Here, we described the inflammatory response during the symptomatic phase of SchS, during treatment with either anakinra or canakinumab, and during relapse after canakinumab withdrawal. During the symptomatic phase, the circulating concentrations of IL-6 were elevated, as were the protein concentrations of IL-6 and mRNA levels of IL-1β in circulating PBMCs. In this phase there was also augmented LPS-induced production of IL-1β and IL-6. Finally, the MRP8/14 and S100A12 concentrations in serum and S100A12 mRNA levels in PBMCs were elevated. All of these were normalized during treatment with either IL-1Ra or anti-IL-1β antibodies, and both therapies led to a shift of the PBMC transcriptome towards the mRNA signature of healthy controls. Clinical relapse several months following canakinumab withdrawal was not associated with a rise in IL-6 serum concentrations, nor with increased ex vivo cytokine production by PBMCs. Interestingly, irrespective of the treatment condition, PBMCs from SchS patients produced less IL-1β and IL-6 when exposed to Pam3Cys or poly:IC when compared to controls.
Taken together, our data point to an IL-1β-driven disorder, which is in line with the clinical efficacy of IL-1β inhibition and with the few reports on in vitro findings. Recently, we reported two variant SchS patients with NLRP3 mosaicism in the myeloid lineage, in whom PBMCs produced high constitutive levels of IL-1β and IL-6, which was abolished by in vitro addition of IL-1Ra [17]. Previously, a few SchS cases were described in which PBMCs or monocytes from symptomatic patients produced more IL-1β and IL-6 upon LPS stimulation compared to control PBMCs [9, 15, 16], and that this could be reversed by in vivo anakinra treatment [15]. Spontaneous IL-1β production by PBMCs was present in one more patient [16], but absent in others [15].
In our patients with NLRP3 mosaicism, the hyperproduction of IL-1β is probably due to an overactive NLRP3 inflammasome. In those without this genetic defect the trigger of the enhanced IL-1β production is still unclear. The increased IL-1β concentrations in turn lead to production of IL-6 and an autocrine or paracrine production of more IL-1β. The latter became clear from the striking reduction in spontaneous IL-1β and IL-6 production by PBMCs from the two NLRP3 mosaic patients when IL-1Ra was added in vitro [17], the reduced IL1B mRNA levels in circulating PBMCs during in vivo IL-1 inhibition, and the lower LPS-induced IL-1β and IL-6 production in PBMCs sampled during treatment.
The increased downstream production of cytokines leads to an enhanced acute phase response with elevation of C-reactive protein (CRP). The amount of IL-6, which is readily measurable in the circulation, is probably responsible for the fever and other signs of systemic inflammation.
How should we envisage the lack of a rise in serum IL-6 and the still downregulated cytokine production when the patients relapsed several months after canakinumab withdrawal? Probably the best explanation is that the relapse is compartmentalized in its early phase, possibly at the level of the skin, which is continuously exposed to pathogen-associated molecular patterns as well as endogenous ligands of pattern recognition receptors (PRRs). Indeed, PRRs were implicated in the pathophysiology of other inflammatory skin diseases, such as AIM2 (absent in melanoma 2) and dectin-1 in psoriasis [30–32]. The IL-1β positive mast cells we recently identified in SchS skin might not only be involved in the chronic urticaria (de Koning et al., submitted), but also in the induction of systemic inflammation.
As the triggers of the IL-1β production are currently unclear, it is of great interest that we find that TLR4 plays a clear role - and not TLR2, TLR3 and TLR6. It implies that either exogenous TLR4 ligands (such as LPS) or putative endogenous TLR4 ligands (like heat-shock proteins, minimally modified LDL, HMGB1, SAA3, MRP8/14, and S100A12 [33, 34]) function as triggers for the attacks. Especially, MRP8/14 and S100A12 are interesting in this regard, as serum levels are associated with disease activity in SchS. Several in vitro studies and mouse models of other inflammatory skin diseases have demonstrated a role for TLR4, e.g., nickel-induced allergic contact dermatitis and graft versus host disease [32]. Intriguingly, expression of both TLR4 and NLRP3 mRNA is extremely low in healthy epidermis, which one might consider a protective strategy preventing continuous stimulation by constituents of the microbiome, for example [31]. We detected a relatively decreased responsiveness to Pam3Cys (TLR2/6 ligand) and poly:I:C (TLR3 ligand) of the SchS patient PBMCs (sampled both during symptoms and anti-IL-1 treatment) compared to control PBMCs. To our knowledge, such divergent responses to TLR2, TLR3, TLR6 and TLR4 ligands have not been reported in inflammatory diseases before. We speculate that the relative hyporesponsiveness to TLR2 and TLR3 ligands might be a protective mechanism in response to the enhanced proinflammatory response to TLR4 ligands.
