Cross-sectional Study of Plasma Axl, Ferritin, IGFBP-4 and sTNFR2 as Biomarkers of Disease Activity in Childhood-onset SLE: A Study of the Pediatric Nephrology Research Consortium

Abstract

Objective:

To evaluate the performance of 4 plasma protein markers for detecting disease activity in childhood-onset systemic lupus erythematosus (SLE) patients

Methods:

Eighty-three consecutive pediatric patients fulfilling ≥4 ACR criteria for SLE and twenty-five healthy controls were prospectively recruited for serological testing of 4 protein markers identified by antibody-coated microarray screen, namely Axl, ferritin, IGFBP4 and sTNFR2. SLE disease activity was assessed using SLEDAI-2000 score. Fifty-seven patients had clinically active SLE (SLEDAI score ≥ 4, or having a flare).

Results:

The plasma concentrations of Axl and ferritin were significantly higher in patients with active SLE than inactive SLE. Plasma Axl levels were significantly higher in active renal versus active non-renal SLE patients. Levels of Axl, ferritin and IGFBP4 correlated significantly with SLEDAI scores. Levels of Axl, IFGBP4 and sTNFR2 inversely correlated with plasma complement C3 levels. Only plasma Axl and ferritin levels correlated with degree of proteinuria. These markers were more specific, but less sensitive, in detecting concurrent SLE activity than elevated anti-dsDNA antibody titer or decreased C3. Ferritin and IGFBP4 levels were more specific for concurrent active lupus nephritis than anti-dsDNA or C3. Plasma ferritin was the best monitor of global SLE activity, followed by C3 then Axl, while both Axl and C3 were best monitors of clinical lupus nephritis activity.

Conclusion:

In childhood-onset SLE patients, plasma ferritin and Axl perform better than traditional yardsticks in identifying disease activity, either global or renal. The performance of these plasma markers should be explored further in longitudinal cohorts of SLE patients.

Introduction:

Systemic lupus erythematosus (SLE) is a chronic multi-systemic autoimmune disorder, resulting from complex abnormalities of both innate and acquired immunity, in which renal involvement is a main determinant of poor prognosis.^{1, 2} Despite its predilection for females of reproductive age, 15–20% of SLE patients have an earlier onset. Compared to adult-onset SLE (aSLE), childhood-onset commonly behaves more aggressively at presentation and over the entire disease course. The term cSLE refers to childhood diagnosis prior to 18 years of age.³ Ethnicity plays a cardinal role in disease incidence, manifestations, and outcome in children. Hispanics and black children show a higher incidence of disease, with greater prevalence and severity of renal as well as neuropsychiatric affection.⁴

Over the past several decades, the overall 10-year survival rates of cSLE have dramatically improved, approaching 90%. Despite medical advances, cSLE continues to be a significant cause of morbidity.⁵ This is mostly due to the complex nature of childhood-onset disease, which requires more comprehensive and multidisciplinary management strategies taking into consideration the child’s growth, educational and emotional needs. Understanding the unique pathogenesis and better recognition of age-specific manifestations and long-term disease complications will hopefully improve outcome and identify more targeted, less toxic therapies.^{6, 7}

There have been significant advances in understanding SLE and lupus nephritis (LN) pathogenesis, course, and long-term outcomes in aSLE. Current conventional serological markers, such as anti-dsDNA and complement levels, cannot be considered a reliable monitor of disease activity, as reflected by SLE disease activity scores (e.g. Systemic Lupus Erythematosus Disease Activity Index, SLEDAI). Furthermore, disease flares may occur without any significant change in their levels.⁸

An ideal cSLE biomarker should show high specificity to facilitate early diagnosis, exhibit significant correlation with either disease activity or damage, and surpass conventional laboratory parameters in predicting disease flares.⁹ Most biomarker studies in cSLE have focused on selected proteins based on their biology. Few comprehensive screens for protein biomarkers have been undertaken in cSLE.

