Sjögren syndrome (SS) is a systemic autoimmune disorder that primarily affects the lachrymal and salivary glands, leading to compromised secretory functions, but may also have systemic manifestations. The pathogenesis of the syndrome is not fully understood, but the hallmark of disease is the lymphocytic infiltration of the exocrine glands.1 In the early histopathological lesions periductal infiltration of activated T cells is characteristic, while in advanced and chronic lesions B cells predominate and the infiltrate extends to the acinar epithelium.2 Macrophages and dendritic cells also appear in intense lesions.2 In this unique environment, local alterations in the synthesis of cytokines and chemokines initiate and/or perpetuate the autoimmune lesions. Systemically, increased secretion of multiple cytokines has been documented, albeit their role in disease remains unclear.3 Among the cytokines detected in the inflamed glands is tumour necrosis factor α (TNFα),4 and in certain autoimmune conditions, inhibition of TNF has been shown to be effective in suppressing tissue inflammation. With the aetiopathogenesis of SS being mainly unknown and the lack of effective treatment,5 the use of anti-TNF agents emerged as a potential therapeutic modality, given the beneficial effect of this treatment in rheumatoid arthritis (RA),6 as well as in other inflammatory arthritides and diseases.7 8

In two recent randomised, placebo-controlled clinical trials, inhibitors of TNF failed to ameliorate signs and symptoms of SS.9 10 To examine the underlying molecular mechanisms responsible for the clinical inefficacy of anti-TNF treatment in SS, we explored systemic immune parameters in a cohort of patients with SS treated with etanercept or placebo in a double-blind, randomised study of 28 patients.10 Failure of etanercept11 suggests alternative mechanisms of immunopathogenesis in SS. Our data demonstrate detectable measures of systemic immune activation in patients with untreated SS and increased cytokine and chemokine levels in the circulation of patients with SS before etanercept treatment. After subcutaneous delivery of etanercept (25 mg) for 12 weeks, not only was there no significant amelioration of multiple disease parameters but also, importantly, etanercept did not reduce the evidence of systemic immune activation and, in fact, resulted in a significant rise in plasma levels of TNFα. Collectively, these and other recent data12 implicate TNFα-independent pathways of pathogenesis in SS and provide insight into intervention strategies.

MATERIAL AND METHODS

Study design

A double-blind, randomised pilot study of etanercept versus placebo therapy included 28 patients (14 in each group) with SS, as determined by the American–European Consensus Group criteria for SS.13 Patients were stratified according to whether they had primary or secondary SS10 and assigned to the active-treatment group (25 mg etanercept, twice weekly) or placebo group for 12 weeks. The protocol was approved by the Institutional Review Board of the National Institute of Dental and Craniofacial Research.

Histopathological evaluations of salivary gland lesions at diagnosis were reported as inflammatory focus scores (FS) according to Chisholm–Mason’s classification14 and to American–European criteria.15 Clinical assessments of disease activity were performed at study entry and at 4, 8 and 12 weeks thereafter.

Immunohistochemistry

Paraffin-embedded tissues were sectioned (four from each group), deparaffinised and rehydrated, followed by heat-induced epitope retrieval. Sections were blocked with the corresponding preimmune serum and incubated overnight at 4°C with primary antibodies to TNFα and TNF receptor (TNFR; Abcam, Cambridge, Massachusetts, USA). Immunolabelling was detected using a biotinylated secondary antibody followed by visualisation with a horseradish peroxidase labeling kit (Invitrogen, Carlsbad, California, USA) and diaminobenzidine staining. Negative controls were carried out by replacing the primary antibody with preimmune serum.

Isolation of plasma and peripheral blood mononuclear cells (PBMCs)

Peripheral blood (60 ml) was collected from healthy volunteers (NIH, Department of Transfusion Medicine, Bethesda, Maryland, USA) and from patients with SS at baseline and at the end of treatment (12 weeks). Blood samples were not available from all patients at every visit. Plasma was removed from heparinised blood after centrifugation and stored at −80°C until analysed. PBMCs were isolated by Ficoll–LSM (Lymphocyte Separation Medium, ICN Biomedicals, Aurora, Ohio, USA) gradient centrifugation for flow cytometry. Whole blood cultures were stimulated with lipopolysaccharide (LPS, O55:B5, Sigma, St Louis, Missouri, USA; 0, 50, 100 ng/ml) overnight and plasma/supernatant was removed after centrifugation for cytokine analysis.

