Novel biomarkers for the assessment of paediatric systemic lupus erythematosus nephritis

Summary

The discovery of serum biomarkers specific for paediatric lupus nephritis (pLN) will facilitate the non‐invasive diagnosis, follow‐up and more appropriate use of treatment. The aim of this study was to explore the role of serum high‐mobility group box 1 (HMGB1) protein, antibodies against nucleosomes (anti‐NCS), complement factor C1q (anti‐C1q) and glomerular basement membrane (anti‐GBM) in pLN. Serum samples of 42 patients with paediatric systemic lupus erythematosus (pSLE) (22 with pLN and 20 without renal involvement), 15 patients with other autoimmune nephritis (AN) and 26 healthy controls (HCs) were examined using enzyme‐linked immunosorbent assay (ELISA). The activity of both pSLE and pLN was assessed by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) tool. The levels of all four biomarkers were significantly higher in pLN compared to AN and to HCs. The anti‐NCS, anti‐GBM and HMGB1 serum levels were significantly higher in pLN than in pSLE without renal involvement. The anti‐C1q and the HMGB1 serum levels were correlated positively with pSLE activity. The HMGB1 serum levels were also correlated positively with pLN activity. These findings suggest that serum anti‐NCS, anti‐GBM and HMGB1 may serve as biomarkers specific for the presence of nephritis in pSLE. HMGB1 emerged as a useful biomarker for the assessment of pLN and pSLE activity, whereas anti‐C1q only of pSLE activity.

Introduction

Approximately 15–20% of all systemic lupus erythematosus (SLE) cases evolve before the age of 18 years and constitute paediatric SLE (pSLE) 1. Paediatric lupus nephritis (pLN) represents the principal determinant of prognosis and mortality of the disease. It is characterized by a higher frequency, of up to 60% in pSLE, a more persistent activity pattern and a worse prognosis, compared to adult cases. As a consequence, pSLE patients receive corticosteroids and immunosuppressive drugs more often than adult patients, in order to achieve early disease remission and hinder damage accumulation 2, 3.

So far, the contemporary mainstay for the diagnosis and the monitoring of the unpredictable course of pLN is renal biopsy, an invasive procedure, which is impractical to use on a serial basis. Markers of renal function such as blood urea nitrogen, creatinine, glomerular filtration rate, the degree of proteinuria and immunological markers (antibodies against dsDNA, anti‐dsDNA, complement factors) are used currently for the assessment of LN activity, but lack sensitivity and specificity. Therefore, there is an urgent need for novel, specific, non‐invasive biomarkers for the prediction and detection of flares, the assessment of therapeutic response and the distinction between active inflammation and permanent renal damage. The main goal is the prompt, short and appropriate use of immunosuppressive treatment, in order to avoid serious side effects and excessive expense 4.

This research is focused upon the complex but blurred pathogenesis of LN. Renal inflammation is initiated by the formation of immune complexes 5. The released nucleosomes during the massive cell apoptosis, characteristic in SLE, act as inducers of the development of pathogenic autoantibodies (antibodies against nucleosomes, anti‐NCS), as well as their targets 6. The complement factor C1q, which recognizes and binds to immune complexes at the glomerular filtration membrane (GBM) and early apoptotic cells, acts itself as an inducer of autoantibody formation, namely antibodies against C1q (anti‐C1q) 7, 8, 9. Antibodies against the GBM (anti‐GBM) are induced due to the ongoing inflammation in the glomeruli, which changes the structure of the GBM, exposing or altering antigens and activating an autoimmune reaction against them 10. High mobility group box 1 (HMGB1) protein, a nuclear protein released in the extracellular space in SLE, might contribute to the inflammation in LN by binding with pathogenic anti‐dsDNA 11.

There is still a paucity of relevant data on pSLE patients, especially of non‐mixed populations. Thus, the objectives of this study were to (a) explore the association of anti‐NCS, anti‐C1q, anti‐GBM and HMGB1 serum levels with the presence of pLN and (b) correlate the levels of the above‐mentioned biomarkers with the pLN and pSLE activity, in a homogeneous Caucasian paediatric population from northern Greece.

Materials and methods

Fifty‐seven patients were studied. Twenty‐two of 57 patients had pSLE with pLN (study group), 20 had pSLE without pLN and 15 had nephritis of other autoimmune origin (the disease control groups). Twenty‐six healthy children constituted the healthy control group.

All the pSLE patients met at least four of the revised SLE classification criteria, either according to the American College of Rheumatology (ACR) in 1997 or to the Systemic Lupus International Collaboration Clinics (SLICC) in 2009, at an age younger than 18 years 12, 13. Paediatric LN was diagnosed either by renal biopsy or by fulfillment of the renal criterion of the SLE classification criteria 12, 13. The histological class of pLN was determined according to the International Society of Nephrology/Renal Pathology Society (ISN/RPS) classification of 2003 14.