Protein microarray is a high-throughput method to assay, in different body fluids, the levels of large numbers of protein biomarkers in parallel, such as autoantibodies, cytokines, chemokines, and adhesion molecules.¹⁰ Using antibody-array-based proteomic techniques, in active aSLE patients with and without LN, a recent study documented elevation of several serum proteins, including Axl, ferritin, insulin-like growth factor binding protein-4 (IGFBP4) and soluble tumor necrosis factor receptor-2 (sTNFR2). These serum proteins were able to distinguish active LN patients and correlated significantly with renal pathology activity and chronicity indices.¹¹ In a further study, these serum proteins predicted disease activity in several aSLE cohorts, compared to healthy controls.¹² In the present study, we explore the usefulness of these four proteins in plasma as monitors of disease activity in a well-phenotyped cSLE cohort.

Materials and methods:

Patients:

Eighty-three consecutive pediatric patients (≤18 years of age) fulfilling ≥4 of the 1997 American College of Rheumatology (ACR) classification criteria for SLE ¹³ were enrolled into this cross-sectional study from the Pediatric Nephrology Research Consortium (PNRC) LN-Autoantibodies study cohort, including patients recruited from pediatric clinics at Texas Children’s Hospital (TCH), Connecticut Children’s Medical Center, and Children’s Healthcare of Atlanta. The study was approved by the institutional review boards (IRBs) at Baylor College of Medicine (H-35050), the University of Connecticut, Emory University, and the University of Houston. Full protocol for the LN-autoantibody study is available upon request. All enrolled patients completed an IRB-approved informed consent form based on good clinical practice and the Declaration of Helsinki. Demographic, clinical data and conventional measures of disease activity, including anti-dsDNA (ordinal variable, based on Crithidia titer), complement C3 and C4 levels (continuous variable, in mg/dL), spot urinary protein-to-creatinine ratio (uPCR), serum creatinine levels, and eGFR (estimated by Bedside Schwartz equation) were collected prospectively. Demographics and clinical characteristics are summarized in Table 1. As controls, healthy individuals of comparable sex and age were enrolled from the Gynecology and Adolescent Medicine Clinic at TCH.

Table 1:

Patient demographics and clinical characteristics:

Features	All SLE	HC*	L+N+	L+N−	L−N−
Number	83	25	28	29	26
Age, mean ± SD	13.6 ± 2.3	14.2 ± 2	15 ± 3	15 ± 3	16 ± 3
Females, N (%)	72 (86.7)	30 (100)	33 (92)	13 (87)	27 (82%)
Race
Hispanic, N (%)	47 (57)	14	16	17	14
African American, N (%)	23 (28)	4	7	10	6
Caucasian, N (%)	7 (8.4)	11	2	1	4
Asian, N (%)	4 (4.8)	1	1	1	2
Mixed, N (%)	2 (2.4)	0	2	0	0
SLE disease duration, median (IQR*), mos	6.8 (0.2–30)	-	0.3 (0.1–2.6)	0.7 (0.2–17)	28 (16–55)
SLEDAI*, median (IQR)	5 (0–10)	0	12 (8–22)	5 (4–11)	0 (0–3)
Historical SLE Class (%)
Neuropsychiatric	12%	0%	18%	10%	9%
Musculoskeletal	60%	0%	64%	57%	56%
Kidney disorder	61%	0%	100%	17%	58%
Mucocutanous	52%	0%	57%	58%	38%
Serositis	22%	0%	32%	14%	21%
Hematological	83%	0%	86%	72%	85%

^*HC: healthy controls; L+N+ active SLE with active nephritis; L+N- active non-renal SLE; L-N- SLE in clinical remission; SLEDAI: Systemic Lupus Erythematosus Disease Activity Index; IQR:

Interquartile range.

