An initial literature search was conducted and 168 articles were retrieved. These articles were graded according to the criteria reported in the Introduction article for this series (1). From this group of 168 articles, 43 that assessed the prevalence of anti-DNA antibodies in patients with SLE, patients with various other diseases, and healthy controls were considered for further review. Of these 43 studies, 11 were graded “A” (Table 1), and they form the basis of the recommendations (13 studies were graded as “B” and the remainder were graded “C” or “D”).
Table 1. Population statistics for anti-DNA antibodies in diagnosis: Systemic lupus erythematosus vs. healthy controls and other diseases*
| || ||Farr||0.4||0.96||10.5||0.63|| |
| || ||Crithidia||0.13||1.0||>18.5||0.87|| |
| || ||Farr||0.57||0.9||5.7||0.64|| |
| || ||Farr||0.6||0.98||26.0||0.41|| |
| || ||Farr||0.68||0.85||4.5||0.38|| |
In almost all studies, the prevalence of elevated levels of anti-DNA antibodies in healthy controls was zero or very low. Of note, in some studies the threshold for a positive test was 2 standard deviations above the mean of the controls; therefore 2.5% of the controls would have a positive test result by definition. It is not surprising that healthy persons may occasionally be found to have detectable anti-DNA antibodies in their serum, because normal B cells have been shown to be capable of producing such antibodies upon stimulation (B cells from SLE patients tend to produce such antibodies spontaneously) (40).
Anti-DNA antibodies have been reported in patients with a variety of rheumatologic diseases and other conditions including: rheumatoid arthritis, Sjögren's syndrome, scleroderma, drug induced lupus, Raynaud's phenomenon, mixed connective tissue disease, discoid lupus, myositis, chronic active hepatitis, other liver diseases, uveitis, relatives of SLE patients, patients hospitalized for nonrheumatologic diseases, Graves' disease, Alzheimer's disease, juvenile rheumatoid arthritis, certain laboratory workers, anticardiolipin antibody syndrome, and persons with silicone breast implants (41–60). The frequency of elevated levels of anti-DNA antibodies in conditions other than SLE is uniformly low (≤5% of patients), and when present, they are often present in low titer. Therefore, outside of a research setting, ordering tests for anti-DNA antibodies is not useful for the diagnosis of any condition other than SLE. However, in a patient without SLE, a positive test result for anti-DNA, particularly at low levels, may be explained by the presence of one of these conditions.
Although anti-DNA antibodies have been reported in a number of other conditions, with rare exception their prevalence was also very low. Therefore in this analysis, healthy subjects and patients with other diseases have been combined, and will be contrasted with SLE patients.
Anti-DNA antibody testing is very useful for the diagnosis of SLE. As can be seen in Table 1 (15, 16, 18, 38, 44–46, 61–64), weighted means (weighted according to numbers of patients assessed) for the use of anti-DNA in diagnosing SLE were 57.3% for sensitivity and 97.4% for specificity. The positive likelihood ratios (LR) for anti-DNA antibodies in the diagnosis of SLE are very high. Most LR were >10, and the weighted mean positive LR was 16.3 for the grade A articles. This indicates that a positive test result will likely have a large impact on the pretest probability; consequently a positive test result will substantially increase the posttest probability of the diagnosis being SLE. Thus, in the setting of some clinical suspicion of SLE, a positive anti-DNA strongly supports the diagnosis.
The sensitivity of the anti-DNA varies substantially among the studies, with a mean sensitivity of 57.3%. This probably depends on a number of factors, particularly the specific population assessed. The negative LR also vary, but most cluster about 0.5 or higher. The weighted mean negative LR was 0.49. Given this small negative likelihood ratio, a negative test for anti-DNA antibodies does not offer strong support to exclude the diagnosis of SLE.
Interestingly, the development of anti-DNA antibodies has been shown to antedate the clinical diagnosis of SLE in some cases (65).
In studies reporting titers or units, the specificity increases with higher concentrations of anti-DNA antibodies. Although uncommonly found in other conditions, clinicians should be aware that a positive anti-DNA may be rarely seen in patients with other conditions, particularly when present in low titer. This is reflected in the range of specificities in Table 1, which overall are close to, but not equal to 100%. Therefore, a positive anti-DNA is not diagnostic of SLE, and it must be interpreted in the context of the clinical presentation.
Several studies have assessed the prevalence of anti-DNA antibodies in racially distinct populations of SLE patients (66–70). Although the reported prevalence of anti-DNA antibodies has varied among the populations, those studies that actually assessed different races in a single study have found comparable prevalences of anti-DNA antibodies among SLE patients (67, 69, 70).