The patients with nephritis of other autoimmune origin had immunoglobulin (Ig)A nephropathy (n = 5), Henoch–Schönlein purpura nephritis (n = 5), post‐infectious nephritis (n = 4) or membranous glomerulonephritis (n = 1), confirmed by biopsy at an age younger than 18 years.

All participants in this study were of Caucasian origin.

A blood sample was drawn from every participant in the study, while some pSLE patients participated with more than one blood sample, drawn at a time distance of more than 3 months.

All patients' demographic, clinical and laboratory data were recorded at the time each blood sample was drawn. The pSLE and pLN activity were assessed simultaneously at the time each blood sample was drawn using the SLEDAI–SELENA tool (Systemic Lupus Erythematosus Disease Activity Index SELENA modification, range = 0–105), which includes a section particular for nephritis 15. Patients with a SLEDAI score > 2 were considered to have active pSLE, while patients with a renal SLEDAI score ≥ 4 (range: 0–16) were considered to have active pLN 16.

All blood samples were centrifuged and the sera were stored at −80ºC. Analysis of patient and control samples was performed simultaneously and each sample was analysed twice.

Serum levels of anti‐NCS, anti‐C1q and anti‐GBM antibodies of the IgG isotype were determined using enzyme‐linked immunosorbent assay (ELISA), according to the manufacturers' instructions (Imtec, Wiesbaden, Germany). The levels of serum HMGB1 were also determined by ELISA, using commercially available kits from Amsbio (Abingdon, UK). The cut‐off values suggested by the manufacturers were 12 U/ml for the anti‐NCS, 2·5 U/ml for the anti‐GBM, 14 U/ml for the anti‐C1q and 2·5 ng/ml for the HMGB1.

The study protocol was approved by the ethics committee of the Aristotle University of Thessaloniki and informed consent was obtained from all parents or older participants according to the Declaration of Helsinki.

Statistical analysis

Results for normally distributed continuous variables were reported using mean values and standard deviation, while the median values and quartiles (Q1–Q3) were calculated for non‐normally distributed variables. The Shapiro–Wilk test was used to test the normality of continuous variables. Differences between patient groups were tested by the non‐parametric Mann–Whitney test. In order to comply with the assumptions of the test, if a patient participated in the study with more than one sample, the median value of the measurements of the samples was used. The non‐parametric Spearman's coefficient was calculated to test for correlations between two continuous variables. The positive and negative predictive values of the biomarkers for the presence of pLN were also evaluated. A P‐value < 0·05 was considered as statistically significant. The statistical analysis was conducted using the statistical package IBM SPSS statistics version 22.0.

Results

The demographic and clinical characteristics of the study participants are shown in Table 1. Renal biopsy was performed in 18 patients with pLN. The pLN classes according to the ISN/RPS classification are shown in Table 1.

Table 1.

The demographic and clinical characteristics of the study participants

	Study group	Disease control groups
	pLN	pSLE without pLN	Nephritis of other autoimmune origin	Healthy controls
Νumber of participants	22	20	15	26
Sex (F : M)	16 : 6	17 : 3	6 : 9	17 : 9
Age (years)a	15·2 (5·1)	16·4 (6)	8·5 (3·1)	14·5 (3·7)
Age at pSLE onset (years)a	11·7 (2·7)	12·4 (3·4)	–	–
Age at pLN onset (years)a	12·5 (3)	–	–	–
Νο. of serum samples	41	31	15	26
In actice pSLE	31	19	–	–
In inactive pSLE	10	12	–	–
In active pLN	24	–	–	–
In inactive pLN	17	–	–	–
pSLE activity (SLEDAI score)b	6 (2–10)	4 (2–4)	–	–
pLN activity (SLEDAI renal score)b	4 (4–8)	–	–	–
No of patients biopsied	18
Histology ISN/RPS classification
II	5
III	7
IV	3
V	3

(a) mean (standard deviation), (b) median (Q1–Q3). pSLE = paediatric systemic lupus erythematosus; pLN = paediatric lupus nephritis; ISN/RPS = International Society of Nephrology/Renal Pathology Society; SLEDAI = Systemic Lupus Erythematosus Disease Activity Index; M = male; F = female.

Serum levels of the biomarkers in the study group, the disease control groups and the healthy control group

The serum levels of the four biomarkers (anti‐NCS, anti‐GBM, anti‐C1q, HMGB1) in the study participants are shown in Table 2.

Table 2.

