Lipid and immune profiles in children with systemic juvenile idiopathic arthritis: a case-control study with paired longitudinal comparison of disease activity

Abstract

Background

Systemic juvenile idiopathic arthritis (SoJIA) lacks specific biomarkers for objective disease activity assessment. Conventional inflammatory markers have limited accuracy for monitoring disease fluctuation. Few studies have integrated lipid metabolic and immune parameters to support SoJIA activity assessment, underscoring the need for simple and efficient biomarker combinations. This study aims to characterize differences in immune parameters between children with active SoJIA and healthy controls; compare the lipid profiles among children with active SoJIA, the same children in remission, and healthy controls; and explore the ability of these parameters to discriminate between disease states.

Methods

This single-center retrospective case–control study included 63 treatment-naive children newly diagnosed SoJIA (PRINTO 2018 criteria). Remission was defined by ACR/PRINTO criteria, with normal C-reactive protein (CRP) ≤10 mg/L and erythrocyte sedimentation rate (ESR) ≤20 mm/h. Forty age- and sex-matched healthy children served as controls. Blood was collected after an overnight fast. Immune, cytokine, lipid, and inflammatory markers were measured using standardized clinical laboratory procedures. Cellular immune markers (CD3⁺, CD4⁺, CD8⁺, CD16⁺CD56⁺, CD19⁺), cytokines [interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), interleukin-10 (IL-10), soluble interleukin-2 receptor (sIL-2R)], humoral immune indicators [immunoglobulin G (IgG), immunoglobulin M (IgM), complement component 3 (C3), complement component 4 (C4)], lipid parameters, CRP, and ESR were measured. The atherogenic index (AI) was calculated.

Results

Compared with healthy controls, children with active SoJIA had significantly higher levels of C3, C4, IL-6, IL-10, TNF-α, CRP, and ESR; higher proportions of CD3⁺ and CD4⁺ T cells; and lower B cells than controls (all P<0.05). Both active and remission patients had lipid abnormalities (lower HDL-C, higher TG and AI), more pronounced during active disease. Total cholesterol (TC) was elevated only in the active group (all P<0.05). Joint swelling and fever were more frequent during active disease than during remission (P<0.05). A CRP + IL-6 + AI model distinguished active SoJIA from controls with exploratory ability (73.0% sensitivity, 95.0% specificity). AI + TG distinguished active from remission disease (area under the curve =0.8650).

Conclusions

Active SoJIA shows prominent immune-inflammatory dysregulation and lipid abnormalities that partially improve with remission. The CRP + IL-6 + AI model showed exploratory ability to distinguish active SoJIA from health, whereas AI alone demonstrated exploratory potential for evaluating disease activity. Adding TG did not significantly improve performance (DeLong test, P=0.80). As only healthy controls were included without disease-mimic groups and models lacked independent validation, these findings are hypothesis-generating and cannot support clinical differential diagnosis. Prospective multicenter studies that include appropriate disease-control groups and independent validation cohorts are needed.

Highlight box.

Key findings

• Systemic juvenile idiopathic arthritis (SoJIA) children present significant immune-inflammatory dysregulation and lipid metabolism disorders, which are more pronounced during active disease.

• The combined model of C-reactive protein (CRP) + interleukin-6 (IL-6) + atherogenic index (AI) showed exploratory ability to distinguish active SoJIA from healthy children [area under the curve (AUC) =0.8361; sensitivity, 73.0%; specificity, 95.0%].

• AI alone showed exploratory ability to distinguish active disease from remission (AUC =0.8595) and warrants further evaluation as a potential marker of disease activity.

What is known and what is new?

• Children with SoJIA exhibit immune-inflammatory dysregulation and lipid abnormalities, and previous studies have linked changes in lipid profiles to disease activity.

• This study systematically evaluated a panel combining CRP, IL-6, and the AI in newly diagnosed, treatment-naive children with SoJIA and compared lipid profiles between paired active and remission states in the same cohort.

What is the implication, and what should change now?

• The combined CRP + IL-6 + AI panel provided an exploratory signal for distinguishing active SoJIA from healthy, and AI showed potential for assessing disease activity. Because no disease-control groups (e.g., infections, malignancies, or other inflammatory conditions) were included, these findings are hypothesis-generating rather than clinically validated. Prospective studies with appropriate disease controls and independent validation cohorts are are required before these biomarkers can be used in practice

