Systemic Juvenile Idiopathic Arthritis-Lung Disease: Characterization and Risk Factors

Abstract

Objective:

Systemic Juvenile Idiopathic Arthritis (SJIA) is associated with a recently recognized albeit poorly defined and characterized lung disease (LD). Our objective is to describe clinical characteristics, risk factors, histopathologic and immunologic features of SJIA-associated LD (SJIA-LD).

Methods:

Clinical data was abstracted from medical records, and epidemiologic, cellular, biochemical, genomic analysis, and transcriptional profiling analyses were performed.

Main Results:

Eighteen patients with SJIA-LD have been evaluated since 2010. Radiographic findings included diffuse ground-glass opacities, subpleural reticulation, interlobular septal thickening, and lymphadenopathy. Pathologic findings included patchy but extensive lymphoplasmacytic infiltrates and mixed features of pulmonary alveolar proteinosis (PAP) and endogenous lipoid pneumonia (ELP). Compared to SJIA patients without LD, children with SJIA-LD were younger at SJIA diagnosis (OR=6.5, P=0.007), had prior episodes of macrophage activation syndrome (MAS) (OR=14.5, P<0.001), have had adverse reactions to biologic therapy (OR 13.6, P<0.001), and have higher serum IL-18 (27,612 vs. 5,413 pg/mL, P=0.047). SJIA-LD patients lacked genetic, serologic, or functional evidence of GM-CSF pathway dysfunction typical of familial or autoimmune PAP. Additionally, bronchoalveolar lavage (BAL) rarely demonstrated proteinaceous material and had less lipid-laden macrophages than seen in primary PAP (66.1% vs 10.5%, P<0.001). SJIA-LD BAL fluid contained elevated levels of IL-18 and IFNγ-induced chemokines CXCL9–10. Transcriptional profiling of SJIA-LD lung tissue identified upregulated type II interferon and T-cell activation networks. This signature was also present in SJIA-LD lung tissue sections lacking substantial histopathological findings, suggesting it may precede and drive lung pathology.

Conclusions:

Pulmonary disease is increasingly detected in children with SJIA, particularly in association with MAS. This entity has distinct clinical and immunologic features and represents an uncharacterized inflammatory LD.

INTRODUCTION.

Systemic Juvenile Idiopathic Arthritis (SJIA) is a chronic inflammatory disease of childhood, representing 10–15% of the total JIA population with a broad peak of onset between 1 and 5 years of age (1). At disease onset, SJIA is characterized by arthritis, daily spiking fevers, an evanescent rash, and a variety of other extra-articular features (1–3). In many SJIA patients, systemic features tend to subside over time, and the long term outcome of the disease depends on the evolution of the arthritis (4,5). About 10–15% of SJIA patients develop life-threatening overt macrophage activation syndrome (MAS), leading to widespread hemophagocytosis and overproduction of cytokines including IFNγ, cytopenias, liver dysfunction, and coagulopathy (6–8).

Before the year 2000, SJIA was treated with non-steroidal anti-inflammatory drugs (NSAIDs), corticosteroids, and methotrexate (MTX) (1). Subsequently, the introduction of biologics neutralizing the activity of IL-1 (anakinra and canakinumab) and IL-6 (tocilizumab) markedly improved the long-term outcome of the disease and dramatically reduced the need for corticosteroids (9–18). Surprisingly, improved control of SJIA did not protect against MAS, rates or mortality of which remain similar (19–22).

While acute pulmonary dysfunction can be seen in SJIA particularly during acute MAS, reports of chronic parenchymal lung disease in SJIA had been exceptionally rare and mild in North America prior to 2013 (23,24). Conversely, in recent years children with SJIA have been increasingly found to have more severe and life-threatening lung disease. In 2013, 25 SJIA patients were reported with chronic pulmonary disease including pulmonary alveolar hypertension (PAH), interstitial lung disease (ILD) and pulmonary alveolar proteinosis (PAP) (25). PAP occurs in a diverse group of disorders leading to alveolar surfactant accumulation due to hereditary or autoimmune disruption of granulocyte-macrophage colony-stimulating factor (GM-CSF) signaling (primary PAP), impaired macrophage numbers or function (secondary PAP), or surfactant production disorders (congenital PAP) (26). The majority of SJIA patients with lung disease were diagnosed after the year 2000, had refractory systemic disease with MAS, and had been exposed to numerus synthetic and biologic disease-modifying anti-rheumatic drugs (DMARDs). Kimura and colleagues suggested that these complications could be related to uncontrolled SJIA disease activity and/or related to medication exposure such as cytokine-directed biologics. A subsequent cohort of 61 new cases expanded this phenotype, and similarly reported a 37% mortality rate (27). However, the immunopathologic basis of this chronic lung disease remains unclear.

