Introduction

Systemic lupus erythematosus (SLE) is a challenging autoimmune disease, which can affect any organ of the body and display a broad spectrum of clinical and immunological manifestations. Childhood-onset SLE has an unpredictable natural history, with epidemiological studies demonstrating the progressive clinical course of SLE and its significant morbidity and mortality rates.

Using the diagnostic criteria of the American College of Rheumatology [1, 2], approximately 10%–17% of proven cases present in childhood with more severe organ involvement than in adults (especially with respect to haematological and renal disease). The outcome of childhood SLE has improved during the last decade, but the morbidity remains high [3]. The renal manifestations of disease activity are the major determinant of the long-term outcome of lupus.

In various epidemiological studies, both at the time of diagnosis and during the clinical course, there are renal manifestations of disease activity in between 20% and 80% of paediatric patients [4, 5, 6]. In prospective studies of childhood-onset lupus, the prognosis is most closely related to the severity of renal disease [3]. Clinically evident nephritis is not only more frequent, but is of greater severity in children than in adults [7].

The classification of lupus nephritis is based on histological patterns of renal biopsies. However, renal biopsies are usually only performed when there is clinical evidence of lupus nephritis with significant proteinuria, haematuria, hypertension or deteriorating renal function [as measured by plasma creatinine and/or glomerular filtration rate (GFR)]. The classification of the severity of lupus nephritis was developed by the World Health Organisation (WHO) [8] with the risk of significant long-term renal impairment relating to the type and extent of renal lesion.

Different classes of disease activity reflect different histological findings in patients’ biopsies with progression over time (especially WHO class III developing into WHO class IV lupus nephritis): WHO class I (normal), II (mesangial), III (focal proliferative lupus nephritis), IV (diffuse proliferative lupus nephritis or DPLN), V (membranous) or VI (end-stage renal disease). There can also be overlap between classes, for example with membranous changes in WHO class IV lupus nephritis. However, these classifications are based on morphological and histological criteria from percutaneous renal biopsies with no evaluation of extraglomerular disease, such as tubular, interstitial or vascular disease. The degree of tubulointerstitial involvement (including mononuclear cell infiltration, interstitial fibrosis and tubular atrophy) as well as the severity of glomerular lesions, both with respect to activity and chronicity indices, must be considered simultaneously to predict renal function of patients with lupus nephritis. Activity indices include intracapillary proliferation, leucocyte exudation, karyorrhexis and fibrinoid necrosis, cellular crescents, wire loops and “hyaline” thrombi and interstitial mononuclear inflammation, whereas chronicity indices include glomerular sclerosis, fibrous crescents, tubular atrophy and interstitial fibrosis.

The role of renal biopsy in the evaluation of lupus nephritis in childhood-onset SLE remains controversial [9]. Historically, the principal candidates for renal biopsies have been children with clinical or laboratory evidence of renal involvement with glomerular dysfunction, in order to identify DPLN, which is the most severe subtype. DPLN adversely affects the prognosis and may influence patient management by requiring an increase in immunotherapy [10]. However, extensive glomerular abnormalities have been identified on biopsies of patients who have no clinical evidence of renal disease [11, 12, 13, 14, 15]. Renal biopsy can also differentiate between proteinuria caused by renal scarring or active disease.

Tubular function is not routinely assessed in lupus patients, despite the fact that SLE may be associated with a variety of tubular defects [16]. However, tubular interstitial disease has a serious bearing on renal outcome and is an important factor in chronicity indices. The presence of proximal renal tubular dysfunction as measured by urinary retinol-binding protein (RBP) is clinically useful for differentiating adult patients with active lupus nephritis from those with stable or absent renal disease [17]. Tubular defects may provide an earlier and more effective indication of renal disease than glomerular defects and disease activity scoring systems. Investigations of tubular function may reveal earlier evidence of renal involvement and have implications for immunosuppressive and other treatment modalities.

There are children with SLE in whom lupus nephritis is not suspected due to the absence of haematuria (microscopic and macroscopic), proteinuria, nephrotic syndrome, hypertension and renal dysfunction but who may have tubular involvement. These patients have not warranted percutaneous renal biopsies.

This observational comparison study of renal tubular and glomerular function in patients with lupus nephritis and those without evidence of renal disease was undertaken in order to assess the prevalence and importance of tubular dysfunction in childhood lupus patients. A comparison was made between the findings and with tubular function of normal children. The relationship between tubular function and glomerular function and the correlation between different markers of tubular function was assessed.