Serum levels of the candidate biomarkers in the study participants

	pLN	pSLE without pLN	Nephritis of other autoimmune origin	Healthy controls
Anti‐NCS (U/ml)	55·6 (39–126·5) a	32·24 (17·7–74)	12·5 (11·5–27·8)	11·25 (10·5–12)
Anti‐GBM (U/ml)	5·2 (3·05–12·51)	2·2 (2·2–2·5)	2·2 (2·2–2·4)	2·4 (2·2–2·5)
Anti‐C1q (U/ml)	33·48 (15·29–61·59)	18·7 (12·65–34·77)	12·5 (12·5–12·5)	12·5 (12·5–13·33)
HMGB1 (ng/ml)	21·03 (9·81–46·12)	8·22 (5·2–12·05)	2·5 (2·5–2·5)	0·85 (0·39–1·44)

(a) Median (Q1–Q3). anti‐NCS = antibodies against nucleosomes; anti‐GBM = antibodies against the glomerular basement membrane; anti‐C1q = antibodies against complement factor C1q; HMGB1 = high mobility group box 1; pSLE = paediatric systemic lupus erythematosus; pLN = paediatric lupus nephritis.

The serum levels of all four biomarkers were significantly higher in the patients with pLN compared to the healthy controls (P < 0·001), as well as compared to the patients with nephritis of other autoimmune origin (P < 0·001) (Fig. 1).

Figure 1.

Serum levels of the candidate biomarkers of the study participants. [Colour figure can be viewed at wileyonlinelibrary.com]

The serum levels of anti‐NCS, anti‐GBM and HMGB1 were significantly higher in the patients with pLN compared to the patients with pSLE without renal involvement (P = 0·023, P = 0·001, P = 0·003, respectively). On the contrary, there was no significant difference in the serum levels of anti‐C1q in these two groups of patients (Fig. 1).

The serum levels of anti‐NCS, anti‐C1q and HMGB1 were significantly higher in pSLE patients without renal involvement compared to healthy controls (P < 0·001, P = 0·001, P < 0·001, respectively). The serum levels of anti‐GBM did not differ significantly in these two groups of participants (Fig. 1).

Associations of the biomarkers with features of pLN

The positive and negative predictive values of the four biomarkers for the presence of pLN are shown in Table 3.

Table 3.

Positive and negative predictive values of the candidate biomarkers for the presence of paediatric lupus nephritis

Diagnostic parameter a	PPV (%) (95% CI)	NPV (%) (95% CI)
Anti‐NCS	55·9 (37·9–72·8)	100 (39·8–100)
Anti‐GBM	78·9 (54·4–94)	78·9 (54·4–94)
HMGB1	64·3 (44·1–81·4)	100 (15·8–100)
Anti‐NCS + anti‐GBM	83·3 (58·6–96·4)	100 (29·2–100)
Anti‐NCS + HMGB1	69·6 (47·1–86·8)	n.a.
Anti‐GBM + HMGB1	86·7 (59·5–98·3)	100 (2·5–100)
Anti‐NCS + anti‐GBM + HMGB1	92·9 (66·1–99·8)	n.a.

^aDiagnostic parameters were calculated for the presence of paediatric lupus nephritis against patients with paediatric systemic lupus erythematosus without paediatric lupus nephritis. PPV = positive predictive value; NPV = negative predictive value; n.a. = not applicable (because of 0 patients with negative test); CI = confidence interval; anti‐NCS = antibodies against nucleosomes; anti‐GBM = antibodies against the glomerular basement membrane; HMGB1 = high mobility group box 1.

The serum levels of HMGB1 in patients with pLN were strongly correlated positively with the pLN activity (ρ = 0·525, P = 0·002), while the serum levels of anti‐NCS, anti‐GBM and anti‐C1q were not correlated significantly with the pLN activity.

The anti‐NCS, anti‐GBM and HMGB1 levels were compared between patients with pLN of class II or III and patients with pLN of class IV or V. The differences did not reach statistical significance.

Correlations between serum levels of the biomarkers and the pSLE activity

The serum levels of anti‐C1q and HMGB1 in all the patients with pSLE, with and without pLN, showed a moderate positive correlation with the pSLE activity (ρ = 0·486, P < 0·001 for HMGB1 and ρ = 0·332, P = 0·011 for anti‐C1q), while the serum levels of anti‐NCS showed no significant correlation with the pSLE activity.

Discussion

The anti‐NCS are present in approximately 65% of pSLE patients 17. According to our findings, the anti‐NCS serum levels were significantly higher in patients with pLN compared to patients with pSLE without pLN, although their levels were not correlated with pLN activity. To our knowledge, this finding has not been reported previously in the paediatric population. On the contrary, four previous studies have not detected an association between the presence of anti‐NCS and pLN 18, 19, 20, 21. Keusseyan et al., in a study of 78 pSLE patients (56% Caucasian), 31 of whom had renal involvement, observed an association between the serum anti‐NCS and the presence of pLN, but only when the disease was active 22.