Flow cytometry

Expression of PBMC cell surface markers was assessed by flow cytometry. Cells were incubated with monoclonal antibodies specific for T cell subsets (CD3, CD4, CD8), B cells (CD20) and monocytes (CD11b) (Becton Dickinson, San Jose, California, USA). Antibodies to activation antigens (CD25 and HLA-DR; Becton Dickinson), costimulatory molecules (CD40; Biosource, Camarillo, California, USA and B7; Becton Dickinson), TNFR I and II (R&D Systems, Minneapolis, Minnesota, USA), chemokine receptors (CXCRI, II), adhesion molecule (l-selectin) and Fc receptor (FcR, CD32; low affinity FcR for aggregated Ig: immune complexes) (Becton Dickinson) were included. The antibodies were conjugated with either fluorescein isothiocyanate or phycoerythrin. Stained cells were analysed by FACScalibur flow cytometer (Becton Dickinson) according to forward- and side-scatter parameters and results were expressed as percentage positive cells.

Multiplex cytokine and chemokine assay

Cytokines and soluble cytokine receptors measured in the plasma with a bead-based multiplex cytokine assay (Biosource International, Camarillo, California, USA) included interleukin 1β (IL1β), IL1 receptor antagonist, IL2, IL2R, IL4, IL5, IL6, IL7, CXCL8 (IL8), IL10, IL12p40, IL13, IL15, IL17, interferon α (IFNα), IFNγ, TNFα, granulocyte-macrophage colony stimulating factor, CCL2 (monocyte chemoattractant protein-1), CCL11 (eotaxin), CCL3 (macrophage inflammatory protein-1α (MIP-1α)), CCL4 (MIP-1β), CCL5 (regulated upon activation, normal T cell expressed and secreted (RANTES)), CXCL9 (monokine induced by IFNγ) and CXCL10 (interferon inducible protein-10). Samples and standards were analysed in duplicate and only variation coefficients <15% were accepted.

Statistics

Evaluation of differential expression of cell surface antigens and differential levels of peripheral cytokines was accomplished with Mann–Whitney U test. p Values of <0.05 were considered significant. Correlations between cytokine values and focus scores were done using both the non-parametric Spearman test and the Fisher exact test after determining cut-off points with receiver operating characteristic curve analysis. All the analyses were performed using the statistical package SPSS version 13.

RESULTS

Clinical outcome

In this double-blind, randomised pilot study of etanercept versus placebo treatment in 28 patients with SS, 14 received etanercept and 14 placebo for 12 weeks. In both treatment groups, 11 of 14 patients had primary SS (pSS) and 3 had SS secondary to RA, systemic lupus erythematosus or CREST syndrome. Throughout the course of treatment, there were no significant differences between the groups in the primary clinical outcomes, which included ocular, oral and laboratory variables, as previously reported.10 12 In one of the patients with SS secondary to RA, a considerable improvement in arthritis symptoms was noted (data not shown), reflecting the documented etanercept efficacy in RA.