Introduction

Background

Juvenile idiopathic arthritis (JIA) comprises a heterogeneous group of immune-mediated inflammatory arthritides of unknown etiology, defined by persistent synovitis as the hallmark pathological feature. It is diagnosed exclusively in children aged under 16 years and requires arthritis to persist for a minimum duration of 6 weeks to fulfill diagnostic criteria (1). Among all categories, systemic juvenile idiopathic arthritis (SoJIA; also termed systemic JIA) accounts for approximately 10–20% of all JIA cases (2). It represents the most severe subtype with the highest risk of complications, distinguished by prominent systemic inflammatory features, including quotidian fever, transient erythematous rash, hepatosplenomegaly, lymphadenopathy, and serositis. A subset of patients with SoJIA may develop life-threatening macrophage activation syndrome (MAS), a severe, potentially fatal complication driven by uncontrolled hyperinflammation (3). Because its early manifestations are nonspecific, SoJIA may be misdiagnosed or overlooked—commonly confused with infections (e.g., viral syndromes, bacterial sepsis, endocarditis), malignancies (e.g., leukemia, lymphoma, neuroblastoma), and other autoimmune or autoinflammatory disorders (e.g., systemic lupus erythematosus, Kawasaki disease, periodic fever syndromes)—resulting in delayed initiation of appropriate therapy and loss of the critical therapeutic window (4), and no standardized, readily available biomarker panel is available for dynamic disease-activity monitoring. Clinicians currently rely mainly on composite clinical scoring systems and conventional inflammatory indicators for disease assessment, while there lacks unified, readily available laboratory biomarker combinations to dynamically monitor disease activity. Therefore, identifying objective, stable, and easily measurable laboratory markers to assist in disease activity evaluation and monitoring holds significant clinical value, although their role in differential diagnosis remains to be established.

Autoimmune diseases are often accompanied by lipid metabolism disorders and immune-inflammatory dysregulation. Studies have shown that in adult patients with chronic rheumatic diseases such as rheumatoid arthritis and systemic lupus erythematosus, frequently exhibit an “atherogenic lipoprotein phenotype” characterized by decreased high-density lipoprotein cholesterol (HDL-C) and elevated triglycerides (TG), accompanied by complement activation and perturbations in multiple inflammatory cytokine levels. Moreover, these alterations are closely associated with disease activity and the risk of long-term cardiovascular events (5,6).

Rationale and knowledge gap

In children with SoJIA, previous studies have established that active disease is associated with significant lipid abnormalities, including decreased HDL-C and elevated TG, which partially reverse during remission (7,8). Active SoJIA is also associated with complement activation, with elevated C3 and C4 reflecting systemic inflammation (9). Altered T-cell subsets, particularly increased CD4⁺ T-cell proportions, have also been reported (10). A systematic review and meta-analysis by Zhao et al. (11) further confirmed that SoJIA patients exhibit the most prominent lipid metabolism disorders among all JIA subtypes, with significantly decreased HDL-C and elevated TG compared with healthy controls. However, most studies have examined lipid or immune parameters separately. Several systematic reviews and meta-analyses have summarized lipid profiles in JIA (11), and prospective cohort studies have described the association between disease activity and lipid alterations (8). To date, few studies have systematically evaluated the combined patterns of lipid and immune alterations within the same patient cohort, and even fewer have examined how these markers evolve from active disease to remission in a paired longitudinal design. The potential additive or synergistic value of combining lipid-derived indices (such as the atherogenic index, AI) with inflammatory markers [C-reactive protein (CRP), interleukin-6 (IL-6)] for disease activity assessment has not been explored in newly diagnosed, treatment-naive SoJIA patients. Nonetheless, the existing literature has several important gaps. First, most prior studies included heterogeneous JIA subtypes or patients already receiving immunosuppressive therapy, which may confound biomarker interpretation. Second, no study has specifically evaluated the combined discriminative performance of CRP, IL-6, and AI simultaneously. Third, the dynamic changes of lipid parameters between paired active and remission states in the same SoJIA patients remain incompletely characterized. These gaps limit the translation of existing knowledge into practical clinical tools for objective disease activity monitoring.

Objective

We conducted an exploratory study to address these knowledge gaps. Newly diagnosed treatment-naive children with active SoJIA were evaluated at diagnosis and again after achieving remission with standard treatment; healthy children as controls. We compared inflammatory cytokines [IL-6, tumor necrosis factor-alpha (TNF‑α), IL-10, soluble interleukin-2 receptor (sIL‑2R)], cellular immunity [CD3⁺, CD4⁺, CD8⁺, natural killer (NK) cells, B cells], and humoral immune parameters [immunoglobulin G (IgG), immunoglobulin M (IgM), complement component 3 (C3), complement component 4 (C4)] between children with active SoJIA and healthy controls. Lipid profiles [total cholesterol (TC), TG, HDL-C, AI] were compared among among healthy controls and patients during active disease and remission. Receiver operating characteristic (ROC) analysis was used to explore the ability of these parameters to distinguish active SoJIA from health and active disease from remission. Because distinguishing SoJIA from healthy children does not address the clinically relevant challenge of differentiating SoJIA from its mimics, the results are intended to generate hypotheses for studies with appropriate disease controls. The principal contributions of this study are the evaluation of the combined CRP + IL-6 + AI panel in newly diagnosed, treatment-naive patients, the paired comparison of lipid profiles between active disease and remission, and the exploratory assessment of AI as an inexpensive complement to conventional inflammatory markers. We present this article in accordance with the STROBE reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0419/rc).