Assessment of SLE disease activity and flares:

Subjects enrolled into the LN-autoantibodies study were either incident patients, with biosamples collected prior to receiving immunosuppression, prevalent patients with new onset lupus flare, or prevalent patients in remission (on or off immunosuppression). At enrollment, patients were categorized into 3 groups; active renal (L+N+, based on urine dipstick and microscopic urine sediment analysis performed by a trained pediatric nephrologist), active non-renal (L+N-, symptoms or organ involvement, but renal SLEDAI=0) and inactive SLE (L-N-, asymptomatic with no findings of organ activity, subclinical hypocomplementemia and/or elevated autoantibodies allowed). A sample size of 25 subjects per group was planned to provide 80% power to detect a standard effect size of 0.95, assuming alpha of 0.0125. After enrollment, SLE disease activity was formally assessed using SLEDAI-2K, a validated tool in clinical practice and research.¹⁴ Renal activity was assessed using renal domain scores of SLEDAI (range 0–16; 0 = inactive LN). Disease damage was assessed using the Systemic Lupus International Collaborating Clinics (SLICC)/ACR Damage Index (SDI) (range 0–47; 0=no SLE damage).¹⁵

Renal Histology:

In the active renal group, the histopathologic features associated with active LN were assessed by performing a renal biopsy, evaluated by a single pediatric nephropathologist (MJH). The International Society of Nephrology/Renal Pathology Society (ISN/RPS) criteria were used to assess specific histologic manifestations of active inflammation of LN as well as features of LN chronicity or degenerative damage ¹⁶. Biopsy activity and chronicity indices (AI, CI respectively) were used for biopsy assessment according to the standards of the National Institute of Health (NIH). AI and CI scores of ≥7 and ≥4, respectively, were considered as risk factors for poor LN outcomes over long-term follow-up.^{17, 18}

Plasma biomarker assays:

Plasma samples were prepared, aliquoted and frozen at −80°C within 2 hours of blood sample collection, prior to batch processing. To avoid multiple freeze/thaw cycles, only one vial was retrieved each time for an assay. Plasma levels of Axl, ferritin, IGFBP4 and sTNFR2 were measured using a human enzyme-linked immunosorbent assay (ELISA) kits. Axl (Cat. #: DY154), sTNFR2 (Cat. #: DY726), IGFBP4 (Cat. #: DY804) were assayed using ELISA kits from R&D Systems (Minneapolis, Minnesota, USA). Ferritin was assayed using an ELISA kit from RayBiotech, Inc (Norcross, Georgia, USA) according to the manufacturer’s manual. Plasma samples were diluted 1:50 for Axl, 1:20 for ferritin, 1:50 for both IGFBP4 and sTNFR2. Optical densities at 450nm were measured using a microplate reader ELX808 (BioTek Instruments, Winooski, VT) and sample concentrations were calculated using a standard curve. All measurements were assayed in duplicate. Performers/readers of biomarker assays were blinded to clinical and reference information.

Statistical analyses:

Data analyses followed STARD reporting guidelines.¹⁹ Unless otherwise stated, ranges, means and standard deviations were calculated for interval and ordinal variables and frequencies and percentages for categorical variables. Continuous variables were expressed as mean-standard error of the mean (SEM). Kolmogorov–Smirnov and Shapiro–Wilk tests were used to test for data normality. If the data was found not to follow a normal distribution, findings were expressed as medians and interquartile range (IQR). Comparison of values among different groups of subjects was performed using the nonparametric Kruskal-Wallis H (continuous variables) or chi-square (categorical variables) tests. Pearson’s correlation coefficient was used for correlation analysis of continuous and normally distributed data. Rho values between 0.2–0.4 were considered weak; 0.4–0.6, good; 0.6–0.8, strong, and >0.8 very strong. Otherwise, the nonparametric Spearman’s correlation coefficient was used. A two-tailed P value less than 0.05 was considered statistically significant.

The diagnostic accuracy of each biomarker as well as conventional markers of SLE were assessed using receiver operating characteristic curve (ROC) analysis, and the corresponding area under the curve (AUC; range 0–1) was calculated. ROC analysis was also used to detect the sensitivity, specificity, positive and negative predictive values, and optimal cut-off values for plasma levels of Axl, ferritin, IGFBP4 and sTNFR2 as well as conventional laboratory measures. All statistical analyses were performed using GraphPad Prism v.6.0 (GraphPad, San Diego, CA, USA).