Markers of immune activation in patients with SS

The lack of significant clinical improvement in the pSS main clinical outcomes during etanercept treatment suggested TNF-independent pathways, despite the abundant expression of TNFα and TNFαR in the immunopathological lesions of SS (figs 1A and B). To explore potential systemically detectable immunological aberrations in patients with SS as a means to evaluate the effect of etanercept treatment on the immune profile of SS, we studied peripheral mononuclear cell populations in SS before and after treatment. By flow cytometric analysis of antibody-stained PBMCs from patients (n = 19) and controls (n = 19), we found no significant differences in the total percentages of T cells (CD3), B cells (CD20) or monocytes (CD11b) between patients and controls (data not shown). The expression of some activation markers (CD25 and HLA-DR) (fig 1C), costimulatory molecules (CD40), FcR (CD32) and adhesion molecules (l-selectin, intercellular adhesion molecule-1) (data not shown) was similar in patients and healthy subjects, although evidence of immune activation in the PBMCs of patients with SS was suggested by a significant increase in expression of TNFRI (p<0.05) and TNFRII (p<0.05). Moreover, chemokine receptors CXCRI and CXCRII were significantly raised, as previously reported.16 Nonetheless, the heightened expression of these activation markers in pSS was not altered by etanercept treatment, at least not during the 12 week evaluation of this study (fig 1C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Peripheral blood mononuclear cell (PBMC) phenotypes in healthy control subjects and in patients with primary Sjögren syndrome (pSS) before and after etanercept (ETN) therapy. Immunohistochemistry for tumour necrosis factor α (TNFα) (A) and TNFRα (B) in lesions from patients with SS (n = 4, representative images shown). (C). FACS staining for cell surface activation markers on freshly isolated PBMCs from healthy controls (n = 19) and patients with pSS at the entry visit before treatment (pre-ETN, n = 19) and after etanercept treatment (post-ETN, n = 9). Levels of CD25 (IL2R), HLA-DR, TNFRI and TNFRII and chemokine receptors CXCRI and CXCRII are shown for individual patients and horizontal lines represent mean levels of expression for each marker (*p<0.05).

Systemic cytokine profile of patients with SS

Based on the identified differences between pSS and control subjects in PBMC phenotype, we next monitored systemic cytokine profiles as an additional measure of immune activation which might be influenced by TNF inhibitors. First, we categorised the multiple detected soluble factors into functionally related groups, including Th1, Th2 and Th17 cytokines, growth factors, interferons, chemokines, soluble receptors and cytokine receptor antagonists. Our studies showed that patients with pSS have an altered pattern of cytokine blood levels compared with healthy controls. Significant increases in the majority of the 25 soluble factors measured clearly indicated evidence of aberrant immune function. Interestingly, rather than a polarised T helper (Th) cell lineage cytokine profile, cytokines associated with all three Th lineages (Th1, Th2 and Th17) were increasingly present in the plasma of patients with SS. For example, of the Th1-related cytokines, IL1β, IL12p40, IL2, IL7 and TNFα were significantly raised in patients with pSS, and Th2-linked IL4, IL5 and IL6 were also significantly upregulated in the patient group (table 1). In addition to IL2, soluble IL2 receptors were also higher in pSS (p = 0.01). Whereas IFNα, which has previously been linked to pSS disease pathogenesis,12 17 18 was remarkably increased (p<0.001), blood levels of IFNγ were lower in patients with pSS than controls (table 1). Finally, an increase of the recently recognised Th17 lineage cytokine, IL17, reflected a more global T-cell activation.

When circulating chemokines were quantified, a heterogeneous pattern of both increased and decreased chemokine levels was seen in patients with pSS. Of the chemokines quantified, both CXCL8 (IL8) and CCL5 (RANTES) were significantly reduced in the plasma of the pSS group compared with healthy controls, concomitant with an appreciable increase in CCL3 (MIP-1α), CCL4 (MIP-1β), CCL11 (eotaxin), CXCL9 (monokine induced by IFNγ) and CXCL10 (interferon inducible protein-10) (table 2).

Correlation of plasma cytokine levels with disease markers

The differential expression of multiple soluble factors between patients and controls was indicative of immune activation, but it remained unclear whether these markers of stimulated immune function in the blood correlated with known markers of SS disease: SSA and SSB antibodies, high erythrocyte sedimentation rate (ESR), increased IgG and FS (histological index for severity of salivary gland involvement according to the American–European classification).14 Levels of cytokines measured in the peripheral blood of patients with untreated pSS, including TNFα, were not associated with the advancement of the local inflammatory lesions (p>0.05). Moreover, cytokine levels were not significantly linked to the presence of anti-SSA or anti-SSB autoantibodies, or to raised ESR (>30 1st mm/h) or IgG (>15.00 g/l) (data not shown).