Methods

Study design and patient data

We conducted this single-center retrospective case-control study at the Children’s Hospital of Nanjing Medical University. All children diagnosed with SoJIA at Children’s Hospital of Nanjing Medical University between January 2024 and December 2025 were screened for eligibility. Of 78 children with suspected or confirmed SoJIA were identified during the study period; 15 were excluded: 9 had previously received glucocorticoids or immunosuppressive therapy, 4 had incomplete clinical data, and 2 were diagnosed with other autoimmune diseases. The final cohort therefore comprised 63 consecutively enrolled, treatment-naive children with newly diagnosed SoJIA. Because this study involved retrospective data extraction, no formal screening log was maintained for patients who were not enrolled.

Inclusion criteria for the SoJIA group: (I) fulfillment of the 2018 Paediatric Rheumatology International Trials Organisation (PRINTO) classification criteria for SoJIA (12); (II) no glucocorticoid, immunosuppressive, or biologic treatment before admission; and (III) complete clinical and laboratory data.

The exclusion criteria were: (I) concomitant infection, malignancy, hematologic disease, or another autoimmune rheumatic disease that could independently affect lipid or immune profiles; (II) severe cardiac, hepatic, or renal insufficiency; and (III) incomplete clinical data. These criteria may have enriched the cohort for more typical or severe SoJIA and may therefore limit the generalizability of the findings.

The final cohort included 63 treatment-naive children with newly diagnosed SoJIA (38 males and 25 females), aged 8 months to 13 years, with a mean age of 5.10±3.29 years.

The same patients were evaluated during active disease at diagnosis (n=63) and again after remission was achieved following standard treatment (n=63). Remission was defined according to the ACR/PRINTO (13) provisional criteria for clinically inactive disease, requiring: absence of active arthritis, fever, rash, serositis, and other systemic manifestations, combined with normal CRP and erythrocyte sedimentation rate (ESR). For the purpose of this study, we operationally defined normal CRP as ≤10 mg/L and normal ESR as ≤20 mm/h, consistent with routine clinical practice at our institution. The healthy-control group comprised 40 children who underwent routine physical examinations during the same period, including 25 males and 15 females, aged 11 months to 14 years, with a mean age of 4.82±3.23 years. Controls were frequency-matched to the case group by age (±1 year) and sex, with a matching ratio of approximately 1.5:1 (63 patients: 40 controls). Medical-record review confirmed that controls had no history of autoimmune rheumatic disease, infection, malignancy, or hematologic disease and had normal physical-examination findings. Group comparability was confirmed by the Mann-Whitney U test for age (P=0.68) and chi-square test for sex (P=0.83) across the three groups. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Children’s Hospital of Nanjing Medical University (No. 202505031-1). Written informed consent was obtained from the parents or legal guardians of all participants.

Because healthy children served as controls, this study could identify biomarkers that differed between active disease and health but could not address the clinically relevant challenge of distinguishing SoJIA from infections, malignancies, or other inflammatory conditions. This limitation is considered further in the Discussion.

Laboratory measurements

Clinical and laboratory data were retrospectively extracted from the electronic medical record and the hospital Laboratory Information System (LIS). Active-stage data were obtained at diagnosis before treatment was initiated, and remission-stage data were obtained from the same patients after remission was achieved with standard treatment. Notably, lymphocyte subsets, inflammatory cytokines, and humoral immunity parameters were not routinely measured during remission, the remission analysis was limited to lipid profiles, inflammatory markers, and clinical data.

Blood samples were collected in the morning after an overnight fast of at least 8 hours. All tests were performed as part of routine clinical care by the Department of Clinical Laboratory according to established standard operating procedures (SOP). Calibrated commercial instruments were used in accordance with the manufacturers’ instructions, and intra- and inter-assay coefficients of variation were maintained within the laboratory’s quality-control limits. This study extracted only the final reported numerical values (percentages, absolute counts, or concentrations) from the LIS; we did not intervene in testing and did not have access to raw flow-cytometry files or gating plots. The assays and platforms are described below.