Results:

Patients’ characteristics and histologic features of active lupus nephritis subjects:

A total of 83 patients with SLE (86.7 % females) were included in this study (Table 1). The mean age was 13.6 ± 2.3 years. The median SLEDAI score of the patients was 5 ranging from 0 to 33. According to their SLEDAI assessment, 28 patients (33.7%) were categorized as active renal (group L+N+), 29 patients (34.9%) were active non-renal (group L+N-) and 26 patients (31.3%) with clinically inactive SLE (group L-N-). SLE disease damage, assessed via SLICC damage index, was evaluated, scored as 0 or 1 in all patients at the time of enrollment. 25 healthy subjects (96 % females, mean age 14.2 ± 2.01) served as controls.

All active renal and non-renal SLE patients were sampled before starting any immunosuppression. Patients were only receiving oral prednisolone or intravenous (IV) methylprednisolone. Inactive SLE patients were on low dose maintenance immunosuppression, including prednisone (59%, median dose 2.5mg/day), hydroxychloroquine (84%), mycophenolate mofetil (50%), azathioprine (25%), or methotrexate (9%). Inactive patients who received rituximab (63%) were sampled a median of 437 days after the last dose (IQR 215–716 days). Inactive patients who received IV methylprednisolone (78%) were sampled a median of 455 days after the last dose (IQR 387–716 days).

The total SLEDAI scores were significantly higher in patients with active renal disease (median 18.5; range 4–33), than those with non-renal and inactive SLE disease (Table 1).Among the active renal group (28 patients, Table 2), the median renal SLEDAI score was 10 (range 4–16). The urinary Protein: Creatinine ratio (uPCR) ranged from 0.08 – 21.5 mg/mg. Pyuria, hematuria and active urinary casts were present in 15 (53.6%), 20 (71.4%) and 12 (42.9%) patients, respectively. Renal biopsy was performed in twenty-two (78.5%) patients of the active renal disease patients. None of the patients was ISN/RPS class IV LN. ISN/RPS classes VI and V were found in 5 (22.7%) patients each, 3 (13.6%) patients showed ISN/RPS class III, while 6 (27.3%) had mixed class LN (III+V or IV+V). Patients showing ISN/ RPS class III/IV ± V were combined as Proliferative LN subgroup (N = 14), while other histological classes of nephritis (ISN/RPS I/II/pure V; N = 8) were combined as the non-proliferative LN subgroup. Histopathologic features of LN activity and chronicity were assessed concomitantly in the same biopsy (Table 2), with a median biopsy activity index of 4 (range 0–17) and chronicity index of 0 (range 0–3).

Table 2:

Clinical and histologic features of the active lupus nephritis patients:

Clinical Presentation (n=28)
- Serum creatinine, mean (SD), mg/dL	0.78 ± 0.11
- eGFR, mean (SD), ml/min/1.73m2	104.9 ± 7.2
- Protein: Creatinine ratio, mean (SD), mg/mg	3.83 ± 0.95
- Renal SLEDAI score, median (IQR)†	10 (4–16)
Histologic features (n=22)
- Activity Index, median (IQR)§	4 (0–17)
○ Endocapillary proliferation, N (%)	12 (54.5)
○ Glomerular WBC infiltration, N (%)	9 (40.9)
○ Hyaline deposits, N (%)	7 (31.8)
○ Karyorrhexis, N (%)	5 (22.7)
○ Cellular crescents, N (%)	5 (22.7)
○ Interstitial inflammation, N (%)	9 (40.9)
- Chronicity Index, median (IQR)	0 (0–3)
○ Glomerulosclerosis, N (%)	6 (27.3)
○ Fibrous crescents, N (%)	0
○ Tubular atrophy and interstitial fibrosis, N (%)	7 (31.8)

eGFR: estimated glomerular filtration rate; IQR: Interquartile range SLEDAI: Systemic Lupus

Erythematosus Disease Activity Index

^†:Range 0–16; 0 = inactive LN

^§:Range 0–24; 0 = no LN activity features

^¶:Range 0–12; 0 = no LN chronic changes

Plasma levels of Axl, Ferritin, IGFBP4 and sTNFr2:

Figure 1 shows the plasma levels of Axl, ferritin, IGFBP2 and sTNFR2 in the four groups of subjects studied. Levels of all four of the protein markers were significantly higher in patients with active renal disease than in the healthy controls. Axl and ferritin levels were significantly higher in the active renal SLE subjects than those with inactive SLE (3765 ± 235.3 vs. 2513 ± 130 pg/ml, P = 0.001) and (110.8 ± 25.5 vs. 18.3 ± 3.9 ng/ml, P = 0.0001) respectively. Only plasma Axl levels were significantly different between active renal (3765 ± 235 pg/mL) and active non-renal disease subjects (2825 ± 201 pg/ml, P = 0.04, Table 3).

Figure 1: Plasma concentrations of Axl, ferritin, IGFBP4 and sTNFR2 in active renal SLE subjects and controls.

The Y-axes show the concentrations of the 4 studied biomarkers. The X-axes display the 4 groups (28 active renal; 29 active non-renal; 26 inactive SLE and 25 healthy controls). Means and SE (error bars) are indicated. Only comparisons attaining statistical significance are indicated with P-values.

Table 3:

Plasma biomarker levels and conventional laboratory measures differentiating active lupus nephritis^*

	Active renal SLE (n=28)	Active non-renal SLE (n=29)	Inactive SLE (n=26)	Healthy controls (n=25)	P-values
Axl (pg/ml)	3765 ± 235.3	2825 ± 200.7	2513 ± 130.0	2091 ± 152.6	<0.0001 § 0.002 † 0.04 ¶
Ferritin (ng/ml)	110.8 ± 25.50	31.2 ± 3.39	18.3 ± 3.96	15.1 ± 2.70	<0.0001 § <0.0001 † >0.999 ¶
IGFBP4 (ng/ml)	180146 ± 355	140049 ± 320	108548 ± 247	157269 ± 411	0.51 § 0.21 † >0.99 ¶
sTNFR2 (pg/ml)	6966 ± 571	7195 ± 1002	5800 ± 568	6040 ± 1163	0.03 § 0.18 † >0.999 ¶
dsDNA Ab (titer)	80 (0–1280)	40 (0–160)	0 (0–17.5)	________	0.02 § 0.003 † 0.16 ¶
C3 (mg/dL)	61.7 ± 7.5	87.2 ± 5.4	113.5 ± 4.5	________	<0.0001 § <0.0001 † 0.09 ¶

^*Values were expressed as mean± standard deviation, except for dsDNA antibody titers (median, interquartile range)

^§:P-value for comparison between all groups

^†:P-value for active SLE vs. inactive SLE patients

^¶:P-value for active renal vs. active non-renal SLE patients

In the active renal disease patients with paired kidney tissue and blood samples, plasma Axl, ferritin and IGBPB4 levels were 15%, 29%, and 10% higher in the proliferative LN classes, compared to the non-proliferative LN classes, although these differences were not statistically significant. There was also a non-significant 12% decrease in plasma sTNFR2 in patients with proliferative LN classes. Serum complement C3 levels were significantly lower among the proliferative LN subgroup (37.1 ± 4.3 vs. 77.5 ± 15.1 mg/dl, P = 0.02). Similarly, patients showing proliferative LN classes had significantly higher levels of pyuria (11 vs.1, P = 0.003) and hematuria than those with non-proliferative classes (14 vs. 4, P = 0.003)

Association of plasma biomarkers with SLE disease activity and renal parameters:

Among patients of SLE (N = 83), plasma Axl showed a good correlation with the renal and total SLEDAI scores (Figure 2). Both ferritin (Rho 0.35, P 0.001) and IGFBP4 (Rho 0.26, P 0.16) showed weaker but significant correlations with SLEDAI. Correlation of our tested plasma biomarkers with conventional disease activity measures such as anti-dsDNA titer revealed weak correlations with only Axl levels (Rho 0.22, P 0.049, and −0.53, P < 0.001, respectively). There was a good correlation between urine protein-to-creatinine ratio (uPCR), the most commonly used marker for assessing renal activity, and Axl (Rho 0.46, P <0.001) as well as with ferritin (Rho 0.42, P < 0.001), after log-transformation.