Plasma cytokine levels before and after etanercept treatment

Having documented a pattern of immunological dysregulation in patients with pSS, we next examined whether etanercept treatment influenced these parameters. To this end, we compared levels of soluble immune mediators before and after treatment. The initially unexpected absence of significant diminution of blood levels of most cytokines and chemokines, as shown for representative Th1 (IL2, IL12p40), Th2 (IL4, IL10) or Th17 (IL17) cytokines, as well as IFNα (fig 2) did, however, correspond to a lack of evidence of a clinical response. Paradoxically, the only cytokine significantly altered by etanercept treatment was TNFα, but the TNFα levels increased rather than declined (p = 0.003) (figs 3A and B). TNF levels in patients treated with placebo did not change significantly (figs 3A and C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Plasma cytokine levels in healthy control subjects (HC) and patients with primary Sjögren syndrome (pSS) before and after etanercept treatment. Plasma cytokine levels were measured with a multiplex ELISA in HC (n = 11), patients with pSS before treatment (pre, n = 22), after placebo treatment (post-PL, n = 11) and after etanercept treatment (post-ETN, n = 11). Levels (pg/ml) for IL2 (A), IL12p40 (B), IL4 (C), IL10 (D), IL17 (E) and IFNα (F) are shown for individual patients and horizontal lines represent median levels of expression for each group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Plasma tumour necrosis factor α (TNFα) levels increase after etanercept treatment. (A) TNFα levels were measured in healthy controls (HC, n = 11), patients with primary Sjögren syndrome (pSS) at baseline (n = 22) and after placebo (post-PL, n = 11) and after etanercept (post-ETN, n = 11); *p<0.05. (B) TNFα levels in individual patients with pSS before and after treatment with etanercept. (C) TNFα levels in individual patients with pSS before and after treatment with placebo.

Because of the significant increase in TNFα in the circulation of patients with pSS treated with etanercept for 12 weeks, we explored potential mechanisms for this drug effect. We obtained peripheral blood from patients before and after etanercept treatment and cultured the blood that was replete with etanercept or placebo in the absence or presence of a Toll-like receptor 4 agonist (LPS) to induce TNFα production. LPS (50 ng/ml) induced TNF secretion in the healthy control cultures, which declined at 100 ng/ml (fig 4A). In the pSS blood, at 50 ng/ml, LPS induced about twofold more TNFα protein than seen in the healthy controls, which further increased at 100 ng/ml. LPS-stimulated whole blood cultures from etanercept-treated patients with pSS released even higher levels of TNF (fig 4A). We evaluated multiple additional immune mediators in these cultures and identified one potential contributing factor to the enhanced expression of LPS-inducible TNFα. In the cultures of whole blood from pretreated patients with pSS, we noted a lack of LPS-triggered secretion of IL10, an inhibitor of TNFα production,19^–21 compared with that seen in the parallel cultures from healthy control subjects (fig 4B). Moreover, in the blood cultures obtained from etanercept-treated patients and thus containing plasma levels of the drug carried over into culture, there was also clear absence of LPS-induced IL10 (fig 4B). Re-assessment of median IL10 levels in the circulation of patients with pSS before and after etanercept showed less IL10 than in healthy control subjects, albeit not significant (fig 2D), which may warrant further consideration. Collectively, our data are consistent with a TNF-independent pathway of disease pathogenesis in pSS and suggest that if TNFα plays a role, it is not a linchpin in orchestrating downstream sequelae.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Tumour necrosis factor α (TNFα) and interleukin 10 (IL10) induction in whole blood cultures from patients and control subjects. (A) Whole blood from healthy controls (HC, n = 8) and patients with primary Sjögren syndrome (pSS) before (pre-ETN, n = 8) and after treatment with etanercept (post-ETN, n = 8) was cultured (1 ml) overnight (18 h) with or without lipopolysaccharide (LPS; 50, 100 ng/ml). TNFα levels in the cultures were measured by a multiplex ELISA. Data represent mean (SEM) for each LPS dose. (B) Supernatants from the LPS-treated whole blood cultures as in (A) were tested for IL10 levels by a multiplex ELISA. Data represent the mean (SEM).