Lipid profiles: serum TC, TG, and HDL-C were measured using a Roche biochemical analyzer (Germany). AI was calculated as (TC-HDL-C)/HDL-C. No unified reference range for pediatric AI has been established in China; Lymphocyte subsets: CD3⁺ T cells, CD4⁺ T cells, CD8⁺ T cells, NK cells, and B cells were measured using a BD FACSCanto II flow cytometer (USA). Fresh EDTA-anticoagulated blood was analyzed within 4 hours of venipuncture according to the SOP of the hospital clinical laboratory. The routine gating strategy used by the clinical laboratory was as follows: lymphocytes were first identified by CD45 expression and side scatter (SSC), gated as CD45⁺/SSClow events. Within this gate, CD3⁺ cells were classified as T lymphocytes, CD3⁺CD4⁺ as T helper cells, CD3⁺CD8⁺ as cytotoxic T cells, CD3⁻CD19⁺ as B cells, and CD3⁻CD16⁺CD56⁺ as NK cells. Raw plots and antibody-clone information could not be retrieved retrospectively because reagent lots changed over time. The laboratory participates in national external quality assessment programs and performs daily quality controls. Inflammatory cytokines: IL-10, IL-6, TNF-α, and sIL-2R were measured using a Siemens IMULITE 1000 chemiluminescence analyzer (Germany). Serum was separated immediately and analyzed fresh without cryopreservation or repeated freeze–thaw cycles. Humoral immunity: IgG, IgM, C3, and C4 were measured using a Siemens BN II specific protein analyzer (Germany). Inflammatory markers: CRP was measured using a Pumen CRP analyzer, and ESR was measured using a VitalMicrosed automatic ESR analyzer (Italy).

Clinical characteristics, including fever, rash, serositis, and joint involvement, were extracted and compared between active disease and remission. Disease activity was defined as described in Section 2.1. Active disease was defined by any active arthritis, fever, rash, or serositis, or by CRP or ESR values above the prespecified institutional thresholds; all newly diagnosed, untreated patients met this definition. Laboratory personnel performing the index tests (lipid profiles, CRP, and IL-6) were blinded to the final diagnosis (reference standard) because the tests were conducted before clinical diagnosis was confirmed. The reference standard (PRINTO criteria) was determined by pediatric rheumatologists who were aware of the index test results, as this was a retrospective study.

Statistical analysis

Statistical analyses were performed using SPSS version 25.0. Continuous variables are reported as the mean ± standard deviation or median (interquartile range), according to their distributions. Two distinct comparisons were undertaken. The active‑versus‑remission comparison is a paired, within‑patient longitudinal analysis of the same 63 patients, whereas the active-versus-healthy comparison was an independent cross-sectional case-control analysis of 63 patients and 40 healthy controls. Statistical tests were selected accordingly. Normally distributed continuous variables were compared between two independent groups using the independent-samples t test, and non-normally distributed variables were compared using the Mann-Whitney U test. Non-normally distributed variables across three groups were compared using the Kruskal-Wallis H test. Categorical variables are reported as n (%) and were compared using the χ² test. All tests were two‑sided, and P<0.05 was considered statistically significant. ROC curves were generated for parameters that differed significantly between groups, and the area under the curve (AUC), sensitivity, and specificity were calculated. Optimal cutoffs were selected post hoc using the Youden index and should be considered exploratory. No formal sample size calculation was performed; the sample comprised all consecutively eligible patients during the study period. No key laboratory variables included in the analyses were missing. The ROC models underwent neither internal cross-validation nor external validation; their performance estimates may therefore be optimistic and should be interpreted cautiously.

Results

The index tests (CRP, IL-6, and lipid profiles) and the PRINTO reference-standard assessment were performed at the same initial visit, before treatment was initiated. No clinical interventions occurred between the two.

Baseline characteristics and immune parameters in active SoJIA and healthy controls

The baseline demographic characteristics and the age and sex comparability of the active SoJIA and healthy-control groups are described in Section 2.1.

Compared with the healthy controls, the active SoJIA group showed significantly higher levels of complement C3 (1.32±0.26 vs. 1.06±0.22 g/L, P<0.001) and C4 (0.27±0.09 vs. 0.20±0.08 g/L, P<0.001); higher proportions of CD3⁺ T cells (68.03%±8.6% vs. 59.44%±15.6%, P<0.001) and CD4⁺ T cells (39.45%±8.16% vs. 35.73%±6.81%, P=0.047); and a lower proportion of B cells (14.79%±6.35% vs. 16.46%±4.92%, P=0.048). The active SoJIA group also had higher levels of IL-6 (33.94±55.83 vs. 5.17±3.57 pg/mL, P<0.001), IL-10 (3.96±2.12 vs. 3.23±1.74 pg/mL, P<0.001), TNF-α (23.49±21.13 vs. 11.44±4.62 pg/mL, P<0.001), CRP (25.10±35.06 vs. 4.06±1.68 mg/L, P=0.006), and ESR (68.9±21.62 vs. 4.1±2.95 mm/h, P<0.001). No statistically significant differences were observed between the two groups in IgG, IgM, IgA, CD8⁺ T cells, NK cells, CD4⁺/CD8⁺ ratio, or sIL-2R levels (all P>0.05), as shown in Table 1.

