human gamma globulin per milliliter of serum. The upper limit of normal for our laboratory is 10 µg/ml (mean ± standard deviation = 5 ± 3, based on measurements of 20 normal subjects). The Raji cell radioimmunoassay was modified15 from that described by Theofilopoulos et al17. The upper limit of normal is 45 µg/ml (mean ± standard deviation = 16.8 ± 12.3 from a group of 20 normal subjects).
DNA binding capacity. Antibodies to DNA were measured using a modified Farr assay18 with125I double-stranded (ds) DNA (New England Nuclear, Boston, MA). Antibodies to single-stranded (ss) DNA were measured using as antigen125I dsDNA heated to 100°C for 30 minutes, and then cooled rapidly on ice. The degree of binding was calculated using the conventional formula: normal, 20%, and 10% for ssDNA and dsDNA, respectively.
Free DNA. Free DNA was detected by counter-immunoelectrophoresis slightly modified from the method of Steinman19. Serum containing antibodies to DNA were applied to wells in agarose in 0.05 M barbital butTer (pH 8.6), together with unknown samples or DNA standards. Samples were subjected to electrophoresis at 40 mamp/slide for 30 minutes. Free DNA in plasma produced a precipitin line midway between the wells. Specificity was tested by digestion with DNase.
Complement. C3c and C4 levels were determined by radial immunodiffusion on M-partigen plates (Calbiochem, La Jolla, CA) according to the technique of Mancini et al20, using the diffusion endpoint. Normal values of C3c are reported as 55-120 mg/dl21-23. Normal values of C4 are reported as 20-50 mg/dl24-26.
C-reactive protein (CRP). CRP was quantitated by radial immunodiffusion27 using goat anti-CRP obtained from Kallestad Laboratories (Chaska, MN). The immunizing antigen had been purified as previously described by James et al28.
Serum protease inhibitors. Alpha1-antitrypsin was assayed in plasma by radial immunodiffusion (Mancini technique) on commercially available plates (Behring Diagnostics, La Jolla, CA) which use endpoint diffusion27.
Fibrin degradation products (FDP). Fibrin degradation products were determined by a commercial latex agglutination test (Thrombo-Wellco test) on thawed serum samples. The sensitivity of the latex reagent was adjusted so that FDP concentrations of 2 mg/dl or greater would give macroscopic agglutination.
Statistical methods. Statistical analysis was performed using the Statistical Analysis Systems program for regression and correlation analysis and the Student's (t-test and Chi-square methods, with Yates' correction when indicated.
Ten of 52 patients (19%) in the present study had an ANA titer less than 1:20 and were considered seronegative (Table 1). Comparison of the incidence of the various ARA criteria descriptives in the current series with those from whom the criteria were developed4 and a subsequent followup study29 revealed similar frequencies with two exceptions.
The average number of Lupus criteria (ignoring ANA as a criterion) fulfilled by the ANA-negative and ANA-positive groups was similar (6.2 versus 6.0).
Lupus criteria are defined by the American Rheumatism Association (ARA).
Alopecia and Raynaud's phenomenon were more frequent in the current study. Comparison of the ANA-negative patients in the current series with those in previous series revealed similar profiles with the exception of hemolytic anemia and alopecia. The differences cannot be evaluated statistically since they were drawn from different populations.
Comparison of ANA-positive and ANA-negative subsets of the current series revealed a statistically significant difference between the groups only for the presence of leukopenia, which was less common (10% versus 36%) in the ANA-negative group (Fisher exact test, 2-tail, P = 0.0321).
The average number of ARA criteria (ignoring ANA as a criterion) fulfilled by the ANA-negative and ANA-positive groups (6.2 versus 6.0) was similar. Evaluation of both groups for differences in age, constitutional signs, differences in system involvement, and differences in disease activity (Tables 2 and 3) revealed no significant differences between the groups:
Seventy percent of the ANA-negative group and 90% of the ANA-positive group were women (P = 0.0424). Twenty percent of the ANA-negative group were black compared with 79% of the ANA-positive group (P = 0.000l).
A comparison of laboratory abnormalities (Table 3) revealed significant differences in antibody titers to both single- and double-stranded DNA and in levels of C4 and alpha1-antitrypsin. A curious observation was that within the ANA-negative group there was a better correlation of ESR with either activity index and of the Raji assay with C4 than in the ANA-positive group (Table 4).
Previous studies of ANA-negative SLE have included populations which did not necessarily meet 4 of the ARA criteria. Arthralgia has often been substituted for arthritis in defining system involvement2, 30-32. Malar rash and photosensitivity are not infrequent complaints in these studies; their relationship to SLE in the absence of other significant symptomatology is unclear.