At Cincinnati Children’s Hospital Medical Center (CCHMC), we have had a sustained increase in patients evaluated for SJIA and severe lung disease since 2014 (Figure S1). The objective of this study was to describe clinical characteristics in patients with SJIA-associated lung disease, define the inflammatory features of this condition, and identify candidate risk factors for its development. Together, we find that this entity has distinct clinical and immunologic features from other disorders, including known causes of PAP, and represents an uncharacterized inflammatory ILD (SJIA-LD). Some of the patients and results of these studies have been previously reported in the form of abstracts (27–30).

METHODS

Study population and design.

This was a prospective cohort study; patients were identified upon lung disease detection or referral to CCHMC for consultation or second opinion. This study was approved by the Institutional Review Board of CCHMC, and written informed consent was obtained from all patients and/or their legal guardians. SJIA was diagnosed based on the International League of Associations for Rheumatology classification criteria (3). Patients were considered to have MAS based on diagnosis of the treating physician, and where data was available satisfied 2016 MAS Classification Criteria (31). Patients were considered to have lung disease if they had objective findings on clinical exam or chest imaging. Patients were considered to have an adverse reaction to biologic therapy if it was listed in the electronic medical record as an allergy and/or led to change in therapy. Patients were enrolled and peripheral blood samples were collected during routine clinic visits. Serum IL-18 levels were determined at time of lung disease detection or initial consultation at CCHMC. Bronchoscopy with bronchoalveolar lavage and/or open lung biopsies were performed through a standardized protocol and as clinically indicated, based on current American Thoracic Society (ATS) guidelines for ILD in children (32). Residual BAL fluid and leftover lung tissue was obtained for research. Chest imaging was reviewed by two experts in pediatric interstitial and rare lung disease (AS and CT), and scored for common radiographic patterns; disagreements (<15%) were jointly adjudicated. Lung biopsies were reviewed by an expert in pulmonary pathology (KW-B) and electron micrographs by an expert in PAP (BCT). Children with SJIA-LD were matched 1:2 with control SJIA patients without clinically apparent lung disease in our registry based on current age and sex. Further details regarding BAL fluid analysis and histopathology are provided in the supplemental methods. Overall outcome was a general measure of SJIA-LD disease activity as determined by the treating physicians (GSS, JH, EMM, TT, CT, AAG).

Whole exome sequencing (WES) and sequence data analysis was performed as previously described; further details provided in supplemental methods (33).

Granulocyte-Monocyte Colony-Stimulating Factor (GM-CSF) autoantibody measurement.

Serum anti-GM-CSF antibody concentration was assessed using a GMAb enzyme-linked immunosorbent assay (GMAb ELISA) as previously described (34).

GM-CSF signaling assessed by intracellular staining for STAT5 phosphorylation.

Heparinized blood was incubated with or without 10 ng/ml of human GM-CSF Leukine for 30 min. Red blood cells were lysed and mononuclear cells were fixed with Phosflow Lyse/Fix Buffer (BD Biosciences). Cells were permeabilized, stained with Alexa Fluor®647 mouse anti-STAT5 antibody (BD Biosciences), and evaluated via flow cytometry. Monocytes were gated based on forward scatter / side scatter.

Cytokine measurements.

Cytokine concentrations in BAL fluid were determined by using Milliplex™ Multiplex kits (MilliporeSigma, Darmstadt, Germany) (35), except for IL-18 which was determined using specific ELISA kits obtained from MBL (Woburn, MA).