Our findings of elevated spontaneous cytokine production are reminiscent of findings in patients with the cryopyrin-associated periodic syndrome CAPS in which systemic inflammation is caused by activating NLRP3 mutations. CAPS patient PBMCs constitutively produce IL-1β, and treatment with IL-1Ra results in both a dramatic clinical improvement and substantive downregulation of LPS-induced IL-1β secretion by the patients’ cells in vitro [35]. Typically, the enhanced proinflammatory response to lower LPS concentrations that we observed in SchS is also seen in other autoinflammatory diseases, whereas at the relatively high concentration of 10 ng/ml, the difference is much smaller or absent [36].
Hence, current and previous findings suggest that an inflammasome is primed in PBMCs in SchS as in CAPS, which explains a substantial IL-1β release in the absence of the second hit that is usually required. Also, the relative hyporesponsiveness to TLR2/6 and TLR3 agonists (in this study) and ATP [16], and low IL18 mRNA levels in monocytes despite high IL-18 serum levels [37] suggest the presence of several negative feedback mechanisms.
We previously reported that in the two patients with NLRP3 mosaicism in the myeloid cell lineage, high constitutive IL-1β and IL-6 production by PBMCs was blocked by in vitro addition of IL-1Ra [17]. Here we show that this high baseline production was not impaired or was only partially impaired while patients were treated with IL-1 inhibitors. This implies ongoing activation of IL-1β, and may explain why these two patients had the most severe phenotypes and their disease quickly relapsed upon cessation of anti-IL-1 treatment [13]. TLR2/6 and TLR4 ligands induced the production of IL-1β and IL-6 in these two patients, but a TLR3 ligand did not.
In this study we additionally monitored S100 proteins, B cell and T cell subsets, M-proteins and serum free light-chains. S100A8/A9 (MRP8/14) and S100A12 are secreted after activation of phagocytes via a so-called alternative secretory pathway and cause strong proinflammatory effects on phagocytes and endothelial cells in vitro [24]. S100A8 and S100A9 and S100A12 have been shown to act as endogenous TLR-4 ligands and binding of S100 proteins to TLR-4 consequently induces NFκB expression via the MyD88-dependent pathway [33, 34]. Besides their pathophysiological role as DAMPs S100 proteins are regarded as markers of systemic inflammation [22–24] and especially in CAPS, both MRP8/14 and S100A12 levels mirrored disease activity, and were suggested as a sensitive marker even for subclinical disease [23, 38]. In CAPS patients responding to canakinumab treatment S100A8/A9 serum levels already dropped to the range of normal controls within 8 days [25].
In our study, S100A12 mRNA levels were significantly higher in circulating PBMCs from symptomatic patients than in PBMCs from treated patients and controls, and serum protein levels of both MRP8/14 and S100A12 correlated with disease activity. In a previous report, serum S100A12 protein levels did not correlate with disease activity in SchS, which might be related to the higher levels found in that study [5]. In most patients, MRP8/14 levels were higher than in healthy controls, even when they were asymptomatic under anakinra or canakinumab treatment [24]. Thus, MRP8/14 and S10012 levels are markers of disease activity in SchS. The patient data showed considerable heterogeneity, and in several patients persistent elevated MRP8/14 levels under treatment were found compared to healthy controls, as was previously reported in CAPS patients [38]. This may indicate subclinical disease activity that is not detected by CRP or IL-6 measurements.
Previously, an increase in transitional B cells, decrease in switched-memory B cells and low levels of peripheral blood plasma cells were reported in a patient with SchS when compared to healthy controls. IL-1Ra treatment had no effect on the patient’s B lymphocytes or the IgM M-protein [39]. Our analyses on T cell and B cell subsets showed no differences between active disease and anti-IL-1 treatment, neither were there any differences in T cell subsets between patients and controls. We do not know if long-term IL-1 inhibition would affect the T and B cell compartments.
An M-protein is one of the diagnostic hallmarks of Schs. We demonstrated that it is not a marker for disease activity as the serum concentration of the M-protein was not affected by IL-1 inhibition. Moreover, in our two most severely affected patients, only unquantifiable IgG kappa was found. Still, it cannot be excluded that long-term anti-IL-1 treatment could halt a progressive increase in M-protein concentration in view of the B cell activating property of IL-1β (the longest treatment duration of SchS patients on continuous anti-IL-1 treatment is currently 8 years). Moreover, long-term follow up of many SchS patients is needed to determine if IL-1 inhibition can prevent progression to a lymphoproliferative disorder. Indeed, in some patients with smoldering or indolent multiple myeloma, who were at risk of progression to active myeloma, concomitant treatment with IL-1Ra and dexamethasone was found to decrease the myeloma proliferative rate [40].
The ratio of serum immunoglobulin kappa and lambda light-chain levels is a prognostic factor for multiple myeloma disease progression [27, 28]. No such correlation was seen in our SchS patient cohort, but in one case, the highest free light-chain level was found prior to starting treatment with IL-1 inhibition. More measurements comparing pre- and post-IL-1 inhibition serum light-chain levels are needed to examine a possible association. We conclude that both the intact M-protein and the free light-chain concentrations are stable biomarkers in SchS patients, and are not affected by disease status or therapeutic intervention.