Table 1. Immune parameters in children with active SoJIA and healthy controls.

Indicators	Healthy control group (n=40)	Active SoJIA group (n=63)	P value
Humoral immunity (g/L)
IgG	9.12±2.94	9.86±2.95	0.22
IgM	1.14±0.5	1.30±0.64	0.15
IgA	0.92±0.68	1.18±0.93	0.22
C3	1.06±0.22	1.32±0.26	<0.001
C4	0.2±0.08	0.27±0.09	<0.001
Lymphocyte immunity (%)
CD3⁺ T cells	59.44±15.6	68.03±8.6	<0.001
CD4⁺ T cells	35.73±6.81	39.45±8.16	0.047
CD8⁺ T cells	26.68±6.11	27.00±6.92	0.91
NK cells	12.11±5.91	13.55±6.83	0.26
B cells	16.46±4.92	14.79±6.35	0.048
CD4⁺/CD8⁺ ratio	1.44±0.53	1.41±0.57	0.67
Inflammatory cytokines
IL-10 (pg/mL)	3.23±1.74	3.96±2.12	<0.001
IL-6 (pg/mL)	5.173±3.57	33.94±55.83	<0.001
sIL-2R (U/mL)	589.9±211.40	649.9±185.90	0.22
TNF-α (pg/mL)	11.44±4.62	23.49±21.13	<0.001
Inflammatory markers
CRP (mg/L)	4.06±1.68	25.10±35.06	0.006
ESR (mm/h)	4.1±2.95	68.9±21.62	<0.001

Data are presented as mean ± standard deviation. The independent-samples t-test was used for normally distributed data, and the Mann-Whitney U test was used for non-normally distributed data. A P value <0.05 was considered statistically significant. CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; Ig, immunoglobulin; IL, interleukin; NK, natural killer; sIL-2R, soluble interleukin-2 receptor; SoJIA, systemic juvenile idiopathic arthritis; TNF-α, tumor necrosis factor-α.

Comparison of serum lipids and AI in healthy controls, active SoJIA, and remission SoJIA

HDL-C was lower in both the active and remission groups than in healthy controls (all P<0.001), and was lower during active disease than during remission (P<0.001). TG was highest during active disease, intermediate during remission, and lowest in healthy controls, with significant differences in all pairwise comparisons (all P<0.01). TC was higher during active disease than in healthy controls (P=0.009) or during remission (P=0.01), but did not differ between remission and healthy controls (P>0.99). Median AI was higher during active disease [2.98; interquartile range (IQR), 2.70–3.41] and remission (2.00, IQR, 1.73–2.39) than in the healthy controls (1.62, IQR, 1.39–1.83; both P<0.01), and was higher during active disease than during remission (P<0.001). Details are shown in Figure 1.

Figure 1.

Lipid profiles in children with active SoJIA, remission SoJIA, and healthy controls. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. AI, atherogenic index; HC, healthy control; SoJIA, systemic juvenile idiopathic arthritis; TC, total cholesterol; TG, triglyceride.

Clinical characteristics during active disease and remission

During active disease, joint swelling occurred in 68.3% of patients, fever in 85.7%, hepatosplenomegaly or lymphadenopathy in 38.1%, and rash in 50.8%. During remission, the frequency of most manifestations decreased substantially, although joint swelling persisted in 42.9% of patients. All listed manifestations except hepatosplenomegaly or lymphadenopathy differed significantly between active disease and remission (P<0.05). See Table 2 for details.

Table 2. Clinical characteristics of children with active versus remission SoJIA.

Clinical manifestations	Number of cases and incidence
	Active SoJIA stage	Remission SoJIA stage	χ2 value	P value
Joint swelling	43 (68.3)	27 (42.9)	8.23	0.004
Fever	54 (85.7)	5 (7.9)	76.53	<0.001
Hepatosplenomegaly and lymphadenopathy	24 (38.1)	15 (23.8)	3.01	0.08
Rash	32 (50.8)	9 (14.3)	19.13	<0.001

Data are presented as n (%). Intergroup comparisons were performed using the χ² test. A P value <0.05 was considered statistically significant. SoJIA, systemic juvenile idiopathic arthritis.

Selection of exploratory markers and ROC analysis for active SoJIA versus health

Because complete inflammatory and immunological profiles were available only for children with active SoJIA and healthy control, this ROC analysis was restricted to those groups. Candidate variables were selected in three steps: (I) identification of parameters that differed significantly between groups, including CRP, IL-6, and AI; (II) consideration of clinical accessibility and reproducibility, with exclusion of complex, nonroutine, or nonsignificant parameters; (III) exploratory integration of CRP, IL-6, and AI in a multivariable logistic-regression model, followed by assessment of sensitivity, specificity, AUC, and the Youden index.

