Demographic data of patients with ASD
We evaluated the data of 47 patients with ASD (88.1% women; median age 40 years, IQR [28–56]). All patients had undergone laboratory tests, including a complete blood count, liver function tests, C-reactive protein (CRP), and ferritin. Serum samples were obtained from patients with ASD in the active state. Table 1 summarizes the baseline characteristics and the laboratory data of the patients. The principal clinical symptoms of included a high spiking fever 68.9%), skin rash (57.8%), arthralgia (57.8%), sore throat (38.9%), and splenomegaly (37.8%). Of these patients, only 3 (6.7%) were diagnosed with reactive hemophagocytic syndrome. Patients with ASD showed elevated median levels of biological markers that represent disease activity of ASD, including CRP (median 6.8 mg/dl, IQR [2.9–10.9]) and ferritin (median 1159 pg/ml, IQR [310–3887]).
Serum levels of Gal-9 and sTIM-3 in patients with ASD
Serum levels of Gal-9 were determined by ELISA in patients with ASD, patients with RA and HCs. As demonstrated in Figure 1A, the levels of Gal-9 were significantly higher in patients with ASD (median: 21.06 ng/ml, IQR [11.86–36.75]) compared to those in patients with RA (7.58 ng/ml, IQR [5.57–10.20] p < 0.001) and HCs (4.51 ng/ml, [IQR; 3.59–5.36], p < 0.001). Similarly, serum sTIM-3 levels in patients with ASD were significantly higher than those in patients with RA or HCs (Figure 1B).
We also compared these checkpoint molecules according to the disease activity of ASD and RA. In the subgroup patients with active ASD or RA, serum levels of Gal-9 and sTIM-3 were significantly higher in ASD compared to those in RA (Figure 2A, B). In the subgroup patients with inactive ASD or RA, serum levels of Gal-9 were significantly higher in ASD compared to those in RA (Figure 3A). Whereas, there was no significant difference in serum levels of sTIM-3 between patients with inactive ASD and those with inactive RA (Figure 3B). In the enrolled ASD patients, three patients were complicated with hemophagocytosis syndrome (HPS). These three patients with HPS exhibited higher levels of Gal-9 or sTIM-3 compared to those without HPS (Gal-9; 48.7 ng/mL versus 22.2 ng/mL, p=0.111. sTIM-3; 7100 pg/m versus 3314 pg/mL, p=0.032.). Similarly, high serum ferritin levels and circulating IL-18 were observed in these ASD patients with HPS (Ferritin; 35147±55227ng/mL versus 4278±9195ng/mL, p=0.043, IL-18; 145650±103368 pg/mL versus 57370±61187 pg/mL, p=0.082),
We evaluated these checkpoint molecules in patients with sepsis. Despite of the limited number of patients (n=3), elevated levels of Gal-9 (18.8±23.0 ng/ml) or sTIM-3 (8799±7274 pg/ml) were observed in these patients. High serum levels of IL-18 (6971±1168ng/ml) were also demonstrated in these patients. It is possible that infection-related Th1 response  may contribute to the high levels of checkpoint molecules in patients with sepsis. We also analyzed the cellular expression of Gal-9 in peripheral blood mononuclear cells (PBMNc) isolated from few patients with ASD or RA. Although we detected the cellular expression of Gal-9 in PBMNc, we could not find a significant difference in the cellular expression of Gal-9 between ASD and RA patients in remission state (data not shown supplementary file1).
Relationship between serum levels of Gal-9 and laboratory parameters in ASD patients
Serum levels of Gal-9 (Figure 4A) or sTIM-3 (Figure 4B) showed a significant correlation with serum ferritin levels (Gal-9; r = 0.77, p< 0.001, sTIM-3; r = 0.71, p< 0.001), but not with CRP levels (Figure 5A, 5B). As shown in Figure 6, serum Gal-9 (Figure 6A) or sTIM-3 (Figure 6B) levels also exhibited a positive correlation with the disease activity score (Pouchot’s score, Gal-9; r = 0.66, p< 0.001, sTIM-3; r = 0.59, p< 0.001). Positive correlations were demonstrated between serum levels of Gal-9 (Figure 7A) or sTIM-3 (Figure 7B) and IL-18 (Gal-9; r = 0.90, p< 0.001, sTIM-3; r = 0.78, p< 0.001) was demonstrated. We also evaluated the correlations between serum levels of Gal-9 or sTIM-3 with Th1 cytokine, IFN-g, in patients with ASD. Although IFN-g was not detected in the sera from HCs, circulating IFN-g was detected in part of ASD patients (21/47). There was a significant correlation between circulating IFN-g and serum levels of Gal-9 (Figure 8A) or sTIM-3 8Figure 8B) in ASD patients.
To determine whether serum Gal-9 could be used to differentiate ASD phenotypes, we further analyzed the distribution of serum Gal-9 in combination with ferritin, since some ASD patients exhibited polarized to high levels of Gal-9 in the correlation with ferritin values in the two-dimensional map (Figure 4A). All patients were subdivided into three groups based on the presence of the systemic or chronic arthritis phenotype. We then compared the serum levels of Gal-9. There was no significant difference in serum levels of Gal-9, sTIM-3 of ferritin among ASD patient with three phenotypes (Figure 9). Although the biology of serum ferritin remains unclear, various immune regulatory roles have been attributed to extracellular ferritin and correlations between inflammatory cytokines and circulating ferritin had been demonstrated . Indeed, the ratios of inflammatory cytokines to serum ferritins were proposed as immune biomarkers in the infections-mediated inflammations . Therefore, we evaluated the ratio of checkpoint molecules to serum ferritin levels. Therefore, we calculated the ratio of Gal-9/ferritin or sTIM-3/ferritin in ASD Patients with different phenotypes. The ratio of Gal-9/ferritin was significantly higher in patients with the chronic arthritis phenotype than in patients without this phenotype (Figure 10A). Similarly, higher levels of sTIM-3/ferritin ratio were observed in patients with ASD with the chronic arthritis phenotype (Figure 10B).
Longitudinal observation of serum levels of Gal-9 or sTIM-3
To explore the longitudinal changes in Gal-9 or sTIM-3, we included 10 patients with two longitudinal samples (at least 1 month apart). In the longitudinal study, 10 patients with active ASD were followed until they became inactive and then resampled. Serum levels of Gal-9 or sTIM-3 decreased significantly in parallel to ferritin and Pouchot’s score after immunosuppressive treatments (Figure11). Therefore, serum levels of Gal-9 or sTIM-3 in patients with active ASD were diminished following successful treatment and clinical improvement. Among the enrolled ASD patients, serum sample were available in 15 patients who received the used medications (steroid alone n=11, steroid plus immunosuppressants n=4). We compared serum levels of Gal-9 and sTIM-3 between these two groups of pretreated ASD patients, however there was no significant difference (Figure 12).