DISCUSSION

With the treatment of SS remaining mainly empirical and efforts continuing to focus on identifying molecular targets for treatment, TNF inhibition emerged as a viable option for treatment. TNFα has been identified systemically and locally in the exocrine glands of patients with SS,22 23 and contributes to the establishment and progression of exocrine gland pathology in animal models.4 24 Furthermore, the use of TNF inhibitors has led to successful treatment of many autoimmune conditions.8 Unfortunately, in SS the use of TNFα antagonists has resulted in a disappointing lack of efficacy.9 10 In an effort to understand this paradox, we explored systemic immune parameters in patients with SS before and after treatment for 12 weeks with the TNFα inhibitor, etanercept. We show evidence of systemic immune activation in SS based on our phenotypic characterisation of PBMCs and on the heightened levels of multiple proinflammatory cytokines. Moreover, we document that etanercept not only failed to ameliorate disease, but was ineffectual in dampening these measures of immune activation.

Based on our patient immune profiling before treatment, we recorded a notable increase in expression of select cell surface markers associated with activation, including the chemokine receptors CXCRI and CXCII and the TNFRs, which may reflect recruitment of activated immune cells to sites of disease. Several prominent Th1, Th2 and Th17 cytokines and chemokines were also upregulated in the blood of patients with pSS, suggesting a global activation of these T-cell lineages, in contrast to previous reports which suggested a Th2 polarisation in patients with SS, with high levels of circulating IL6 and IL10.25 26 In addition to evidence of raised levels of TNF, we identified heightened circulating IL17, a cytokine secreted by a recently defined lineage of Th17 cells.27^–29 In producing IL17, these cells provide defence against bacteria and mediate inflammation, but aberrant Th17 responses have been implicated in a growing list of autoimmune disorders.27 30 31 Another noteworthy difference between peripheral cytokine levels of healthy controls and patients with pSS was the increase in patient levels of IFNα. Studies in salivary gland biopsies of patients with SS have previously shown increased expression of IFNα-inducible genes and the presence of a large number of IFNα-producing plasmacytoid dendritic cells within the lesions compared with control tissues.18

In an effort to establish possible correlations of this differential cytokine pattern in SS with disease progression, we compared clinical measures of disease, including ESR, serum IgG, anti-SSA (Ro) and/or anti-SSB (La) with the degree of glandular destruction (FS) in our cohort. Importantly, TNFα secretion was neither associated with the presence of systemic clinical markers nor with increasing severity of the pathogenic lesions, further indicative of a lack of connection between TNFα and disease pathogenesis.

After the 12-week course of etanercept, raised extracellular activation markers on PBMCs remained unchanged. Moreover, no decline in peripheral cytokines was evident in the treated patients, consistent with the lack of efficacy of this agent for SS in clinical outcomes. Most surprising was the statistically significant increase in TNFα after treatment with the inhibitor, although a further detriment in clinical outcome was not substantiated during this interval of treatment. The ability of etanercept, similar to naturally produced soluble TNFRs,32 not only to bind and neutralise TNFα, but also to stabilise and prolong its half-life has previously been demonstrated. The possibility of rheumatoid factor binding the Fc portion of etanercept and further increasing the plasma half-life of etanercept-bound TNFα was not evident in that increased levels of TNFα did not correlate with high plasma levels of rheumatoid factor (data not shown). In several other clinical trials with etanercept, a similar paradox was noted with raised circulating levels of TNFα.33^–35 Finally, an etanercept-mediated increase in T-cell TNF secretion was seen in patients with ankylosing spondylitis, even when they experienced excellent pain relief and clinical resolution.36

How an antagonist of TNF might augment its levels in pSS has remained a puzzle. To explore the mechanism further we cultured blood from the patients before and after treatment in the presence of a Toll-like receptor agonist, LPS, to induce TNF secretion. Increased production of TNFα after treatment with etanercept indicated that the high levels of TNFα in the patient serum may not be solely the result of cytokine retention in the periphery, but may also stem from a negative feedback pathway or may be associated with reduced suppressive mechanisms, such as inadequate IL10. In this regard, we detected a failed induction of IL10 in the blood cultures of patients with pSS, which may reflect an imbalance in proinflammatory and anti-inflammatory cytokine levels.

Collectively, we show that multiple immune pathways are activated in pSS, but not necessarily linked with disease advancement. This hyperactivated immune profile is not modulated by etanercept, nor are clinically detectable measures of disease. Based on these data and that of others, it appears that TNF does not represent a key mediator of disease and thus may not be an appropriate intervention target. Our data suggest that other immunological targets must be examined locally and systemically for consideration in future approaches to alleviate the pathogenesis of this disease.