For distinguishing active SoJIA from healthy controls, the AUCs for CRP, IL-6, and AI were 0.6720 [95% confidence interval (CI): 0.576–0.768], 0.7454 (95% CI: 0.657–0.834), and 0.6052 (95% CI: 0.506–0.704), respectively (Figure 2). The logistic-regression model combining CRP, IL-6, and AI yielded an AUC of 0.8361 (95% CI: 0.762–0.910). At the optimal threshold, sensitivity was 73.0%, specificity was 95.0%, and the Youden index was 0.6802. The AUC of the combined model was significantly greater than those of CRP (P<0.001), IL-6 (P=0.02), and AI (P<0.001) alone (DeLong test; Table S1). Detailed results are presented in Table 3. Because the model was developed and evaluated in the same dataset without internal or external validation, these estimates are likely optimistic. Moreover, this analysis distinguishes active SoJIA from health rather than from inflammatory, infectious, or malignant mimics; the observed differences may therefore reflect nonspecific inflammatory and metabolic changes rather than disease-specific biology.

Figure 2.

ROC curves of CRP, IL-6, and AI alone and in combination for distinguishing active SoJIA from healthy controls. combined model, the combination of the three indicators; samples with CRP values “<0.2” were excluded, while all valid samples were included for the remaining indicators. AI, atherogenic index; AUC, area under the curve; CI, confidence interval; CRP, C-reactive protein; IL-6, interleukin-6; ROC, receiver operating characteristic; SoJIA, systemic juvenile idiopathic arthritis.

Table 3. Discriminative performance of CRP, IL-6, and AI alone and in combination for distinguishing active SoJIA from healthy children.

Diagnostic Indicator	AUC	95% CI	Optimal cutoff value	Sensitivity (%)	Specificity (%)	Youden’s Index
CRP, mg/L	0.6720	0.576–0.768	9.32	52.83	100	0.5283
IL-6, pg/mL	0.7454	0.657–0.834	10.00	66.67	85	0.5167
AI	0.6052	0.506–0.704	2.008	46.03	72.50	0.1853
CRP + IL-6 + AI combination	0.8361	0.762–0.910	9.009	73.00	95.00	0.6802

The combined model was constructed by binary logistic regression, and the predicted probability was used for ROC analysis. The optimal cutoff value was determined by the maximum Youden’s index. All AUC values were statistically significant compared with the random guess (AUC =0.5000, all P<0.05). AI, atherogenic index; AUC, area under the curve; CI, confidence interval; CRP, C-reactive protein; IL-6, interleukin-6; SoJIA, systemic juvenile idiopathic arthritis.

Selection of markers and ROC analysis for active disease versus remission

Because immune and inflammatory biomarkers were not measured during remission, analyses distinguishing active disease from remission were limited to lipid parameters. AI and TG were selected because they differed significantly between disease states and are routinely available and reproducible. For distinguishing active disease from remission, TG yielded an AUC of 0.7287 (95% CI: 0.6812–0.7762), with an optimal cutoff of 1.420 mmol/L, 48.5% sensitivity, and 87.9% specificity. AI yielded an AUC of 0.8595 (95% CI: 0.8201–0.8989), with an optimal cutoff of 2.461, 84.8% sensitivity, and 78.8% specificity. The combined TG + AI model yielded an AUC of 0.8650 (95% CI: 0.8264–0.9036), with 84.8% sensitivity and 78.8% specificity (Figure 3 and Table 4). The AUCs for AI and the combined model were both greater than that for TG (P<0.001 and P<0.001, respectively), whereas the combined model did not outperform AI alone (P=0.80; DeLong test; Table S2). These unvalidated results are exploratory and apply only to the distinction between active disease and remission within this cohort. Independent validation is required before clinical use.

Figure 3.

ROC curves of TG and AI alone and in combination for distinguishing active from remission SoJIA. Combined model, TG + AI for discrimination; a total of 66 samples were included, comprising 33 active cases and 33 remission cases. AI, atherogenic index; AUC, area under the curve; ROC, receiver operating characteristic; SoJIA, systemic juvenile idiopathic arthritis; TG, triglyceride.

Table 4. Discriminative performance of TG and AI alone and in combination for distinguishing active from remission SoJIA.

Diagnostic Indicator	AUC	95% CI	Optimal cut-off	Sensitivity (%)	Specificity (%)	Youden Index
TG, mmol/L	0.7287	0.6812–0.7762	1.420	48.5	87.9	0.3636
AI	0.8595	0.8201–0.8989	2.461	84.80	78.80	0.6364
TG + AI combination	0.8650	0.8264–0.9036	0.4507	87.90	75.80	0.6368

AI, atherogenic index; AUC, area under the curve; CI, confidence interval; SoJIA, systemic juvenile idiopathic arthritis; TG, triglyceride.

