In this study, profoundly depressed NK cell activity was observed in a large subgroup of patients with sJRA and in only 1 of 20 patients with the polyarticular form of the disease. The extent of NK dysfunction in this group of patients was similar to that seen in patients with MAS [17
] or HLH [12
]. The two study groups (sJRA versus other JRA subtypes) were well matched in terms of age, duration of the disease, and treatment regimens with the exception of a slightly higher proportion of patients with sJRA receiving steroids. Steroids have been reported to suppress the cytolytic activity of NK cells [22
], and this might potentially have contributed to the observed differences in NK function. However, the logistic regression analysis did not show significant differences between groups defined on the basis of treatment regimens. In addition, several patients with sJRA who demonstrated profoundly depressed NK cell cytolytic activity were receiving only non-steroidal anti-inflammatory drugs. Owing to the limitations of the statistical power with the numbers of study subjects enrolled, it is possible that some effects of the immunosuppressive medications might have been underestimated. Nevertheless, NK dysfunction seems to be a distinguishing feature of sJRA that is intrinsic to the disease itself.
Further analysis of the flow cytometry data revealed that some of the patients with sJRA had a rather selective disappearance of the circulating immunoregulatory CD56bright
subset of NK cells, a pattern previously seen in patients with MAS or HLH [25
]. These cells express low levels of perforin and are, in general, poorly cytotoxic [24
]. The disappearance of CD56bright
NK cells from peripheral circulation is therefore unlikely to account for the defects in cytolytic activity of NK cells. In contrast, CD56bright
NK cells might have a function in regulating the CD56dim
cells, and in this case their disappearance might have an effect on cytolytic activity in some sJRA patients. Alternatively, the apparent absence of immunoregulatory NK cells in peripheral circulation might reflect their active recruitment to sites of inflammation.
Although the observed NK dysfunction in a subgroup of patients with sJRA might not be of primary etiological significance for JRA itself, the similarities to the immunologic abnormalities seen in MAS and HLH suggest that depressed NK cell activity is likely to be relevant to the pathogenesis of MAS in sJRA. It is important to mention that low NK cell activity has been noted in many rheumatic diseases [28
], most notably in systemic lupus erythematosus [29
]. In our study, however, in a subgroup of patients with sJRA, the extent of NK dysfunction was profound, with an almost complete absence of cytolytic activity. This parallels the fact that, although MAS has been described in association with almost any rheumatic disease and it is not uncommon in systemic lupus erythematosus [30
], it is by far most common in sJRA [4
Other groups have noted low levels of perforin expression in cytotoxic cells from patients with sJRA in comparison with other clinical forms of the disease, suggesting that this feature might be responsible for the increased incidence of MAS [26
]. In our study, when patients with sJRA were analyzed as a group, perforin levels were not significantly different from those in patients with other JRA types. However, the examination of the individual patterns of perforin staining revealed a small subgroup of JRA patients with a very low perforin content in NK cells. Most of these patients had sJRA. Furthermore, one of them had profoundly decreased proportions of perforin-positive cells in all three major cytotoxic cell populations, a pattern that has been previously reported in patients with MAS [17
] and in the carriers of perforin-deficient FHLH [14
]. Although no overall correlation between perforin expression and NK cell cytolytic activty was noted in our study, we still cannot exclude the possibility that, at least in some patients, reduced perforin expression might have functional significance. In other words, there might be some heterogeneity in the mechanisms underlying NK cell dysfunction in sJRA. The existence of such heterogeneity was also noted in our previous study of MAS patients [17
] that included ethnically diverse Caucasian, African American, and Latin American patients. The ethnic heterogeneity of the patients with JRA included in this study might also underlie the discrepancy between our results and the study by Wulffraat and colleagues [26
], which showed that patients with sJRA as a group had lower perforin expression in cytotoxic effector lymphocytes. That study included a much more ethnically homogeneous population of Dutch children.
Granzyme B is another important component of the perforin-mediated cytotoxicity pathway. In our study both patient groups had granzyme B expression patterns indistinguishable from those seen in healthy controls, suggesting that the observed NK dysfunction is not likely to be related to abnormal granzyme B expression.
The cytolytic activity of NK cells in our study was measured by using NK-sensitive K562 cells, which are lymphoblasts derived from a patient with chronic myelogenous leukemia. The exact receptors involved in the NK-mediated lysis of K562 cells have not yet been identified. However, the lysis of some similar cell lines has been recently shown to be mediated through the natural cytotoxicity receptors (NKp46, NKp30, and NKp44) [31
]. These receptors have important biologic functions in the innate immune system, and their abnormal expression might have a function in the development of NK dysfunction in sJRA.
On the basis of our data, the feature that distinguishes systemic onset JRA from other forms of JRA, and is common to the major hemophagocytic syndromes, is NK cell dysfunction. The exact mechanisms that would link deficient NK cell function and, in some cases, depressed perforin expression with the expansion of activated macrophages are not clear. One possible explanation is that decreased NK function might be responsible for a diminished ability to clear the infecting pathogen and remove the source of antigenic stimulation at early stages of infection [32
]. This would lead to persistent antigen-driven T cell activation associated with an increased production of cytokines, such as IFN-γ and granulocyte/macrophage colony-stimulating factor, that stimulate macrophages. Subsequently, the sustained macrophage activation would result in tissue infiltration and in the production of high levels of TNF-α, interleukin-1, and interleukin-6, which have a major role in the various clinical symptoms and tissue damage.
Several recent studies using perforin-deficient and NKcell-depleted mice indicate that NK cells and perforin-based systems are also involved in the downregulation of immune responses through a direct effect of NK cells and/or perforin-based systems on the survival of activated lymphocytes [33
]. NK dysfunction might therefore lead to a failure to provide homeostatic signals for the removal of activated T cells. For instance, Su and colleagues [33
] demonstrated that the infection of NK-depleted mice with murine CMV results in an exaggerated immune response associated with more persistent expansion of cytotoxic CD8+
T cells that secrete IFN-γ, an important macrophage activator. Another possible explanation is related to the recently discovered ability of NK cells to lyse autologous antigen-presenting cells such as dendritic cells, thus limiting the magnitude of an immune response [37
]. Interestingly, this interaction might involve the above-mentioned natural cytotoxicity receptors [38