Analysis of complement biomarkers in systemic sclerosis indicates a distinct pattern in scleroderma renal crisis

Plasma samples were collected from 122 patients who fulfilled the American College of Rheumatology (ACR) criteria for SSc [16]. The disease was classified as dcSSc (n = 41) or lcSSc (n = 81), based on the extent of skin involvement [3]. Samples were collected within a mean (±SD) of 7.5 (8.5) years of disease onset, which was defined as the first non-Raynaud’s manifestation. A summary of baseline characteristics of patients with SSc is presented in Table 1. Healthy control plasma samples were collected from 47 age-matched (mean 52.3 ± 8.7 years) and sex-matched (14 males, 33 females) healthy volunteers with no history of rheumatologic disease. All plasma samples were retrieved in a standardised fashion (non-fasting) and were stored at −80 °C after centrifugation until the experiments. Additionally, we included plasma samples from age- and sex-matched patients with RA (n = 81), PsoA (n = 22) and AS (n = 20). All patients with RA fulfilled the 1987 ACR criteria [17]. Diagnoses of PsoA and AS were based on clinical judgment by the treating physician and included radiographic examinations when applicable. All patients with AS had axial involvement but no clinical signs of peripheral arthritis, whereas the patients with PsoA had peripheral arthritis but no clinical signs of axial involvement.

Anti-nuclear antibodies (ANA) and anti-centromeric antibodies (ACA) were analysed by an indirect immunofluorescence technique using the human Hep-2 cell line as a substrate; anti-topoisomerase I antibodies (ATA) were analysed by ELISA; and anti-RNA polymerase III antibodies (ARA) were analysed by immunoprecipitation. Serum cartilage oligomeric matrix protein (COMP) was measured with a commercial sandwich ELISA using two monoclonal antibodies directed against separate antigenic determinants (AnaMar, Lund, Sweden) [18]. Subjects were classified as having dcSSc or lcSSc according to LeRoy et al.’s criteria [3]. The severity of skin involvement was determined by the modified Rodnan skin score (mRSS) [19]. Disease duration was determined as the time from the first non-Raynaud’s manifestation.

Plasma concentrations of C4d [12], C3bBbP [20] and sTCC [20] were measured by specific sandwich ELISAs in plasma ethylenediaminetetraacetic acid (EDTA) samples as described previously. Readout of each of these assays was given in complement activation units (CAU), a defined arbitrary unit set for the International Complement Standard #2 (ICS#2) sample, which is serum-pooled from about 1000 healthy individuals and incubated with activators of all three complement pathways [20].

Plasma EDTA samples collected from patients with SSc were tested for activity of classical or alternative complement pathways by haemolytic assays performed as described elsewhere [21], with small modifications. Briefly, the classical complement pathway was activated on antibody-sensitised sheep erythrocytes. To overcome the inhibitory effect of EDTA, plasma samples were diluted 1:100 in DGVB²⁺ buffer (2.5 mM veronal buffer, pH 7.35, 72 mM NaCl, 140 mM glucose, 0.1% gelatin, 1 mM MgCl₂ and 0.15 mM CaCl₂). For the alternative pathway, plasma samples were diluted 1:20 in Mg-ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (Mg-EGTA) buffer (2.5 mM veronal buffer, pH 7.3, 70 mM NaCl, 140 mM glucose, 0.1% gelatin, 7 mM MgCl₂ and 10 mM EGTA) and added directly to rabbit erythrocytes.

Kidney and skin biopsies collected from patients with SSc were fixed in formalin and embedded in paraffin. Sections of 4-μm thickness were cut and automatically pre-treated using the PT Link system (Dako/Agilent Technologies, Santa Clara, CA, USA) and then stained in an Autostainer Plus (Dako) with rabbit anti-human C3c polyclonal antibody (P0062; Dako) at a final dilution of 1:5000 for 30 minutes. Subsequently, the EnVision Flex HRP kit (Dako) was applied to the sections for 20 minutes to detect primary antibodies, followed by 3,3′-diaminobenzidine (DAB) reagent for visualisation.

Obtained data on the appearance of complement activation markers and haemolytic activity were not distributed normally; in each complement measurement there were outliers, and some patient subgroups were rather small. Therefore, we used non-parametric statistical methods throughout the whole analysis. The Kruskal-Wallis and Mann-Whitney U tests were used to compare multiple groups or two groups, respectively. Spearman’s correlation was used to analyse relationships. p values <0.05 were considered significant. Calculations were performed with Prism 5.0 (GraphPad Software, La Jolla, CA, USA) and IBM SPSS Statistics version 20 (IBM, Armonk, NY, USA) software.

We tested three different complement activation markers: (1) C4d, corresponding to activation of the classical pathway; (2) the fluid-phase alternative C3 convertase (C3bBbP), reflecting activation of the alternative pathway as well as the amplification loop enhancing the cascade at the level of convertases; and (3) sTCC, which evaluates the lytic (terminal) pathway (Fig. 1). All autoimmune patients, regardless of diagnosis, had 5- to 15-fold increased levels of all three markers, thus confirming ongoing complement activation. Regarding SSc, C4d was increased to a similar degree as for AS and PsoA, but to a lesser extent than in RA (p = 0.05 by Kruskal-Wallis test). Interestingly, the alternative pathway was activated significantly more strongly in patients with SSc than in patients with RA (p < 0.001). All groups of patients showed similar levels of sTCC. As expected, healthy control subjects showed low levels of all three markers.