Gene expression analysis in lung tissue biopsies.

Total RNA was extracted from formalin-fixed, paraffin-embedded (FFPE) SJIA biopsies and control lung tissue sections using miRNeasy FFPE Kit (Qiagen). Control lung samples represented tissue procured but deemed unsuitable for transplant. Gene expression analysis was performed by Nanostring Technologies. Raw count data was obtained and analyzed with nSolver (Nanostring Technologies) and AltAnalyze to determine differentially expressed genes (fold change ≥2, FDR corrected p<0.05) and perform pathway analysis (36). Quality control analysis was performed by principal component analysis, variance vs mean normalized signal analysis, and pairwise variation during housekeeping gene selection.

RESULTS

Clinical presentation.

Since 2010, 18 children with SJIA-LD have been evaluated at CCHMC (Table S2 and Figure S1). Of these, 13 patients received initial consultations and/or evaluations (second opinions) due to concerns of SJIA-LD, and five patients were diagnosed and managed at CCHMC prior to lung disease detection. Between 2010–2018, 74 patients with SJIA received their primary rheumatologic care at CCHMC, suggesting a prevalence of SJIA-LD as high as 6.8% (5/74). At the time of initial diagnosis of SJIA-LD, most patients had only subtle respiratory symptoms such as mild resting tachypnea with normal oxygen saturation; however, 78% (14/18) did have digital clubbing (Table S2 and Table S3). One patient (SLD3) had evidence of pulmonary hypertension on echocardiogram and required sildenafil therapy, but improved over the course of treatment with resolution of these symptoms. No patients had evidence of esophageal dysmotility as determined by swallow studies where available (n=2), or autoantibody production including myositis-associated antibodies such as MDA5 (n=1). Two SJIA-LD patients had prior history of clinically and radiographically diagnosed pneumonia; however, it is unclear whether this represented infectious pneumonia or early chronic lung disease. In general, patients with SJIA-LD reported similar levels of pain, disability, and health-related quality of life as the controls. There was a trend towards more reports of fever or typical SJIA rash in the past month in the SJIA-LD as compared to the control group (Table S3).

SJIA-LD patients were followed for a median of 1 year after diagnosis (range 0.5–13 years; Table S2). All patients continued to receive biologic therapy after lung disease detection. Over this period, pulmonary disease was largely stable in 9 patients (50%), while 4 patients showed overall worsening of their disease and 5 patients overall improvement. There were no clear differences between age of onset, clinical features, radiographic or histopathological findings between patients who improved and those who did not. However, patients whose lung disease improved all achieved excellent control of their underlying SJIA based on both clinical and laboratory features.

Chest – HRCT findings.

We performed a comprehensive review of the chest CT scans from 18 patients with SJIA-LD. Of these, 16 patients demonstrated at least one of seven key features: (1) pleural thickening (n=11), (2) septal thickening (n=10), (3) bronchial wall or peribronchovascular thickening (n=11), (4) “tree-in-bud opacities” (n=2), (5) “ground glass” opacities (n=8), (6) peripheral consolidation (n=5), and (7) lymphadenopathy (n=6) (Figure 1; Table S4). Four SJIA-LD patients demonstrated at least 5/7 features, and of these lung disease worsened in one and remained stable in three patients. Two patients, one with small areas of air trapping (SLD9) and another with nodular opacities and mild bronchiectasis (SLD16), did not have any of the key six features. Both remained stable over time. Three patients had evidence of pulmonary artery enlargement on chest CT, including SLD3 noted above to have clinically significant pulmonary hypertension; the other two patients had no clinical or echocardiogram evidence of pulmonary hypertension.

Figure 1:

High-resolution Chest CT images from SJIA-LD patients. A, Focal areas of pleural thickening and bronchovascular centric tree-in-bud opacities in a mild case. B, More extensive pleural thickening with associated interlobular septal thickening in a moderately severe case. C, Ground-glass opacities with interlobular septal thickening and peripheral areas of consolidation in a severe case.

Histopathology.