Figure 2: Association of Axl with conventional and novel biomarkers in childhood-onset SLE.

The Y-axes show the concentrations of Axl (pg/mL). Pearson coefficients (r) and P-values are indicated for each marker. The X-axes display: IGFBP4, C3 complement (mg/dL), double stranded DNA antibody titer, Insulin Growth Factor Binding Protein 4 (ng/mL); SLEDAI: Systemic Lupus Erythematosus Disease Activity Index; sTNFR2: soluble Tumor Necrosis Factor Receptor 2 (pg/mL); uPCR: urine protein-to-creatinine ratio (mg/mg).

Among the patients with paired kidney tissue and blood samples, plasma biomarker levels showed only weak to no correlation (P > 0.05) with activity or chronicity index. The only statistically significant association between the plasma biomarkers and the specific components of the LN activity index was a modest inverse correlation between plasma IGFBP4 and interstitial inflammation (Rho −0.49, P < 0.05). There was weak to no correlation between the plasma biomarkers and specific components of the LN chronicity index.

There was good correlation between Axl and sTNFR2 levels (Rho 0.45) as well as weak correlations between Axl and ferritin (Rho 0.30) and IGFBP4 (Rho 0.35) respectively. There were no additional correlations between our novel plasma protein markers.

Performance of biomarkers in monitoring SLE disease activity and renal parameters:

To assess the performance of the four plasma biomarkers in discriminating active renal from active non-renal and active SLE from inactive SLE subjects in comparison with serum anti-dsDNA and low complement C3 levels, ROC analysis was performed (Table 4).

Table 4:

ROC^* analyses of lupus protein markers

	Active SLE vs. inactive SLE
Markers	AUC (95% CI)	Optimal cut-off	Sensitivity	Specificity	P-value
Axl	0.71 (0.60–0.80)	>3058 pg/ml	0.52	0.89	0.002
Ferritin	0.81 (0.70–0.90)	>27.3 ng/ml	0.62	0.88	<0.0001
IGFBP4	0.58 (0.45–0.72)	>77820 ng/ml	0.65	0.58	0.2
sTNFR2	0.60 (0.48–0.73)	>7244 pg/ml	0.40	0.89	0.13
Anti-dsDNA	0.67 (0.55–0.79)	titer ≥1:40	0.77	0.59	0.01
C3	0.79 (0.73–0.90)	<86.5 mg/dL	0.96	0.65	<0.0001
	Active renal vs. active non-renal
Markers	AUC (95% CI)	Optimal Cut-off	Sensitivity	Specificity	P-value
Axl	0.72 (0.58–0.85)	>3000 pg/ml	0.78	0.62	0.004
Ferritin	0.62 (0.47–0.77)	>94.5 ng/ml	0.41	0.99	0.007
IGFBP4	0.56 (0.41–0.72)	>275428 ng/ml	0.32	0.89	0.4
sTNFR2	0.52 (0.36–0.67)	>4761 pg/ml	0.82	0.35	0.8
Anti-dsDNA	0.55 (0.39–0.71)	titer ≥1:160	0.69	0.48	0.5
C3	0.72 (0.57–0.87)	<65 mg/dL	0.85	0.67	0.005

^*ROC = receiver operating characteristic curve; AUC = area under the curve; SLE = systemic lupus erythematosus; CI = confidence interval

Ferritin showed an excellent ability in discriminating cSLE patients with disease activity (AUC 0.81, P <0.0001), while both Axl and C3 exhibited good performance (AUC 0.71, P 0.002; AUC 0.79, P <0.0001), respectively. Again, plasma Axl and C3 showed similar performance in differentiating active renal from active non-renal SLE subjects (AUC 0.72 both). However, ferritin did not perform as well in predicting renal disease activity.

Discussion:

In this cross-sectional study of cSLE, we analyzed the performance of four candidate plasma protein biomarkers previously identified by antibody-coated microarray screening of aSLE patient plasma. With the exception of ferritin, levels of these biomarkers are not known to differ between plasma and serum sampling. We showed that plasma levels Axl and ferritin were elevated in active SLE compared to inactive SLE subjects. Plasma Axl, ferritin, and sTNFR2 were significantly higher in patients with active renal lupus than inactive patients or healthy controls. However, only plasma Axl was significantly higher in active renal than active non-renal SLE.