Several studies have addressed the potential utility of testing for anti-DNA in patients without a positive antinuclear antibody (ANA) test (41, 64, 71–73). Using the Hep-2 substrate for ANA, the prevalence of patients with a positive anti-DNA assay despite a negative ANA test has been reported to be 0–0.8% (41, 71). In older studies that used less sensitive rodent substrates, frequencies of anti-DNA among ANA negative patients has been reported as 3–8% (64, 72, 73). Therefore, unless there is reasonable suspicion that the ANA may be falsely negative, anti-DNA antibody testing is not generally indicated in ANA-negative patients.
Anti-DNA antibodies are very useful for the diagnosis of SLE. They are particularly useful to confirm the diagnosis for a patient whose clinical presentation already suggests a reasonable pretest likelihood of the diagnosis of SLE being present (e.g., 5% or more). While offering very strong support in the correct clinical setting, anti-DNA antibodies have been rarely described in a variety of other conditions; therefore a positive anti-DNA is not diagnostic of SLE by itself. Not all patients with SLE have positive anti-DNA antibodies; therefore a negative anti-DNA does not exclude the diagnosis of SLE. Anti-DNA antibodies are not useful for the diagnosis of other conditions. In general, anti-DNA antibody testing should be reserved for patients with a positive ANA.
From the literature search, 31 studies assessing the correlation between anti-DNA antibodies and some aspect of prognosis in patients with SLE met the criteria for further review. Of these 31 studies, 8 were graded “A” and 11 were graded “B” (Table 2) (15, 16, 18, 37, 38, 44, 50, 61, 74–83), and form the basis of the recommendations.
Table 2. Use of anti-DNA antibodies for prognosis among systemic lupus erythematosus patients
| ||Farr||0.98||0.97||25.2||0.02||NA||NA||NA||NA||NA||NA||NA||NA|| |
| ||Crithidia||0.56||0.97||24.1||0.45||NA||NA||NA||NA||NA||NA||NA||NA|| |
| ||Crithidia||0.14||0.91||1.55||0.94||NA||NA||NA||NA||NA||NA||NA||NA|| |
| ||Farr||0.41||0.73||1.5||0.81||NA||NA||NA||NA||NA||NA||NA||NA|| |
| ||Farr||0.73||0.72||2.6||0.38||NA||NA||NA||NA||0.76||0.72||2.7||0.33|| |
| ||Crithidia||0.43||0.6||1.07||0.95||0.46||0.6||1.15||0.9||NA||NA||NA||NA|| |
| ||Farr||0.89||0.4||1.48||0.28||NA||NA||NA||NA||NA||NA||NA||NA|| |
|Weighted means|| ||0.66||0.66||4.14||0.51||0.65||0.41||1.7||0.76||0.86||0.45||1.7||0.3|| |
After review and grading of the literature, it was considered that there were sufficient studies to assess the utility of anti-DNA antibodies in the measurement of several aspects of prognosis in patients with SLE. Measures of prognosis that were considered include: active versus inactive overall disease, presence of renal involvement, and active versus inactive renal disease.
There are several important caveats relevant to the interpretation of these studies. The presence of anti-DNA antibodies was often included as part of the criteria by which active disease was defined. This tautology makes it harder to separately assess the correlation of anti-DNA antibodies with disease activity. Also, there has been no single, universally accepted definition of active disease for SLE patients. Because of the substantial variability on definitions of activity, the authors' definitions of activity were accepted, provided they were explicitly defined. In many studies, any potential effect of immunomodulatory therapy or other treatment on the disease activity and/or laboratory test results is impossible to extract, as such information is often not specifically provided. Therefore, this has not been specifically addressed. Many studies provide limited information on patient selection criteria for the SLE patients reported, thus raising the potential for bias in the results.
Considering overall activity of SLE, as variously defined among the different studies, the data from these studies indicate that the presence of anti-DNA antibodies is associated with increased disease activity. However, the sensitivity and specificity vary among studies (weighted mean sensitivity and specificity were both 0.66). Positive LRs varied from approximately 0.88 to more than 10. This indicates that in some studies the presence of anti-DNA antibodies was strongly predictive of active disease while in others it had no prognostic significance whatsoever. This may result from heterogeneity in the populations of SLE patients assessed in the various studies. Thus, although anti-DNA antibodies can be associated with disease activity, there are clearly populations of SLE patients who have persistently elevated anti-DNA antibodies but do not have active disease (84, 85).
The weighted mean positive LR was 4.14. This implies that the results of the test may alter the pretest likelihood of the determination of disease activity, but that the effect will be relatively small. A positive anti-DNA would be anticipated to be useful in assessing prognosis only in the correct clinical setting (i.e., if there are indications of disease activity from other clinical assessments). Of note, with one exception (75), higher titers of anti-DNA antibodies were more strongly associated with active SLE in the studies reporting titers (15, 41, 44, 50, 76, 86). Therefore, a higher threshold for a positive anti-DNA would be expected to increase its specificity, and perhaps be more predictive of SLE disease activity.