Open lung biopsies were performed on eight SJIA patients suspected to have lung disease based upon ATS Guidelines, i.e. for clinical symptoms that were rapidly worsening or progressing despite current therapy (32). Notably, patients who underwent lung biopsy had significantly more HRCT features than those who did not undergo lung biopsy (median 4 vs 2, p=0.014). All eight biopsies displayed histopathology in the endogenous lipoid pneumonia/pulmonary alveolar proteinosis (ELP/PAP) spectrum (Figure 2 and Table S2 and S5). Histopathologic features included alveolar space and conducting airway luminal filling with eosinophilic proteinaceous material, foamy alveolar macrophages, and cholesterol clefts. Aggregates of cholesterol clefts associated with a granulomatous response and fibrosis previously described as pulmonary interstitial and intraalveolar cholesterol granulomas were present (37). The ELP/PAP findings were diffuse, patchy but extensive, or focal in a subpleural and peribronchovascular bundle distribution. The proportion of ELP versus PAP features varied among the patients (Figure 2F–H) but the abundance of foamy macrophages and cholesterol clefts as well as the degree of inflammation and/or fibrosis were more prominent than typically seen in autoimmune PAP (APAP, autoantibody-mediated disruption of GM-CSF signaling) or hereditary PAP (HPAP, mutations in the GM-CSF receptor chain) (26). Additionally, while PAP features were present in all lung biopsies, diagnostic PAP features were not present in the BAL with only rare or minimal eosinophilic material observed in two of four BAL fluids obtained concurrently with the biopsies (Table S5).

Figure 2:

SJIA-LD histopathogy includes ELP/PAP features, vasculopathy, chronic inflammation & fibrosis. Gross focal lesions (A,*^, **) are histologically comprised of peribronchiovascular (A,*) and subpleural (A**) air space filling with an associated inflammatory infiltrate. Disease distribution in other cases include patchy but extensive regions of lung involvement with intervening areas of normal lung (B, arrow), and diffuse involvement with marked interstitial, pleural and interlobular septal collagenous fibrosis (C, arrow; Trichrome stain). Vasculopathy includes intimal hyperplasia and fibrosis of veins (C, arrowhead) and mild arterial medial and/or intimal thickening (D, arrowhead and inset; Movat pentachrome stain) to marked occlusive mural thickening of small arteries (E, arrowhead and inset, Trichrome-VVG stain). Features in the spectrum of ELP and PAP include airspace filling with foamy macrophages (F, inset; G, arrowhead) and eosinophilic proteinaceous material (G-H,*) with a variable abundance of cholesterol clefts (F-H, arrows) including cholesterol cleft aggregates associated with a granulomatous reaction and surrounding interstitial fibrosis (F-G, arrows). The ELP/PAP features are associated with Type II alveolar epithelial cell hyperplasia (F, arrowhead) and a mild to marked lymphoplasmacytic infiltrate present in all cases. a=artery; b=bronchiole. Original magnification 20x (A), 40x (B-C), 200x (D-F), H), 400x (F insert; G). (I), Ultrastructural features of SJIA-ILD. High-resolution view of a surgical lung biopsy showing cholesterol clefts (double asterisks), collagen fibrils (arrowheads) comprising interstitial fibrosis, and an alveolar macrophage containing lipid droplets (single asterisks) and engorged with numerous intracytoplasmic lamellar structures typical of internalized surfactant (arrows). Uranyl acetate-lead citrate staining, original magnification 6000x.

Lymphocyte predominant inflammation was present in all biopsies along with evidence of ongoing and chronic injury exemplified by alveolar type II epithelial cell hyperplasia and collagenous fibrosis with lobular remodeling. Variable, nonuniform arterial wall thickening was identified in all but one case (SLD14) with a marked vasculopathy involving both arteries and veins being present in one biopsy (SLD08; Figure 2D–E). The arterial abnormalities in this case consisted of marked mural thickening of small pulmonary arteries by medial hypertrophy, intimal hyperplasia and mural fibrosis with severe luminal narrowing (Figure 2E) as well as muscularization of arterioles. Pulmonary veins also had intimal hyperplasia with fibrosis with luminal narrowing (Figure 2C). The degree of inflammation varied among the biopsies but was uniformly lymphocyte predominant. The lymphocyte population was comprised predominantly of CD3+ T-cells with fewer CD20+ B-cells (Table S5 and Figure 3). Helper CD4+ T-cells predominated over cytotoxic CD8+ T-cells in all but one case which had similar numbers of CD4+ and CD8+ T-cells (Figure 3).