Plasma Axl levels correlated significantly with SLE disease activity, renal activity, urine protein levels, and conventional activity measures such as serum anti-dsDNA and complement C3 levels. Similarly, ferritin levels correlated with SLEDAI scores and proteinuria, while IGFBP4 correlated with SLEDAI scores and complement C3 levels.

Axl belongs to the Tyro3-Axl-Mer (TAM) family of tyrosine kinase receptors. The three receptors have similar structures, especially in their extracellular domain. This extracellular domain is usually cleaved off by proteases, thus releasing soluble Axl (65 kDa, similar in size to albumin), which is measured in circulation.^{12, 20} Being widely expressed on monocytes, macrophages, antigen presenting cells, vascular endothelial and smooth muscle cells, Axl plays a cardinal role in homeostatic regulation of immunity, apoptotic cell clearance, and cell adhesion: processes implicated in lupus pathogenesis. Axl is released into bloodstream in patients with inflammatory bowel disease, liver cirrhosis, hepatocellular and ovarian cancers, carcinomas associated with colitis, acute coronary syndrome, and chronic kidney disease.^21–24

Several recent studies have evaluated Axl concentrations in aSLE. Plasma Axl levels were elevated in patients with active SLE, especially active LN, and correlated with SLEDAI scores as well as other conventional activity measures.^{12, 25, 26} Our results are consistent with these studies, with significant differences in Axl levels between active and inactive SLE patients, active renal and non-renal SLE patients, and active renal versus healthy controls. In addition, in the present study of cSLE, plasma Axl levels correlated with SLEDAI, renal SLEDAI, anti-dsDNA titer, hypocomplementemia, and proteinuria. However, isolated studies have failed to reproduce this.²⁷

Ferritin (474 kDa, consisting of 24 polypeptide subunits) is an iron-binding molecule which serves as the primary reservoir of iron, in biologically available form, for numerous vital cellular processes. Ferritin helps in iron storage, recycling and detoxification, and protects proteins and lipids from the potential toxicity of iron.²⁸ Plasma ferritin levels are typically 50% of serum levels.²⁹ Due to its large size, urinary excretion is minimal in the healthy state.³⁰

Being an acute phase reactant, ferritin is often elevated in bacterial infections, chronic inflammatory conditions, and several malignancies.²⁸ In addition, high concentrations of ferritin have been observed in serum or plasma in several autoimmune disorders including SLE, rheumatoid arthritis (RA), multiple sclerosis, thyroiditis and polymyositis/dermatomyositis (PM/DM).³¹ Hyperferritinemia syndrome highlights its importance in autoimmune diseases, not only as a pivotal factor in inflammation, but also as an acute phase reactant. The syndrome occurs in Still’s disease, macrophage activation syndrome, catastrophic antiphospholipid syndrome, and septic shock.³²

Several studies have examined ferritin levels in aSLE patients of different ethnicities. Serum and urinary ferritin are both higher in active SLE compared to inactive SLE, and in active LN compared to active non-renal SLE.^33–35 Of studies comparing ferritin levels in SLE versus RA (serving as diseased controls), ferritin levels are higher in SLE patients.^32,36

Although our findings confirmed the role of ferritin as a potential marker of SLE activity, being a robust monitor of SLE disease activity (AUC 0.81) with strong correlation to SLEDAI scores, its role in detecting active renal involvement is limited in children. Plasma ferritin levels were not significantly higher in active renal SLE compared to active non-renal SLE. However, levels were significantly elevated in active renal SLE patients compared to inactive SLE and healthy controls.

Insulin-like growth factor binding proteins (IGFBPs, 23–34 kDa) are a group of transport proteins for Insulin-like growth factors (IFGs), which include six binding proteins, IGFBP-1 through IGFBP-6. They function through either IGF-dependent or IGF-independent effects. IGFBPs have a vital role in the development and progression of various cancers and serve as important diagnostic and prognostic tumor biomarkers.³⁷ However, the role of IGFBPs in systemic autoimmune disorders such as SLE remains unclear.

