Introduction
Antinuclear antibodies (ANAs) are measurable in approximately 25% of the population, and the prevalence of significantly elevated levels may be 2.5% [
1]. Findings from numerous studies show remarkable consistency across ethnically and racially diverse study populations despite the use of many different methods for ANA measurement. The persistence of this type of autoreactivity in the human population suggests that antinuclear antibodies may be an important component of the normal immune response. Most individuals with a positive ANA do not have an autoimmune disease and most also are unlikely to develop one. This is consistent with the fact that the prevalence of all autoimmune disorders is 5 to 7% [
2]. Furthermore, the disease that is most closely linked to ANA positivity, systemic lupus erythematosus (SLE), is relatively rare, affecting no more than 1 to 1.5 per 1,000 persons (0.1 to 0.15%) in the United States [
3]. Nevertheless, since ANA positivity is for all practical purposes a requirement for SLE diagnosis, it must also be assumed that individuals who are in preclinical disease stages are represented in the ANA positive healthy population. Although many consultations for ANA positivity seen in rheumatology practice are not associated with any identifiable pathology, it is also true that if early detection of SLE is to become feasible, focus on the ANA positive population will be necessary.
We have considered the possibility that other blood markers could be used to differentiate benign ANA positivity from that which carries a high risk of autoimmune disease. These markers may include other autoantibodies, since it is well-known that autoantibody positivity increases in quantity and complexity in years preceding a diagnosis of SLE [
4]. Gene dysregulation in peripheral blood cells has been closely associated with SLE diagnosis and disease status, so changes in gene expression may also signal a condition with enhanced risk.
To address these questions, we studied healthy individuals and patients with autoimmune diseases who had been enrolled in the Dallas Regional Autoimmune Disease Registry (DRADR). A subgroup of healthy controls that were found to have high ANA levels was examined in greater detail using autoantigen and gene expression arrays. The findings suggest the feasibility of identifying risk markers for development of SLE in the setting of ANA positivity, using both demographic features and profiling of autoantibodies and expressed genes in peripheral blood.
Discussion
Detecting autoimmune disease in early or preclinical stages is clinically important because the institution of treatment prior to the onset of organ damage has a greater chance to ameliorate or even cure the disease [
12]. However, early and reliable diagnosis of lupus is a challenge, in large part due to the performance profiles of available diagnostic tests. The optimal test would be sensitive enough to detect all individuals who have a disease while at the same time delivering sufficient specificity to have reasonable predictive probability that the disease is likely. The classic screening test for SLE is the presence in serum of antinuclear antibodies (ANAs) measured by immunofluorescence staining of a cellular substrate [
13]. ANA positivity is for all practical purposes required to make a diagnosis of lupus since more than 99% of patients with SLE have significant levels of this autoantibody detected at some time during the course of disease. However, since the prevalence of SLE is low, most individuals presenting to a physician with ANA positivity do not in fact have lupus and are not at high risk for developing this disease. But there are few available quantitative and objective measures to establish prognosis for an individual with a positive ANA. This contrasts to the tools that are available for determining the risk of cardiovascular disease, where multifaceted profiles including elements of personal and family history, social habits, body measurements and lab tests can generate a reasonable and very personalized risk prediction for an individual patient [
14].
In practice, physicians actually do employ some profiling to assess the risk associated with ANA positivity. Young women are more likely to develop lupus than old men, for example, so ANA positivity in the former is of greater concern. The present study confirms our previous observation that female gender is a risk factor for significant ANA positivity [
1]. This result is also consistent with other findings in healthy control populations including a study of 500 normal individuals in Brazil showing that ANA positivity was almost twice as prevalent in females as in males [
15]. Similar findings were reported in a rural Canadian population, with the gender difference being greatest at higher ANA levels, as was also noted in the present study [
16]. The enhanced female risk profile does not appear to extend to the anti-CCP antibody that is associated with RA, another female-predominant autoimmune disorder. Reasons for the association of female gender with strong ANA positivity remain obscure.
Age was not correlated with ANA levels in HCs, which seems to contradict the generally-accepted hypothesis that immunosenescence is associated with increased autoantibody production due to decreased self-regulatory mechanisms. The present findings are consistent, however, with other reports [
15] and suggest that in a cross-sectional analysis such as this there are many reasons for ANA positivity. The younger individuals may in fact have abnormal immune regulation that predisposes to SLE-like disease while older persons may develop autoreactivity as part of aging immune responses that do not lead to development of pathology.
