In NZB/W F1 mice, the administration of TNF-α reduces the severity of the lupus-like illness [6
]. This observation has not been repeated in other lupus-prone mouse strains and the effect of TNF-α in NZB/W F1 mice depends on the dose and the age of the mice [7
Rizgar Mageed (University College London, UK) described a series of experiments designed to clarify the effect of TNF-α in the NZB/W F1 strain. Immunization of young NZB/W F1 mice with phosphatidylcholine/ovalbumin conjugate leads to the production of anti-double-stranded (ds)DNA antibodies. This effect is reduced by administration of recombinant TNF-α and enhanced by anti-TNFα. Histological examination of lymphoid tissues of these mice showed that TNF-α reduces the size of T cell areas, whereas anti-TNF-α promotes T cell function but disrupts B cell migration.
This work led to the hypothesis that different results would be obtained in neonatal mice. It was postulated that, in these mice, anti-TNF-α would enhance T cell function in such a way as to promote tolerance and reduce autoimmunity. However, the results of the experiment did not support the hypothesis. Neonatal mice treated with anti-TNF-α developed increased numbers of T cells, more anti-dsDNA and antinucleosome antibodies, and increased proteinuria in comparison with mice treated with a control antibody.
The concept that TNF-α protects against the development of SLE was also supported by studies conducted in TNF-α-deficient mice, described by Rachel Ettinger (National Institutes of Health, Bethesda, MA, USA). These mice develop antinuclear and anti-DNA antibodies after 15 weeks of age, but do not develop lupus-like illness. However, the actual mechanism of this effect is uncertain because mice that lack both the TNF-55 and TNF-75 receptors do not develop these autoantibodies. Moreover, the effect is highly dependent on genetic background. TNF-/- mice on a B6 background do not develop autoantibodies, whereas those on a mixed B6 × B129 background do. Therefore, it appears that a gene on the 129 background is required to predispose the TNF-deficient mice to autoimmunity. The development of autoantibodies is dependent on T cells and on IL-6, because autoantibodies do not develop in TNF-α-/- mice that lack either T cells or IL-6.
These experiments in murine models suggest that administration of anti-TNF-α might predispose to the development of SLE in humans. Because anti-TNF-α drugs are now in widespread use in the treatment of rheumatoid arthritis and Crohn's disease, there are clinical data relating to this issue. These data were reviewed by Sir Ravinder Maini.
Anti-dsDNA antibodies occur very rarely in patients with rheumatoid arthritis who have not received anti-TNF-α therapy but were reported in 7% (11/156) of patients who had received such treatment [8
]. After a single infusion of infliximab, anti-dsDNA antibodies first develop after a mean of 6.3 weeks and disappear 4–6 weeks later. When repeated infusions are given, the anti-dsDNA antibodies may not disappear until after the last infusion. Anti-dsDNA antibodies have been reported after treatment with either infliximab or etanercept, and the dose of anti-TNF-α given does not affect the likelihood of an anti-DNA response.
The majority of these patients develop IgM but not IgG anti-dsDNA antibodies and do not develop clinical features of SLE. However, clinical SLE can occur following anti-TNF-α treatment, and there are a number of well documented cases [9
]. The disease is mild, remits when the drug is stopped, and neither cerebral nor renal involvement has been reported.
Why is the prevalence of clinical SLE after anti-TNF-α treatment so low (0.04–0.2%) when the prevalence of anti-dsDNA antibody production following such treatment is much higher (16%)? Similarly, why do TNF-α knockout mice develop autoantibodies but not a lupus-like illness? One possibility is that TNF-α exerts two opposing effects. The first effect operates at the level of T lymphocytes to suppress autoantibody formation. The second effect operates at the level of the target tissues to promote inflammation. For example, TNF-α is known to activate endothelium, which could lead to transmigration of leucocytes into the tissues.
The concept that TNF-α could have two opposing effects in SLE was aptly summarized by Josef Smolen (University of Vienna, Austria) who dubbed it the Janus cytokine in honour of Janus, the double-faced god of Roman mythology. He pointed out that levels of TNF-α are raised in the serum of patients with SLE [10
] (although levels of the soluble inhibitor TNF receptor are also raised) and that it has been detected in renal biopsies of patients with lupus nephritis. These findings suggest that TNF-α blockade might be useful as a treatment for SLE.
Smolen reported his experience with four patients with SLE who had been treated with 5 mg/kg infliximab and concomitant azathioprine. In this open trial, all four patients showed signs of clinical improvement, even though levels of anti-dsDNA rose in two cases.
At this point, therefore, the place of TNF-α blockade in the treatment of SLE is unclear. Although there is a large body of evidence pointing to protective effects of TNF-α against the development of autoimmunity in both humans and mice, there is also evidence that anti-TNF-α could be used as an agent to reduce tissue damage in patients with SLE.