The present study is the first to compare ANA-negative patients (fulfilling at least 4 criteria) with ANA-positive patients drawn at the same time from the same population in a consecutive (unbiased) sampling of patients clinically diagnosed as having SLE.
Gladman et al33 reported a 10% incidence of seronegativity for ANA among 160 patients with SLE. They observed an increased incidence of skin disease, Raynaud's phenomenon, arthritis, polyserositis, and. positive family history of connective tissue disease in this subgroup. However, only 7 of 16 of their patients met four or more criteria.
Twenty-eight patients with ANA-negative SLE, meeting at least 4 ARA criteria, have been reported with sufficient detail to evaluate1, 30-32, 34-40. The interesting study by Maddison et al2 of 66 patients was not included in this analysis because 42 patients in this group did not fulfill 4 or more criteria. The 24 patients who did fulfill the criteria were not described separately from the total group.
The present study is the first to compare ANA-negative patients (fulfilling at least 4 criteria) with ANA-positive patients drawn at the same time from the same population in a consecutive (unbiased) sampling of patients clinically diagnosed as having SLE. The study by Maddison et al of 66 “ANA-negative” patients drew the ANA-positive patients from a different population from the negative patients, and therefore did not lend itself to statistical evaluation.
Nineteen percent of patients in the present study were ANA-negative. The proportion of patients found to be ANA-negative in other studies41-44 ranges between 2% and 36%. Although there was no significant difference in age between the ANA-negative and ANA-positive groups, the ANA-negative group was predominantly white and the positive group, predominantly black. Whites with SLE more commonly had a negative ANA. While SLE occurred more frequently in women (independent of ANA reactivity), ours is the first series of ANA-negative patients (meeting at least 4 criteria) to contain men.
Our ANA-negative and ANA-positive groups were similar with respect to ARA criteria fulfilled with the notable exception of leukopenia, which was significantly less common in the ANA-negative group. The cause for this finding is not clear at this time. The slightly different frequencies of Raynaud's phenomenon, arthritis, and leukopenia and the moderate difference in frequency of hemolytic anemia probably reflect the rheumatologic population in the current study, in contrast to the predominantly dermatologic population of the previously reported studies.
Our study indicates that ANA reactivity is not a significant indicator of severity since ANA-negativity does not delineate milder disease.
ANA-negativity in SLE has been reported too often to dismiss.
Alopecia and Raynaud's phenomenon were accepted as present if a positive history was obtained. The variation from the aforementioned studies may relate to other investigators' dependence on physical findings or provocation of Raynaud's phenomenon using ice water. Such tests would be more restrictive than those applied in the current study.
Review of our ANA-negative and ANA-positive patients' characteristics (Table 2) showed no significant differences with respect to system involvement, constitutional signs, either activity index (Table 3), or number of criteria fulfilled. Previous suggestions of mild disease in SLE patients who were ANA-negative2, 3 probably relate to the selection process (since most previous series were derived from dermatologic presentations) and especially to the nonrandom selection of ANA-positive SLE patients as comparison groups.
Our study indicates that ANA reactivity is not a significant indicator of severity since ANA-negativity does not delineate milder disease. Donadio's observation that 22% of his patients with membranous lupus nephritis were ANA-negative further substantiates this8.
There are several possible explanations for ANA-negativity in patients with SLE. The choice of 1:20 as the cutoff dilution for ANA-reactivity is standard2 and was reestablished with normal blood donors. The assay becomes quite nonspecific below 1:20. While faulty technical performance of the assay could produce false negative ANA results, ANA-negativity in SLE has been reported too often to dismiss. In the present study, this possibility was ruled out by reevaluating, under controlled conditions, the ANA originally determined by the hospital laboratory. Recently a primary standard for ANA reactivity has been prepared by the WHO; this standard was used to titer the conjugate for the substrate used in the studies presented here.
A proportion of patients with SLE have ANA which include some antibodies of the IgE class that may not be identified by the currently employed polyspecific antisera.
Investigations by Fessel31, Provost30, and Maddison2 have demonstrated no evidence of a prozone phenomenon. While liver or kidney from rats or mice is the standard substrate for assessing ANA reactivity, other systems have been used3, 5, 45. Specificity is not well established for these other substrates. Gladman and colleagues33 suggested substrate insensitivity since 31% of their group of ANA-negative patients had anti-dsDNA. In the present study, 1 of 10 ANA-negative patients with SLE had antibodies to dsDNA.
Use of monospecific antisera (e.g., antihuman IgG) could result in false negatives since 4% of ANA is IgM46. In this and other reports of ANA negativity, a polyspecific antihuman gamma globulin, which would identify non-IgG antibodies, was used. A proportion of patients with SLE have ANA which include some antibodies of the IgE class47 that may not be identified by the currently employed polyspecific antisera.