Discussion

Key findings

The principal findings were as follows. First, compared with healthy controls, children with active SoJIA had lipid abnormalities (lower HDL-C, increased TG and AI) and immune‑inflammatory dysregulation, including complement activation and elevated inflammatory cytokines. Second, HDL-C, TG, and AI differed between paired active and remission states, suggesting that they may reflect changes in disease activity. Third, the CRP + IL-6 + AI panel showed exploratory ability to distinguish active SoJIA from health (AUC =0.8361; sensitivity, 73.0%; specificity, 95.0%). Fourth, AI alone distinguished active disease from remission (AUC =0.8595), and adding TG did not significantly improve performance (AUC =0.8650; DeLong test, P=0.80).

Strengths and limitations

A key strength of this study was the evaluation of active disease, paired remission measurements, and healthy controls within the same cohort, which allowed biomarker changes across disease states to be examined. In addition, AI is calculated from routine lipid measurements and may offer an inexpensive complement to conventional inflammatory markers.

Several limitations should be considered. First, and most critically, the inclusion of healthy controls alone means that the study assessed differences between active SoJIA and health, not the clinically relevant distinction between SoJIA and infections, malignancies, or other inflammatory disorders. The results therefore cannot establish differential diagnostic utility. Second, the ROC models were developed and evaluated in the same dataset without cross-validation or external validation, which increases the risk of overfitting and likely inflates performance estimates. Third, the small, single-center, retrospective cohort limits generalizability and statistical power. Fourth, immune and inflammatory parameters were not routinely measured during remission, precluding a comprehensive evaluation of immune recovery. Fifth, long-term follow-up was unavailable, so the ability of AI to predict atherosclerosis or cardiovascular events remains unknown. Finally, the institutional thresholds used to define normal CRP (≤10 mg/L) and ESR (≤20 mm/h) may not apply in other settings. Prospective multicenter studies with disease controls, larger samples, and independent validation cohorts are needed.

Comparison with similar research

Our lipid findings are consistent with previous studies (11), which showed that in children with SoJIA, HDL-C levels were significantly decreased and TG levels were significantly increased (P<0.05), whereas lipid profiles in other JIA subtypes (oligoarticular, enthesitis-related) showed no significant differences compared with healthy controls, suggesting that SoJIA is the subtype with the most prominent lipid metabolism disorders among all JIA subtypes. Breda et al. (14) reported greater carotid intima-media thickness and higher blood pressure in children with active JIA, suggesting subclinical cardiovascular alterations. Meanwhile, Bohr et al. (15) reported that children with SoJIA exhibited a higher risk of cardiovascular disease and subclinical atherosclerosis compared with other JIA subtypes, which deserves clinical attention. The association between disease activity and lipid alterations in JIA has been previously reported (8). However, to our knowledge, no previous study has specifically evaluated the combination of CRP, IL-6, and AI as a combined panel in newly diagnosed, untreated SoJIA patients, which represents the unique contribution of the present study.

Explanations of findings

SoJIA is the most heterogeneous subtype of juvenile idiopathic arthritis, with a pathogenesis distinct from other JIA subtypes. It is characterized predominantly by dysregulation of innate rather than adaptive immunity (12,16). Specifically, SoJIA is driven by persistent, excessive activation of the monocyte/macrophage system, which secretes large quantities of proinflammatory cytokines including IL-1, IL-6, IL-18, and TNF-α, culminating in a “cytokine storm” (17). These cytokines contribute to fever, evanescent rash, and serositis and may also disrupt lipoprotein-metabolizing enzymes, thereby promoting secondary dyslipidemia (18). Furthermore, complement activation and altered T-cell subsets may further amplify inflammation in SoJIA (9). Previous studies have shown that children with SoJIA often exhibit an “atherogenic lipoprotein phenotype” characterized by decreased HDL-C and elevated TG during active disease, and that these abnormalities are closely associated with an increased risk of long-term cardiovascular events (7).