There are still controversial data regarding the association of serum anti‐NCS with the overall pSLE activity 18, 19, 21, 22, 23. Our study, in line with the study by Keusseyan et al., demonstrated no correlation between the anti‐NCS serum levels and the SLEDAI score 22.

Although a relation between high serum anti‐C1q levels and the presence of pLN, especially in its active state, has been proposed by several researchers, in the present study this relation was not confirmed 24, 25, 26, 27, 28, 29, 30. The serum anti‐C1q levels did not differ significantly between the pSLE patients with pLN and those without renal involvement. Moreover, there was no correlation between the serum anti‐C1q levels and the pLN activity. This is in line with an Italian and two Brazilian studies, which did not demonstrate any association between either the presence or titer of serum anti‐C1q and the presence of pLN 21, 31, 32. The lack of any association between anti‐C1q and the presence of pLN prevented us from performing any evaluation of its predictive value.

The present study supports the possible utility of serum anti‐C1q as a marker of pSLE overall activity and aligns with the studies by Wu et al. and Gilliam et al. 25, 27. The latter study has appraised it as a highly specific and sensitive biomarker for active pSLE, more reliable than serum anti‐dsDNA 27. Conversely, Ravelli et al. and Jesus et al. have not verified such a relation 21, 31.

The presence of anti‐GBM in animals alters the microenvironment of the glomeruli and leads to an accelerated development of nephritis, similar to LN in humans 33. Data on the role of anti‐GBM in human LN are still rare. In one adult study, the serum anti‐GBM has been related to the presence of LN and its activity 34. In the paediatric population, in a cohort of 21 patients with pSLE the serum anti‐GBM levels could distinguish the 13 pSLE patients with a history of nephritis from the 8 non‐renal pSLE patients 35. The present study has also verified this observation. Of note, the anti‐GBM serum levels in our pSLE patients with no history of kidney involvement were similar to those of the healthy controls, indicating that the anti‐GBM seems to be related to pLN development and not generally with pSLE.

The HMGB1 serum levels were significantly higher in pLN patients compared to pSLE patients with no history of kidney involvement, and were correlated positively with pLN activity. These findings confirm the possible role of HMGB1 in pLN that was indicated by a previous report in 26 pSLE Greek patients and by a recent small study from Croatia 36, 37.

HMGB1 serum levels were also correlated positively with the pSLE activity, as demonstrated previously by Kanakoudi et al. 36.

To our knowledge, this is the first study that compared the serum levels of anti‐NCS, anti‐GBM and HMGB1 between patients with pLN and patients with nephritis of other autoimmune origin, in order to assess their specificity for pLN. Indeed, we found that these biomarkers were specific for pLN and not markers of renal involvement and damage in general. Kozyro et al. have published two studies concerning the role of serum anti‐C1q in acute post‐streptococcal glomerulonephritis (APSGN) 24, 38. Nineteen of 50 children with APSGN were positive for anti‐C1q, and these patients had significantly higher proteinuria, serum creatinine, more often oliguria, hypertension and delayed resolution of the disease 38. In the present study none of the four patients with post‐infectious nephritis had high levels of serum anti‐C1q.

Limitations of this study include the relatively restricted patient population, which did not allow a multivariable data analysis. However, when considering the population of Greece and the rarity of pSLE, the number of patients can be considered appropriate. Even so, the predictive values of the biomarkers for the presence of pLN should be interpreted cautiously. The number of patients with each histological class of pLN was very small to allow reliable associations with the biomarkers. Moreover, some serum samples were drawn later than the kidney biopsy. Our Centre's policy is to perform a kidney biopsy at the time of the pLN diagnosis and not to repeat it routinely during the follow‐up period. Taking into account the fact that in pLN a change in histological pattern is common during the disease course, some of the serum samples from patients with class II or III pLN at diagnosis may actually correspond to a more severe pLN class 39. We also did not manage to retrieve paired serum samples in terms of active and inactive disease phase for all patients.

However, it is worth mentioning that this is the first study assessing the role of four novel biomarkers in a homogeneous Caucasian sample of pSLE Greek patients. Our results need to be confirmed further in similar ethnicities. It could be helpful to study these serum biomarkers in pSLE patients before the onset of pLN, as well as before a flare of pSLE or pLN, in order to validate their role as predictors of future renal activity or pSLE activity. Moreover, the investigation of a possible relation between the levels of anti‐NCS, anti‐GBM and HMGB1 and the histological activity and chronicity indices would be important, in order to elucidate if they can distinguish recent from chronic kidney lesions and to determine optimal treatment algorithms.

Disclosure

The authors have no disclosures to declare.