ILD is a life-threatening and increasing recognized complication in children with SJIA [8]. A screening algorithm recommended pulmonary screening for newly diagnosed patients with SJIA with “red flag features” due to increased risk for LD. These “red flag features” included baseline characteristics (young age of SJIA onset, HLA type, high disease activity), respiratory symptoms and features of drug hypersensitivity-like reactions [16]. The incidence of ILD in children with SJIA in this two-center study was 19.9%, consistent with 16% incidence rate of ILD in our previous single-center series [17]. Regardless, these were much higher than the rates reported in prior studies in other countries [4, 7]. The reason for this difference may be that all patients in our cohort underwent HRCT at the diagnosis, and we also found that some patients developed ILD during the course of the disease rather than at the onset. Among them, 4 cases were diagnosed through HRCT due to the lack of significant respiratory symptoms and X-ray findings. The relative low rate of clinical inactive disease due to under-use of biologics may be another potential reason.

Approximately 85.7% of patients with SJIA-ILD presented with arthritis at the onset of the disease in this study. Notably, the presence of axial joint involvement like the cervical spine and large joint involvement like the hip, known as poor prognostic risk factors, was more frequently observed in the ILD group. Laboratory examinations indicated that patients diagnosed with SJIA-ILD exhibited a marked inflammatory response with higher CRP, WBC, ANC, and ESR values and were more likely to exhibit hyperferritinemia and lymphopenia than those without ILD. These clinical features suggested higher disease activity and poor prognostic factors. Patients with lymphopenia were prone to secondary infections. Therefore, we adopted TMP-SMX prophylaxis in reducing the incidence of pneumocystis pneumonia with a favorable safety profile for these patients, consistent with other reports [18, 19].

Systemic juvenile idiopathic arthritis stands apart from other forms of JIA due to its propensity for life-threatening complications. A prime example of such a complication is MAS, a severe cytokine storm syndrome characterized by rampant systemic inflammation, profound hyperferritinemia, coagulopathy, and multiorgan dysfunction. In our study, about 40% of patients with ILD had at least one episode of MAS, significantly higher than that in the non-ILD group, as previously reported [8, 20]. Among these, 28.6% experienced more than two episodes of MAS. Furthermore, patients with SJIA-MAS exhibited a heightened susceptibility to ILD compared with those without MAS in a prior study [4]. Recent investigations had underscored the pivotal role of interferon-gamma (IFN-γ) in the pathogenesis of MAS. Promisingly, preliminary clinical trials evaluating emapalumab, an anti-IFN-γ monoclonal antibody, had yielded encouraging results. However, the reasons for the close association between SJIA-ILD and MAS were yet unclear. Gao et al. [21] discovered the crucial role of IFN-γ in triggering lung inflammation associated with MAS in a mouse model. Indeed, some patients with SJIA-ILD displayed elevated levels of IL-18 and IFN-γ-induced protein chemokine 9 (CXCL9), both specific markers of MAS [22, 23]. Moreover, our study revealed that more than 75% of patients with ILD exhibited SF level exceeding 684 ng/mL, including some cases of subclinical MAS. Therefore, we recommend every patient with SJIA-ILD should serve as a potentially crucial indicator for early screening of MAS.

In this study, only one patient received IL-6 blockade prior to SJIA-ILD onset and none received IL-1 blockade. There was no significant difference in the use of biologics between the ILD and non-ILD groups. The lower usage rate of biologics in our cohort was due to the fact that IL-6 antagonist was not available in China until the end of 2013 and IL-1 antagonist was still not available so far. Additionally, some children and their family could not afford the cost of biologics and had to rely on glucocorticoids or combined DMARDs treatment. A global case study involving 41 patients with SJIA-LD revealed some of which were linked to previous treatments with biological agents [6]. Additionally, patients with SJIA-ILD were reported with frequent occurrences of adverse reactions to biologics, such as delayed-type hypersensitivity or drug reactions with eosinophilia and systemic symptoms (DRESS-like reactions) [6, 8]. DRESS-like reactions typically present with fever, rash, lymphadenopathy, eosinophilia, atypical lymphocytosis and organ involvement [24, 25]. Recent studies have also connected these DRESS-like characteristics to the presence of the HLA-DRB1*15 alleles. Saper et al. [25] had associated these drug allergic reactions with the presence of HLA-DRB1*15 alleles. Interestingly, when comparing patients who stopped using IL-1/IL-6 antagonist permanently to those who maintained their treatment, there were no significant differences observed in terms of mortality rates or Physician Global Assessment of Lung Disease Activity (PGALD) scores [6]. However, DRESS-like reactions had not been observed in our cohort, and whether this was related to ethnic differences remains to be confirmed by further research.

This study assessed the differences in treatment targets between ILD and non-ILD groups based on the short-term (Goal one), the mid-term (Goal two), and the long-time target (Goal three) proposed by Hinze et al. in 2018 [26]. The ILD group required a prolonged time to control inflammation, and the proportion of children achieving inactive status within one year was lower than that of the control group. Lower remission rates indicated a poor prognosis, and 17% of patients with ILD remained in an unremitted state by the end of follow-up, including two cases of death. The mortality rate in our cohort was 5.7%, which was substantially lower than previous estimates ranging from 58–68% [5, 7]. Recent data showed patients with SJIA-ILD had frequent hypoxia and requirement for respiratory support with lower mortality than previously reported [6], similar to the results of our study. Intensive treatment in the early stages of the disease to control disease activity as early as possible may be beneficial in reducing the development of ILD. Considering the variety of treatment approaches within our cohort, it was not feasible to assess the efficacy of any single therapy. The diversity in treatment approaches highlighted the critical need for targeted clinical trials.

Our study has several limitations. First, as a retrospective analysis, it may introduce statistical biases, and treatment strategies were not standardized. Second, the cohort was from only two centers, with a relatively small sample size, highlighting the need for larger-scale data from domestic multicenter studies. Third, while we summarized clinical characteristics and prognosis, we did not investigate the pathogenesis of ILD, including cytokine profiling such as IL-18, CXCL9, MMP7 and HLA type [8, 27, 28]. Further prospective studies are essential to better understand the underlying mechanisms and develop targeted interventions to improve patient outcomes.