In terms of immunology, children with active SoJIA showed significantly elevated complement C3 and C4 levels (C3: 1.32±0.26 vs. 1.06±0.22 g/L, P<0.0001; C4: 0.27±0.09 vs. 0.20±0.08 g/L, P<0.001), suggesting complement activation, which may be closely associated with vascular endothelial injury and systemic inflammatory response (19). Regarding Tcell subsets, the proportions of CD3⁺ T cells (68.03%±8.6% vs. 59.44%±15.6%, P<0.001) and CD4⁺ T cells (39.45%±8.16% vs. 35.73%±6.81%, P=0.047) were significantly increased, while the proportion of B cells was significantly decreased (14.79%±6.35% vs. 16.46%±4.92%, P=0.048), reflecting a predominance of Tcellmediated cellular immunity, consistent with the autoimmune pathological features of SoJIA. The levels of inflammatory cytokines IL-6 (33.94±55.83 vs. 5.17±3.57 pg/mL, P<0.001) and TNFα (23.49±21.13 vs. 11.44±4.62 pg/mL, P<0.001) were markedly elevated. IL6 is not only a core driver of systemic inflammation but also induces the production of acutephase proteins such as CRP (6), which is consistent with the significantly elevated CRP levels observed in this study (25.10±35.06 vs. 4.06±1.68 mg/L, P=0.006). Of note, these immune abnormalities are intrinsically linked to lipid metabolism disorders: inflammatory cytokines such as IL6 and TNFα can inhibit HDL-C synthesis and promote TG accumulation, leading to an “atherogenic lipoprotein phenotype” that further increases cardiovascular risk (20).

In the context of this immune-lipid interplay, this study further assessed lipid metabolism disorders and their value in discriminating disease activity. The results showed that children with SoJIA exhibited significant lipid metabolism disorders, characterized by decreased HDL-C and increased TG and AI levels, and the degree of abnormalities was highest in the active phase, intermediate in the remission phase, and lowest in healthy controls. The significant decrease in HDL-C during the active phase may be related to the suppression of HDL-C synthesis and the promotion of its catabolism by systemic inflammation (20). The elevation of TG levels is associated with decreased lipoprotein lipase activity induced by inflammatory cytokines (such as IL-6 and TNF-α) (16). As an important indicator of atherosclerosis risk, the median (interquartile range) AI levels in the healthy control group, active phase group, and remission phase group in this study were 1.62 (1.39–1.83), 2.98 (2.70–3.41), and 2.00 (1.73–2.39), respectively, with a statistically significant difference among the groups (P<0.001), indicating a more atherogenic lipid profile during active disease; however, AI alone does not demonstrate vascular injury or predict future cardiovascular events. Breda et al. reported greater carotid intima-media thickness and higher blood pressure in children with active JIA, suggesting subclinical cardiovascular alterations.

Implications and actions needed

In this single-center dataset, the combined assessment of CRP, IL-6, and AI provided an exploratory signal for distinguishing active SoJIA from healthy controls. However, its value for clinically relevant differential diagnosis remains unknown because the study did not include other inflammatory, infectious, autoimmune, or malignant conditions. Disease specificity and differential diagnostic performance must therefore be evaluated in appropriately controlled, independently validated studies.

AI alone showed promising exploratory ability to distinguish active disease from remission (AUC =0.8595). This result requires prospective validation in larger independent cohorts before clinical application. Because adding TG did not significantly improve performance, the combined model is not supported by the present data.

Future studies should (I) validating the CRP + IL‑6 + AI panel prospectively across multiple centers; (II) include infections, malignancies, other inflammatory disorders, and relevant JIA subtypes as disease controls; and (III) use long-term follow-up to determine whether AI predicts subclinical atherosclerosis or cardiovascular events in children with SoJIA.

Conclusions

In this exploratory single‑center case‑control study, active SoJIA was associated with lipid abnormalities (lower HDL-C and higher TG and AI) and immune‑inflammatory dysregulation including complement activation and elevated IL-6, compared with health. HDL-C, TG, and AI also differed between paired active and remission states, suggesting that they may reflect changes in disease activity. The CRP+IL‑6+AI panel showed exploratory ability to distinguish active SoJIA from healthy controls (AUC =0.8361; sensitivity, 73.0%; specificity, 95.0%). Its relatively high specificity suggests that a positive result could support clinical suspicion in selected settings, however, the modest sensitivity indicates that the model cannot reliably exclude SoJIA when negative. AI alone distinguished active disease from remission (AUC =0.8595), and adding TG did not significantly improve performance (AUC =0.8650, DeLong’s test P=0.80).

Because the study included healthy controls but no disease-mimic groups, the findings are hypothesis-generating and cannot establish differential diagnostic utility. In addition, the ROC models were developed and evaluated in the same dataset without internal or external validation, so their performance is likely optimistic. Prospective multicenter studies with appropriate disease controls and independent validation cohorts are required to determine whether these markers have clinical value for diagnosis or disease-activity monitoring.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Children’s Hospital of Nanjing Medical University (No. 202505031-1). Written informed consent was obtained from the parents or legal guardians of all participants.

Data Sharing Statement

Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0419/dss

tp-15-9-382-dss.pdf ^{(71.7KB, pdf)}

DOI: 10.21037/tp-2026-0419