Also, in the 2 grade “A” studies that directly compared all 3 methods for anti-DNA determination (19, 38), the performance characteristics for the Farr and Crithidia assays were superior to that of the ELISA (Table 2). The weighted mean negative LR for the association between anti-DNA and overall SLE disease activity was 0.51. An interpretation of this would be that although a negative test result does not exclude disease activity, it has a small but potentially relevant association with the lack of disease activity. Therefore, a negative test for anti-DNA would be most helpful in the setting of a low pretest probability of disease activity.
Studies analyzing the correlation of anti-DNA antibodies with the presence of renal disease had weighted mean sensitivity of 0.65, specificity of 0.41, positive LR of 1.7, and a negative LR of 0.76. Therefore, although there was some variability among the studies, the presence of a positive test for anti-DNA antibodies only slightly increases the likelihood that a SLE patient has renal disease; a negative anti-DNA test, by itself, does little to exclude the presence of renal disease. By extension, testing for anti-DNA in order to assess whether renal involvement was present in an SLE patients would be of most value (and should be limited to) patients with a reasonable pre-existing suspicion of renal involvement.
Several studies have assessed the association of anti-DNA antibodies with specific histopathologic measures of renal disease in SLE (some of these studies report only correlation coefficients or other summary statistics, but not primary data). Anti-DNA antibodies have been reported to correlate with renal activity (but not chronicity) index (83, 87), and to have some correlation with WHO type IV (diffuse proliferative) glomerulonephritis (87, 90). However, although correlations may have been statistically significant, anti-DNA antibody positive patients were noted among all histopathologic types.
In the one study (87) from which data could be clearly extracted according to WHO pathologic changes, the statistics for the association of anti-DNA antibodies with type IV glomerulonephritis (GN) (as compared to all other types) were: by Crithidia sensitivity 0.93, specificity 0.6, positive LR 2.3; and by ELISA sensitivity 1.0, specificity 0.04, positive LR 1.04. Higher titers of anti-DNA were more clearly associated with type IV GN. While determinations of serum complement protein concentrations were not specifically considered in this analysis, it has been suggested that the combination of increased titers of anti-DNA in conjunction with depressed levels of complement proteins may be more predictive of active lupus nephritis (91, 92). This issue was not specifically addressed in this review.
Considering the activity of known renal disease in SLE patients, weighted means for anti-DNA were 0.86 sensitivity, 0.45 specificity, 1.7 positive LR, and 0.3 negative LR. Thus, as for overall disease activity, the presence of anti-DNA antibodies may increase the pretest-to-posttest likelihood of active renal disease in a patient with SLE, but the effect will be small. Although the number of studies was small, the negative LR was more relevant for determining activity of known renal disease as opposed to determining whether renal involvement was present at all. In the studies reporting titers (15, 44, 76, 86), higher titers were often seen in patients with more active disease.
Several studies assessed the impact of anti-DNA antibodies on overall outcome or survival of SLE patients (75, 79, 93–96). Although the numbers of patients in each group in some of the studies were small, there was no significant effect of anti-DNA antibody on outcome or survival.
Anti-DNA antibodies correlate with overall activity of disease in patients with SLE, and they are useful for this purpose. However, the LRs for this correlation are relatively small. This implies that the finding of anti-DNA will have a limited impact on the pretest likelihood of active disease for a given patient with SLE. The anti-DNA should optimally be used for patients with some pretest likelihood of active SLE (i.e., those patients for whom active SLE is suspected on clinical grounds otherwise, placing the patient at a pretest likelihood of disease activity of 10% or greater). In SLE patients with no other evidence of disease activity, a positive anti-DNA is unlikely to make important changes on the overall impression of disease activity. In patients with a positive anti-DNA, higher titers were more closely associated with disease activity. As a guide to overall disease activity, anti-DNA should be interpreted in SLE patients only in conjunction with suspicion of disease activity based on history, physical examination, or the results of other laboratory tests.
Anti-DNA antibodies correlate with the presence and activity of renal disease in patients with SLE, and they are useful for this purpose. However, the LRs for this association are very small, implying that the finding of a positive anti-DNA may add little to the overall impression of renal disease or its activity. Although anti-DNA antibodies have also been associated with histopathologic type of lupus nephritis as well as a pathologic activity index, the literature in this area is not extensive, and precludes analysis that would be of the greatest assistance to the practitioner. In patients with a positive anti-DNA, higher titers were more closely associated with the presence and activity of renal disease. As a guide to renal disease and its activity in SLE patients, anti-DNA may be useful; however, it must be interpreted in conjunction with other measures of renal disease.
Anti-DNA antibodies do not correlate with overall survival or outcome in SLE, and are not useful for this purpose.