Figure 3:

SJIA-LD has a lymphocyte predominant infiltrate with a predominance of CD4+ helper T-cells. Lung biopsies from two patients with SJIA_lung (A-B) with lymphocyte predominant inflammation distributed diffusely as single cells as well as aggregates, some with follicle formation (B, arrow). The inflammatory infiltrate is CD3+ T-cell predominant (C-D) admixed with fewer CD20+ B-cells (E-F). CD4+ helper T-cells (G-H) predominant over CD8+ cytotoxic T-cells (I-J). Original magnification 200x (left panel) and 100x (right panel).

Findings of an infectious process were not identified in any biopsy. No epithelioid granulomas or viral cytopathic changes were identified, and no fungal or acid fast microorganisms were identified by Grocott’s Methenamine-Silver (GMS) and acid-fast bacilli (AFB) stains, respectively. Four biopsies were cultured with growth of very few Streptococcus viridans group in one biopsy and no growth of anaerobic, acid fast or fungal organisms. BAL specimens in four patients grew only oropharyngeal flora and few Staphylococcus aureus species. No fungus, including Pneumocystis, was identified by special stains or culture except for Candida albicans being isolated from one BAL sample. Cultures were negative for Legionella species and AFB. A PCR panel for 15 viruses and Mycoplasma pneumoniae was positive only for HHV6 in two patients and EBV in one patient.

Electron microscopy examination of a lung biopsy from SLD4 revealed ultrastructural features of the lung parenchyma including fibrosis, cholesterol clefts, and alveolar macrophages engorged with lamellar structures and occasional intracytoplasmic lipid droplets (Figure 2I). Lamellar bodies are typical of surfactant phagocytosed from the alveolar surface but here were more abundant that normal. The intracytoplasmic lipid-droplets are typical of cholesterol-/cholesterol ester-rich inclusions, and could typically be seen in small numbers. However, in this patient droplets were much less numerous than in APAP/HPAP where cholesterol export is reduced due to impaired GM-CSF signaling and such droplets fill the cell (38,39).

Clinical risk factors.

Clinical risk factors present in SJIA-LD were identified by examining cases matched by age and sex to SJIA patients in our patient registry without clinically apparent lung disease. Compared to controls, SJIA-LD cases were more likely to be diagnosed at age 2 years or younger (Figure S2A and Figure S3, OR 6.5 (CI: 1.73–22.67), p=0.0070). In our cohort we also observed a striking number of adverse events to cytokine-targeted biologics exposure (Table S6). Most of these reactions were to tocilizumab, and were described variously from pain and feeling unwell, to difficulty breathing, to “anaphylaxis”. Indeed, we found a statistically significant association between adverse events to cytokine-targeted biologics, including immune-mediated reactions, and SJIA-LD cases (Figure S2B, OR 13.6 (CI: 2.35–67.42)). We also identified a statistically significant association with history of MAS when compared to controls (Figure S2C, OR 14.5 (CI: 3.35–47.21), p=0.001). Finally, we compared serum IL-18 levels as a biomarker of disease activity. Patients with SJIA-LD had significantly higher total IL-18 levels (27,612 pg/mL (IQR 10,391–62,972)) when compared to SJIA patients without lung disease (5,413 (IQR 310–31303)) (Figure S2D, p=0.047).

Other differences between SJIA-LD and primary PAP.