Because sTCC is a marker of the terminal complement pathway, which is fuelled by all complement pathways, we analysed which of the early complement markers correlate to appearance of sTCC in patients with SSc and other autoimmune diseases. Only C3bBbP, and not C4d, was significantly correlated to sTCC in all groups of patients. However, in patients with RA, C4d was significantly correlated to C3bBbP, which may indicate the role of the amplification of the classical pathway via alternative convertases in overall complement activation in this patient group. Correlation coefficients and p values are given in Table 2.

To find out whether complement activation correlates to the severity of fibrosis expressed by mRSS, we performed Spearman’s correlation analysis and found weak, borderline significant correlation between mRSS and C4d (Table 3). Furthermore, COMP, which is expressed by fibroblasts in SSc, is associated with mRSS [22] (in the present cohort, Spearman’s r = 0.47, p = 0.0005) and is known to activate the alternative pathway [23]. However, we did not observe a statistically significant correlation between complement activation markers and COMP levels in plasma, implying that it is likely not a major complement trigger in these patients, at least not when released into blood.

We classified the patients into subgroups depending on the form of disease (lcSSc vs. dcSSc); typical complications, such as pulmonary arterial hypertension (PAH), scleroderma renal crisis (SRC), pitting scars, ulcers, telangiectasias; or the presence of certain types of anti-nuclear autoantibodies, including ACA, ATA, ARA, ANA without ACA, ATA or ARA, and finally ANA-negative. We did not observe significant differences in C4d, C3bBbP and sTCC levels between men and women or between patients with lcSSc and those with dcSSc (Table 4). Also, concentrations of these markers were equally distributed within patients with and without PAH, pitting scars, ulcers, and telangiectasias. Importantly, a different distribution of complement activation markers was detected in patients with SRC, who had significantly higher amounts of C4d (p = 0.036) and significantly lower levels of C3bBbP (p < 0.001) and sTCC (p = 0.003) than patients without kidney involvement. When overall distribution of autoantibodies was analysed by the Kruskal-Wallis test, we found a significant difference for sTCC (p = 0.016) and a difference on the border of significance for C3bBbP (p = 0.055). Although not statistically significant (p = 0.095), there was a trend for C4d, according to which patients without any antinuclear autoantibodies presented with low levels of this marker. Furthermore, patients with ARA positivity showed a pattern of complement activation markers similar to that of the patients with SRC (high C4d, low C3bBbP and TCC). Additional Kruskal-Wallis analysis performed for a single type of autoantibody revealed that the ARA-positive group differed significantly from patients without any antinuclear antibodies (ANA-negative) at a p level <0.05 for all complement markers. Also, the ARA-positive group had significantly lower amounts of C3bBbP and TCC than ARA-negative patients (p < 0.01). Interestingly, 8 (47%) of 17 patients with ARA presented also with SRC, a group that accounted only for 15% of all patients with SSc (18 of 122). Taken together, patients with ARA were, as expected, overrepresented among those with SRC and, importantly, manifested the same pattern of complement activation markers.

The levels of C3bBbP and sTCC in the SRC group (means 656 CAU and 4.22 CAU, respectively) were lower than in other patients with SSc; however, these levels were still considerably higher than analogous means for healthy control subjects (79.7 CAU and 1.64 CAU, respectively), indicating that complement activation takes place during the renal crisis. At the same time, C4d, which is a marker of classical pathway activation and the end-degradation product of C4b component, increased to significantly higher levels than in the non-SRC group. One possible explanation for these observations is that during SRC the classical pathway is strongly initiated by autoantibodies, and then an amplification loop driven by the alternative pathway takes over the cascade, leading to systemic depletion of its components, including C3 and factor B proteins. Such depletion is also reflected in a reduction of soluble sTCC compared with patients with no SRC episodes. To confirm the scenario of ongoing complement consumption in SRC, we analysed haemolytic activity of plasma samples from patients with SSc. Residual complement activity was examined in assays testing the classical and alternative pathways (Fig. 2). Of all classifications made for the whole SSc cohort, only the SRC-positive group showed significantly lower haemolytic activity in the alternative pathway (p = 0.0107) and a borderline significant drop of haemolytic activity in the classical pathway (p = 0.063), which confirmed systemic complement depletion, most probably at the level of alternative convertases.

Because the SRC subgroup showed a distinct pattern of complement activation markers, we searched for the primary source of a putative strong complement activation deduced from lower levels of C3bBbP and lower haemolytic activity of plasma samples. Kidney biopsies from 5 of 18 patients with SRC included in the study were available, and we examined these for C3b deposition by immunohistochemistry. C3b deposits were detected in glomeruli of two patients and in tubules of one more patient (Fig. 3), whereas C3b staining was negative for kidneys of two remaining patients. These findings indicate that ongoing complement activation in kidneys may be one of the hallmarks of SRC. However, it does not explain all cases, and other sources of activation are also plausible.

Searching for further explanations for complement activation in patients with SRC, we followed the overrepresentation of ARA-positive patients in the SRC group and analysed C3b deposition in skin biopsies obtained from ARA⁺, ARA⁻ and ARA⁻/ACA⁺ patients. Material from four patients in each group was available. We observed C3b deposition in skin tissue from all ARA⁺ patients, but other groups also had at least two of four slides that were positive (Fig. 4), so there was no dramatic difference between the groups. The staining was localised to the dermal layer and most prominent in the endothelium of small dermal arteries. Less staining was associated with the collagenous fibres of the dermis. In the basal layer of the epidermis, cells were discovered containing what seemed to be stain. However, these cells were melanocytes, and the pigment was melanin, not DAB chromogen (Fig. 4).