Background/Purpose: Children with Systemic Juvenile Idiopathic Arthritis (SJIA) who develop lung disease (LD) are at significantly increased risk of serious complications and even death. Early detection and intervention are critical for improving outcomes and stabilizing disease progression. However, diagnosing LD in children presents many unique challenges. No single pulmonary measurement can be easily applied for diagnosis and conventional methods are not always safe or feasible in young children. Krebs von den Lungen 6 (KL-6) is a promising diagnostic and prognostic blood-based biomarker of LD in adult studies, however its utility in pediatric LD is unclear. This study evaluates KL-6 as a diagnostic biomarker for LD in pediatric patients with SJIA.

Methods: Thirty-three patients enrolled in the Childhood Arthritis and Rheumatology Research Alliance (CARRA) Registry (from 18 US clinical sites) SJIA-LD cohort were included. As comparators, we included 26 patients with SJIA and no LD (SJIA-no LD) from the CARRA First-Line Options for SJIA Treatment (FROST) study, 6 children with primary protein alveolar proteinosis (PAP) as positive controls and 10 healthy children as negative controls. The primary outcome was plasma KL-6 concentration in each group as quantitated by ELISA. Secondary outcomes included clinical characteristics associated with elevated KL-6 levels.

Results: Median plasma KL-6 level in SJIA-LD was 593 U/mL (interquartile range [IQR]:268-5,896), compared to 194 U/mL (IQR:152-253) in the SJIA-no LD group (p< 0.0001) and 172 U/ml (IQR:89-245) in healthy controls (p=0.0003). No significant difference was observed between SJIA-LD and PAP patients (593 vs. 1430, p=0.258). At a cut-off of 260 U/mL by receiver operating curve (ROC), plasma KL-6 yielded a sensitivity of 79% and specificity of 81% for SJIA with LD (area under the curve = 0.83, p< 0.0001). At a cut-off of 421 U/mL, sensitivity was 63% with specificity reaching 100%. Clinical features in SJIA-LD associated with higher mean KL-6 levels included tachypnea on examination (7,627 vs. 1,719, p=0.03), hypoxemia on pulse oximeter (9,769 vs. 3,096, p=0.049), abnormal chest-Xray findings (3,177 vs. 198, p=0.04), and steroid requirement at any point (6,156 vs. 449, p=0.01). The Physician Global Assessment of Lung Disease (PGALD) was strongly correlated with mean KL-6 (r=0.75, p< 0.0001). KL-6 levels were not significantly associated with the presence of macrophage activation syndrome (p=0.66) or markers of systemic inflammation: C-reactive protein, ferritin, interleukin (IL)-18, CXCL9, or sIL2r. We were unable to evaluate the relationship between computed chest tomography (CT) abnormalities and KL-6 levels as there was an insufficient number of CT scans performed within a month of the blood sample.

Conclusion: KL-6 is a useful diagnostic biomarker of LD in pediatric SJIA; a cut-off level of 421 U/mL yielded a specificity of 100% in our cohort. Future efforts should focus on the utility of KL-6 for monitoring disease progression, evaluating treatment response, and for predicting long-term disease trajectories.

Box and Whisker plot of Plasma KL-6 Concentrations within each group – median, interquartile range (IQR), and outliers displayed. Median plasma KL-6 concentration in SJIA-LD was 593 U/mL (IQR:268-5,896); 194 U/mL (IQR:152-253) in the SJIA-no LD group (p < 0.0001) and 172 U/ml (IQR:89-245) in healthy controls (p=0.0003).

**** = p < 0.0001

ROC analysis of the plasma KL-6 concentration in systemic juvenile idiopathic arthritis-associated lung disease (SJIA-LD) with SJIA-no LD as the control. At a cut-off of 260 U/mL, plasma KL-6 yielded a sensitivity of 79% and specificity of 81% for SJIA-LD (area under the curve = 0.83, p < 0.0001). A cut-off of 421 U/mL yielded a sensitivity of 63% with specificity reaching 100%.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/krebs-von-den-lungen-6-kl-6-as-a-potential-diagnostic-biomarker-of-lung-disease-in-pediatric-systemic-juvenile-idiopathic-arthritis-preliminary-findings-from-a-multisite-us-cohort/