Due to the shared histopathologic features between SJIA-LD and primary APAP/HPAP, we explored possible overlapping pathophysiologic mechanisms between these entities. We performed WES on six patients with SJIA-LD using our previously described discovery pathway (33). However, we did not find any pathogenic variants in genes associated with primary hereditary PAP or congenital PAP. Notably, we also did not identify known pathogenic variants in hemophagocytic lymphohistiocytosis-related genes which have been linked to MAS (40), although one patient did have a novel variant of unknown significance in UNC13D (c.358G>C, p.A120P) inherited from a healthy parent. We also failed to identify high-levels of autoimmune PAP-associated neutralizing anti-GM-CSF autoantibodies in patients with SJIA-LD (Figure 4A). Finally, to assess whether patients with SJIA-LD have uncharacterized dysfunction in the cellular response to GM-CSF, fresh whole blood from patients with SJIA-LD was stimulated with GM-CSF to assess downstream signaling. As shown in Figure 4B–C, blood monocytes demonstrated normal STAT5 phosphorylation, supporting intact GM-CSF signaling.

Figure 4:

SJIA-LD patients lack characteristic features seen in PAP. A, serum titers of anti-GM-CSF antibodies detected in patients with autoimmune PAP (n=5), hereditary PAP (n=1), or SJIA-LD (n=6). B, representative flow cytometry detection of P-STAT5 in monocytes from control patient (top) and patient with SJIA-LD (bottom) before or after simulation with GM-CSF. C, dose response of P-STAT5 in monocytes from healthy controls (blue) and SJIA-LD patients (red) to GM-CSF. Data are expressed as mean fluorescence index (MFI at 10ng/ml ÷ MFI at 0 ng/ml x100. D-F, Analysis of BAL fluid obtained from patients with SJIA-LD (n=6) and hereditary or autoimmune PAP (n=7). (D) shows percent of lipid-laden macrophages detected by ORO staining, (E) shows BAL fluid cell count differential. (F) shows cytokine levels as determined by specific ELISA or luminex as described.

In order to further define the alveolar pathology seen in SJIA-LD, we compared patient BAL fluid to that from patients with APAP/HPAP. As noted above and despite histologic features of alveolar filling, BAL fluid from children with SJIA-LD did not consistently show abundant lipid-laden macrophages. While proportion of lipid-laden macrophages was variable in patients with PAP, it was significantly higher than that seen in SJIA (Figure 4D; 66.1% vs. 10.5%; P<0.001). The BAL fluid cell composition in SJIA-LD was typically macrophage predominant (Figure 4E). However, examination of cytokines and chemokines present in BAL fluid showed that children with SJIA-LD had significantly higher IL-18 levels than those in PAP (p=0.017, Figure 4F). Recent work has highlighted the role of IL-18 in SJIA and particularly in development of MAS, including through inducing IFNγ (41–43). Interestingly, we found that some children with SJIA-LD had very high BAL concentrations of the IFN-induced chemokines CXCL9 and CXCL10 (although most did not), which were not elevated in PAP patients (Figure 4F). BAL fluid was not normalized, but notably these findings do not reflect a global changes in inflammatory mediators, as levels of IL-6, IL-8, and TNF were comparable between SJIA and PAP patients (Figure 4F). Taken together, SJIA-LD has features of a novel inflammatory lung disease distinct from well characterized forms of primary PAP.

Gene expression analysis in lung tissue biopsies.

Given the findings above showing significant pulmonary inflammation in SJIA-LD and an association with MAS, we utilized the Nanostring NCounter platform to define transcriptional responses in lung tissue. nCounter has been shown to have excellent correlation between FFPE and freshly frozen tissue (44). We examined eight tissue samples from five SJIA patients; three patients had two lung regions biopsied simultaneously. SJIA lung sections were compared to three tissue control lung samples donated but deemed unsuitable for transplant. Samples were analyzed using the Human Autoimmune Panel, which targets 750 genes related to inflammation and autoimmunity. We identified 37 differentially expressed genes (fold change >2, corrected p<0.05), 36 with increased expression in SJIA-LD and one (PADI4) with decreased expression (Figure 5A and Table S7). Many of the upregulated targets are in gene pathways related to IFNγ response, including HLA-D family members and other interferon-related genes (Figure 5B). Two of the most highly upregulated non-HLA genes were CXCL10 (9-fold increase) and CXCL9 (7-fold increase), IFN-induced chemokines whose serum levels are strongly associated with MAS (45,46). Indeed, pathway analysis of the larger group of genes with uncorrected p<0.05 identified a transcription factor network including STAT1 (Figure S4). In addition, at least significantly upregulated genes (CD27, CD276, CTLA4, IL2RG, SLAMF7; FDR corrected p<0.05) were reflective of T cell activation (Figure 5B). The most significantly downregulated gene, PADI4, has been shown to modulate immune responses in lupus, and to be associated with risk for ILD in rheumatoid arthritis (47). Critically, two patient biopsy sections from radiographically and grossly unaffected areas had much less severe histopathology, with only focal areas of inflammation. However, these samples clustered with other SJIA-LD samples, and were without clear differences in inflammatory gene expression levels from affected samples, suggesting that this signature precedes and may even drive pathology (Figure 5C–D). Taken together, this gene expression analysis supports significant IFNγ-induced pulmonary inflammation in children with SJIA-LD.