In one study, elevated IGFBP-4 was found in pre-pubertal children with Type-1 diabetes mellitus (T1DM).³⁸ Another study in T1DM patients showed that IGFBP4 strongly correlated with cardiovascular mortality.³⁹ In the synovial fluid of RA patients, IGFBP-4 levels were significantly higher than in osteoarthritic or healthy joints.⁴⁰ In a further validation study, plasma IGFBP-4 was superior to traditional activity markers in differentiating LN activity, as well as features of renal histopathologic chronicity.⁴¹

In children with lupus nephritis, we show an inverse correlation between plasma IGFBP-4 and interstitial inflammation, but no correlation with features of chronicity. Although our results showed IGFBP-4 levels to correlate well with SLEDAI scores and hypocomplementemia, we found no significant differences between levels in active renal, non-renal SLE, inactive SLE, or healthy controls. In addition, the performance of this biomarker in discriminating SLE and LN activity were close to that of conventional activity measures.

TNF-α, a potent pro-inflammatory cytokine, is known to mediate its actions via binding to its two receptors (TNFR1 and TNFR2). The two receptors are expressed on most cell surfaces and their extracellular portion is commonly shed into circulation. The soluble forms sTNFR1 (55 kDa) and sTNFR2 (42 kDa) are readily measured in most body fluids. Being able to compete with the cell-attached TNF receptors for TNF-α, sTNFRs are suggested to have the ability to act as inhibitors of TNF activity or as carriers of the cytokine, prolonging its biological effect.^42–45

In the past decade, the development of anti-TNF therapy has been a milestone in the treatment of RA and spondyloarthropathies, but their use is reported to be associated with new or aggravated forms of autoimmunity, making its use in SLE controversial. There is less certainty about the precise role of TNF-α in SLE. However, several studies have documented the presence of significantly elevated levels of TNF-α and sTNFR2 in SLE patients and reported their correlation to SLE disease activity, lupus nephritis activity, as well as histological changes on renal biopsy.^{11, 42, 46, 47} Furthermore, in a longitudinal cohort, sTNFR2 exhibited superior concordance with current disease flare, compared to conventional disease measures.¹¹

Our study showed sTNFR2 levels to be significantly elevated in active renal cSLE compared to healthy controls; however, no difference was noted between active renal and non-renal SLE. Levels of sTNFR2 inversely correlated with complement C3 levels, but did not correlate with SLEDAI scores. Regarding the ability to distinguish active SLE from inactive SLE and active renal from active non-renal SLE, it performed no better than conventional activity measures (anti-dsDNA and C3).

There are limitations to our study. Due to the challenges of diagnosing SLE in children, and the urgency for treating symptoms at initial presentation before diagnosis can be formalized, the protocol allowed for steroid administration prior to sample collection. However, incident subjects had plasma collected prior to all immunosuppression, which limits the confounding effects of medications on biomarker expression seen in many other published cohorts. Moreover, paired serum samples were not collected to allow direct comparisons with plasma levels. Finally, our cross sectional design does not allow for interrogation of biomarkers for prognosis, or for capturing changes in disease activity over time.

To our best knowledge, this is the first study to examine the role of plasma Axl, IGFBP4 and sTNFR2 in a cSLE cohort. Of the markers examined, plasma Axl appears to have great potential as a biomarker of SLE and lupus nephritis activity, as it is more specific than anti-dsDNA for concurrent disease activity. Serum ferritin has been examined before in several cSLE cohorts, supporting our finding that plasma level is an excellent marker of disease activity in a patient already diagnosed to have lupus. On the other hand, it cannot be used for making the initial diagnosis of SLE as it is elevated in other diseases. Further longitudinal studies are needed to validate the performance of these plasma proteins as predictors of disease flares compared to traditional markers, and to ascertain if combining these plasma protein markers with anti-dsDNA and complement levels would result in better sensitivity and specificity profiles in predicting early relapse of SLE.