Blomjous and Feltkamp-Vroom48 have suggested that in some circumstances ANA may bind to circulating antigen, forming immune complexes, thus rendering the patient transiently ANA-negative. If ANA-bound complexes were the cause of ANA-negativity, it might be anticipated that circulating free DNA might be found in ANA-negative patients but not in ANA-positive patients. Our series demonstrates the opposite (Table 3), although the numbers are too small to achieve statistical significance.
It is possible that all free DNA could be bound in our ANA-negative patients or that small fragments of DNA, not measurable by our assay, could be present. ANA has been reported less frequently and at lower titers in patients with complement deficiencies36, 39, 49. A detailed assessment of deficiencies of complement was not possible in the current series.
Absence of ANA from patient serum may be related to absorption of all ANA to specific patient tissues31. Koffler et al50 and Appel et at51 eluted antibody to native DNA from the glomeruli of patients who had no demonstrable ANA in their serum. Tan and Vaughan45 noted that antibody to single-stranded DNA failed to stain interphase nuclei, such as those of standard liver and kidney. This was probably because such substrates lack exposed sites of single-stranded determinants2,52, which are primarily purine and pyrimidine bases of denatured DNA53.
The present study compares, for the first time, ANA-negative and ANA-positive patients drawn from the same population.
Analysis of features of the ANA-positive and ANA-negative groups failed to demonstrate a significant difference with respect to severity of disease.
The presence of certain anticytoplasmic antibodies (Ro and La)54,55 has been reported to correlate negatively with the presence of certain antinuclear antibodies (e.g., No and Sm)56-58. The relationship, however, is probably indirect since antibodies to Ro or La are found in 30% of patients with SLE31. Reichlin and Mattioli58 observed anti-Ro antibodies in 3 patients who were ANA-negative. Provost30 has also reported anticytoplasmic antibodies in some ANA-negative patients.
The last possibility, of course, is that the ANA-negative patients do not actually have SLE. A clinico-pathologic conference reported in the American Journal of Medicine59 is relevant to this point. A 51-year-old woman had fever, rash, arthritis, pericarditis, pleural effusion,. proteinuria, hematuria, and an associated disorder, hypothyroidism. While the patient's symptoms superficially simulated those of SLE (and actually met 4 criteria), pathologic evaluation revealed granulomatous vasculitis.
The present study compares, for the first time, ANA-negative and ANA-positive patients drawn from the same population. Analysis of features of the ANA-positive and ANA-negative groups failed to demonstrate a significant difference with respect to severity of disease.
Supported in part by BGSDRR/NIH Grant, R.R. 05366.
We are grateful to Peter Panayiotou and Debora Bork for skilled technical assistance and to Melody Davis for help in preparing the manuscript.
|Lee (43) *||110||11|
|Gladman (33) †||153||10|
|Present study †||52||19|
*Rat liver substrate
†Mouse kidney substrate
*Facial erythema, discoid lesions, alopecia, photosensitivity, or mucosal ulcers
†Reynaud's phenomenon, tissue infarction, or vasculitis
(n = 10)
(n = 42)
|Average age||40 ± 4||39 ± 2|
|% male †||30||10|
|% black ‡||20||79|
|Global index||2.5 ± 0.3||2.7 ± 0.2|
|Clinical activity index||1.91 ± 0.13||1.92 ± 0.11|
|ESR (mm/hr)||66 ± 14||60 ± 9|
|C3c (mg/dl)||86 ± 7||63 ± 5|
|C4 (mg/dl) §||31 ± 6||15 ± 2|
(% binding) ¶
|4.8 ± 2.6||22.4 ± 4.1|
(% binding) #
|0.8 ± 0.8||11.3 ± 4.0|
|Raji cell assay for IC
(µg AHGG eqv/ml)
|27 ± 15||49 ± 8|
|C1q assay for IC
(µg AHGG eqv/ml)
|40 ± 12||24 ± 4|
|28 ± 8||18 ± 4|
|301 ± 30||236 ± 11|
|Free DNA (titer)||0 ± 0||2 ± 1|
*ESR = erythrocyte sedimentation rate
ssDNA = single-stranded DNA
dsDNA = double-stranded DNA
IC = immune complex
AHGG = aggregated human gamma globulin
†Student's t-test, P = 0.0424
‡Student's t-test, P = 0.0001
§Student's t-test, P = 0.0074
¶Student's t-test, P = 0.0007
#Student's t-test, P = 0.0042
^Student's t-test, P = 0.0152
|ANA −||ANA +|
*ESR = erythrocyte sedimentation rate
NS = not significant
Copyright © 2008-2018 Georgena Sil. All Rights Reserved.