Figure 5:

IFNγ response and T cell activation gene expression is upregulated in SJIA-LD lungs. (A), unsupervised clustering of samples based upon differentially expressed genes between SJIA-LD samples (blue) and controls (green) (corrected p<0.05). (B), Log relative expression of significantly upregulated genes involved in HLA-D (left), IFNγ response pathways (center) and T cells (right). (C), principle component analysis of Nanostring gene expression data. (D), H&E sections of lung samples from two simultaneously obtained lung biopsies from radiographically and grossly affected and unaffected regions from two patients used to determine SJIA-LD associated transcriptome. Original magnification 40x.

DISCUSSION

Here, we describe 18 patients with SJIA and severe, chronic lung disease. Despite mild clinical features these children had striking findings on chest CT including pleural and septal thickening, “tree-in-bud” and “ground glass” opacities, and peripheral consolidation. Histologically, this lung disease was characterized by features in the ELP/PAP spectrum but associated with lymphocyte predominant inflammation, variable vasculopathy, and evidence of ongoing and chronic injury. The patients lacked genetic, serological, or functional evidence of primary or congenital PAP. In contrast, both BAL cytokine analysis and gene expression profiling of lung tissue revealed substantial inflammation including elevated IL-18, IFNγ pathway activation and T cell function. Taken together, we describe a disorder with distinct clinical and immunologic features, representing a previously uncharacterized inflammatory lung disease (SJIA-LD).

In the past, there were rare reports of children with SJIA and chronic pulmonary manifestations, but the lung disease was relatively mild. For example, in 1980, describing the largest cohort of these patients, Athreya and colleagues noted that patients with parenchymal lung disease typically had only minimal abnormalities in pulmonary function. (24). Over the last 10 years, however, we and others observed increasing numbers of SJIA patients with severe persistent lung disease associated with high mortality (25,27) suggesting a new entity. Based upon our cohort, the prevalence of this emerging lung disease may exceed 5% of all SJIA patients.

The histopathologic pattern observed in SJIA-LD is different from the nonspecific interstitial pneumonitis that had been previously reported in SJIA and adult onset Stills disease, supporting this lung disease as a distinct entity (24,48). The lipoproteinaceous material and lipid-laden macrophages in SJIA-LD are features shared with PAP, where dysfunction of alveolar macrophages leads to accumulation of pulmonary surfactant in alveolar spaces (26). These patterns resemble that seen in biologic pre-exposed patients in a large international patient registry, which includes seven of the patients reported herein (27). We failed to identify genetic causes of primary hereditary PAP or congenital surfactant disorders, and patients lacked autoantibodies associated with primary autoimmune PAP. As such, SJIA-LD could be considered among the secondary causes of PAP. However, clinical and histopathologic features of SJIA-LD differ from PAP and support a distinct pathologic mechanism. Digital clubbing was a key presenting manifestation of SJIA-LD but is not usually seen in PAP (26). Additionally, lung biopsies in SJIA-LD had moderate to marked fibrosis with lobular remodeling in contrast to the typically well-preserved alveolar architecture characteristic of PAP (49).