Other clues are available to suggest approaches to stratifying risk in the ANA-positive population. One is the well-recognized presence of other autoantibodies that accumulate prior to SLE diagnosis [
4]. The prevalence of any one of these is low, however. For example, ribosomal P autoantibodies are highly specific for SLE but are present in less than one-third of patients [
17]. This finding predicts that if autoantibodies are included in risk profiling, multiplexed assays will be required. One such approach that is available clinically, the ENA panel, was not useful in the high ANA HC individuals in the present study because very few positives were found. This result is consistent with our previous experience indicating that only the high ANA positives are likely to have ENA specificities [
1] and even then, the prevalence of other autoantibodies in HCs is very low. The alternative approach which has been applied in the present study is to greatly expand the repertoire of autoantibodies that can be probed by use of an autoantigen array, which in other studies has been shown to provide novel insights into expressed autoantibody repertoires [
18]. The array data in the present study revealed increased autoreactivity in a group of high ANA HCs. One of the elevated autoantibodies was thyroglobulin, and since autoimmune thyroid disease is probably about 10-fold more common than SLE [
19], this result suggests that a significant proportion of the ANA positivity seen by rheumatologists is related to thyroid autoimmunity. Longitudinal studies suggest that thyroid autoreactivity, especially in women, may be predictive of thyroid dysfunction [
20]. Autoantibodies to cartilage proteoglycan can be measured in several systemic and joint-specific rheumatic diseases including Sjogren's Syndrome, rheumatoid arthritis, lupus and ankylosing spondylitis [
21], suggesting that undetected or preclinical joint inflammation may contribute to ANA positivity. The relative increase in skin autoreactivity in the high ANA HC group might be related to a relative enrichment for skin antigens on the arrays, so this finding should be interpreted with caution. However, it does raise the interesting question of whether some early autoimmune events might take place in cutaneous structures. Transient ANA positivity has been observed in patients with polymorphous light eruption [
22] and exposure to sun in susceptible individuals can trigger major organ-damaging lupus [
23]. The interface dermatitis that characterizes the pathology in SLE as well as in other skin disorders may precede a diagnosis of lupus [
24] and is associated with activation of the Type I interferon gene signature [
25]. The triggering of this set of genes in skin, however, is not limited to inflammatory disease but can also occur as part of the immune response to viruses, raising the question of whether cutaneous reactions to environmental agents in susceptible hosts might generalize to a systemic response.
The autoantibody arrays show that although the autoreactivity spectrum in SLE is broad, not all specificities are elevated. Upregulated autoantibodies to gliadin, and to T lymphocytes and neuroblastoma cells were present only in the ANA high HC group, and not in the SLE patients. This result confirms in part our previous report that gliadin autoreactivity is associated with incomplete forms of lupus that are associated with myopathies [
26]. Whether these autoantibodies are actively protective and lower the risk of lupus or alternatively are predictive of other autoimmune diseases developing in these individuals will require further longitudinal investigation.
Upregulated genes observed in high ANA HC individuals include some in the Type I IFN signature that are associated with SLE [
27,
28]. While some of these genes, notably
IFI27, were only elevated in SLE, others such as
MX-1 showed an intermediate level in the High ANA HC group. In addition to being the most highly-upregulated of the IFN genes in our sample, other data suggest that
IFI27 is relatively more specific for lupus than at least some of the other IFN-inducible genes. A recent study demonstrated that
IFI27 is more likely to be upregulated in lupus than in another autoimmune condition, idiopathic thrombocytopenic purpura [
29]. The specific functions of many of the proteins associated with IFN-related genes are obscure, but
MX-1 is closely associated with the response to the influenza virus, so upregulation of this gene in normal individuals might reflect the ubiquitous exposure to this pathogen.
The present data suggest possible components of a lupus risk profile. As in the cardiovascular risk profiling equations, gender will be a factor but in lupus the risk will be associated with females rather than males. The age category will be inverted from that of cardiovascular disease, with greater weight given to younger ages. The lupus risk is correlated with ANA levels, not just positivity, with values in the upper quartile having a two- to three-fold elevation of risk [
1], so this will be an important component. Other autoantibodies that may add to risk include those that are clinically well-known like anti-dsDNA and anti-Sm, as well as other novel specificities including the skin determinants identified in the present study. On the other hand, antineuronal, anti-thyroid or gliadin autoantibodies might steer attention away from SLE towards other autoimmune disorders. Elevated expression of genes related to the Type I IFN signature is likely to add points to the risk equation.
This study has several limitations. One is the lack of information regarding use of medications, especially hormones, by the HC. Whether administration of estrogen in the form of oral contraceptives or postmenopausal replacement therapy might induce high ANA levels in a healthy individual cannot be ascertained from our data. Another limitation is the cross-sectional design which does not permit insights into changes that evolve over time. And it would be of interest to determine reactivity to foreign antigens such as infectious agents to further interpret the significance of the autoreactive responses. Ultimately, validation of risk profiles will require longitudinal studies.
Finally, since SLE is a relatively rare disorder, the probability of finding a new onset patient is low even after ANA positivity has been identified. One approach to increasing the likelihood of useful results would be to follow individuals who already have some of the identified risk profile components. For example, studies could be carried out in individuals who have been sent for ANA testing for any reason. The very fact that the individual sought medical attention and that an ANA was ordered is likely to increase the pre-test probability of disease. Such an approach has shown that in the population of individuals sent for rheumatoid factor testing, the pre-test probability for RA is 17% [
30].
Competing interests
NJO has equity interest in ArthroChip LLC and has received research grants from Medimmune, Novo Nordisk, Genentech/Roche and Human Genome Sciences. DRK has received a research grant from Human Genome Sciences. All other authors declare that they have no competing interests.
Authors' contributions
QZL supervised the protein and gene expression array assays and analyses. DRK organized the patient registry. JQ organized and performed laboratory assays. JZ performed the protein array assays, while YL performed data analyses for protein and gene expression array assays. VKB coordinated patient recruitment for the registry. BFC contributed to design and validation of the autoantibody array. EKW conceived of the study design and NJO assembled the patient cohorts and supervised clinical phenotyping. QZL, DK, EKW and NJO wrote the manuscript with input and consensus from all authors. All authors read and approved the final manuscript.