In SJIA-LD, patients present with a striking dissociation between relatively subtle clinical features including mild tachypnea, dyspnea, chronic cough, and clubbing, and the severity of the inflammatory process in the lungs based on chest CT and biopsy. In addition, BAL fluid often did not contain proteinaceous material or a predominance of lipid-laden macrophages, and as such is insufficient to exclude SJIA-LD. This may also suggest that approaches such as whole lung lavage may be less beneficial in SJIA-LD. As such, there is an urgent need to define patients at risk for this lung disease as well as strategies to screen for early onset. Based on our cohort of SJIA patients, the risk factors for development of SJIA-LD include (1) age at SJIA onset <2, (2) recurrent MAS, (3) adverse reactions to tocilizumab, which may or may not be immune-mediated, and (4) prominently elevated IL-18 levels. Indeed, strikingly high serum levels of IL-18 distinguish patients at risk for MAS (41–43), and this cytokine has been recently identified as a potential therapeutic target in MAS (50). Children with this clinical pattern should be carefully monitored for signs of SJIA-LD with non-invasive testing such as six-minute walk test or pulmonary function testing with diffusing capacity in the lungs for carbon monoxide (DLCO), and should have a low threshold for obtaining HRCT with any suspicious signs or symptoms.

Our histologic and immunologic findings may provide the basis to understand the underlying mechanism of SJIA-LD. Our data suggest that although PAP-like features are indicative of macrophage dysfunction, it is not likely to be caused by intrinsic macrophage defects as in primary PAP. Alternatively, alveolar macrophage function may be altered by the external inflammatory milieu of SJIA and MAS, promoting the deviation of macrophage differentiation away from the phenotype required for surfactant recycling. Lymphocytic infiltration with T cell predominance characterizes inflammatory lesions in both SJIA-LD and MAS (45). Several translational studies have implicated IFNγ as a pivotal cytokine in MAS (46,51), and our gene expression profiling in lung biopsies from SJIA-LD patients similarly identifies a strong IFN-induced signature. It is important to note that there is significant overlap in genes induced by type I and type II interferon, and as such the relative involvement of these pathways is unclear. However, these observations suggest overlapping pathophysiologic pathways between MAS and SJIA-LD, where the inflammatory cytokine milieu that drives MAS also promotes alveolar macrophage dysfunction (Figure 6). Consistent with this hypothesis, mice with T-cell restricted overexpression of T-bet driving IFNγ production demonstrated both dysfunction of bone marrow macrophages, resulting in erythrophagocytosis and alveolar macrophage dysfunction with development of PAP-like lung pathology (52).

Figure 6:

Proposed mechanism of SJIA-LD. The cytokine storm during acute MAS, where IFNγ and IL-18 play a pivotal role, activates resident immune cells in the lungs including macrophages. These cells subsequently release chemokines which may include CXCL9–10 and initiate recruitment of Th1 cells into the lung interstitium. Accumulation of immune cells then promotes persistent and self-sustained inflammatory response. The inflammatory milieu associated with this response deviates polarization of macrophages away from the phenotype necessary for recycling surfactant leading to the emergence of PAP-like features. The role of biologic therapy and/or other environmental and genetic factors in this process remains unclear.

A critical outstanding question is the role of biologic therapy in SJIA-LD. Kimura et al noted that the emergence of chronic lung disease in SJIA coincided with the introduction of IL-1 and IL-6 blocking biologics and decreasing use of corticosteroids, suggesting potential contribution of drug exposure (25). We report a similar dramatic increase in SJIA-LD recognition, and 94% (17/18) developed lung disease after exposure to at least one biologic. However, biologic treatment has become near universal for SJIA patients (53), making such epidemiologic comparisons challenging. In addition, SJIA-LD has not been reported in patients with unrelated inflammatory diseases also treated with these biologics including hereditary periodic fever syndromes, rheumatoid arthritis, and inflammatory bowel disease, and seems to be rare in Southern Europe where the utilization of biologics is comparable to the US (DeBenedetti, Ravelli, personal communications). It is also notable that in our cohort, all patients continued to receive biologic therapy, and lung disease substantially improved in at least three patients while they continued IL-1 and IL-6 blocking biologics. Together these observations underscore the critical need for well-designed multi-center epidemiologic studies including different geographic areas with different patterns of medication utilization to define risk factors and support strategies to prevent development of this disorder.