Subjects and methods

This was a cohort study of paediatric and adolescent patients with SLE reviewed at the tertiary referral clinics of the Departments of Nephrology and/or Rheumatology at Great Ormond Street Hospital for Children NHS Trust. Due to the nature of tertiary referrals, these patients tended to be at the severe end of the spectrum of childhood disease activity. All the patients were seen at additional outpatient appointments in a joint nephrology and rheumatology clinic specifically set up for lupus patients involved in this study. The subjects of the study were identified by a search of the Patient Information Management Systems (PIMS) of Great Ormond Street Hospital for Children NHS Trust and the Nephrology Department’s Patient Database.

Of 21 children with SLE, 10 had lupus nephritis (clinical and biopsy proven) and 11 had no evidence of nephritis and were designated “lupus non-nephritis” (no clinical or laboratory evidence of renal involvement at the time of analysis). The controls consisted of 94 normal children who were of similar ages [18]. Only urine from these patients was analysed for tubular function for ethical reasons. The controls provide valuable information concerning the normal ranges of tubular function in childhood. Controls were excluded if autoimmune disease, renal pathology or infection was diagnosed or clinically suspected. The literature has well-established data on tubular function in normal children [19] and those with nephrotic syndrome [18, 20] and insulin-dependent diabetes mellitus [21].

Full assessment with history and examination was undertaken for all patients. Blood tests included investigative markers of disease activity [with erythrocyte sedimentation rate (ESR), C-reactive protein, anti-nuclear antibodies, anti-double-stranded DNA antibodies, C3, C4, and anti-cardiolipin IgG and IgM antibodies]. Other investigations included serum electrolytes, creatinine, and full blood count. Calculations were made of the estimated GFR from serum creatinine (Cr) levels, although some patients had had measured GFR.

Urine investigations included urinalysis and quantification of proteinuria with urine albumin/Cr ratios. The markers of tubular function were analysed: namely urinary RBP, urinary NAG (N -acetyl-β- d -glucosaminidase) and TRP (tubular reabsorption of inorganic phosphate); and urine amino acids were estimated in all cases. Urine specimens taken from the controls were only analysed for RBP and NAG [18] (as this was the initial control protocol and they would have required blood to be taken for phosphate levels to calculate the TRP). All the urine samples were from the second episode of micturition during the day in order to avoid increased biochemical values found in early morning urine samples [19]. The NAG and RBP data were expressed as a ratio over urinary Cr in units of μmoles 2-methoxy-4-(2’nitrovinyl)-phenol (MNP)/h per millimole and micrograms per millimole, respectively.

The data were logarithmically transformed as control RBP and NAG levels are known to be logarithmically distributed [18]. The methods of analysis of patients for both urinary RBP levels and urinary NAG levels were identical to those previously published [18]. The urinary RBP levels were analysed by enzyme-linked immunosorbent assay using rabbit antisera (Dako, High Wycombe, UK) and urinary NAG levels with a commercially available kit (Praill Price Richardson Diagnostics, London, UK) on an IL Monarch centrifugal analyser (Instrumentation Laboratory, Cheshire, UK).

The data were analysed on SPSS 11.0 software for Windows. Statistical significance was derived using non-parametric tests such as Mann-Whitney U, Wilcoxon and Kruskal-Wallis tests. To quantify the relationships between the continuous variables, a column scatterplot was first obtained and then calculation of the correlation either by Spearman’s rank or Pearson correlation methods was performed. Ethical approval was obtained for this study from the Institute of Child Health and Great Ormond Street Hospital for Children NHS Trust Research Ethics Committee.

Results

Baseline characteristics

Baseline characteristics of the groups with similarities in ages and sex ratios are shown in Table 1. Due to the expected severity of disease activity in the lupus nephritis group, patients in this group had a statistically significant increase in the number of admissions over a similar time interval since diagnosis (P <0.001) and an increased relapse rate (P =0.03). Using the eight-system involvement of the British Isles Lupus Assessment Group (BILAG) disease activity index [22], there were 8 lupus nephritis patient systems with the most-severe organ involvement (category A) compared with 3 in the lupus non-nephritis group; this was statistically significant with the renal BILAG scores (P =0.013), although not with the total BILAG index (P =0.20). All lupus nephritis patients had evidence of active disease at the time of biopsy at 0.1–1.8 (median of 0.8) years prior to assessment. The immunosuppressive and non-steroidal anti-inflammatory drug usage at the time of analysis is shown in Table 2.

Table 1 Baseline characteristics of controls and lupus patient groups
Full size table
Table 2 Immunosuppressive and non-steroidal anti-inflammatory drug (NSAID) use in lupus patient groups
Full size table

Laboratory indices

The laboratory indices (Table 3) showed no statistically significant differences between the groups. The lupus non-nephritis patients were more likely to have higher anti-cardiolipin IgG antibodies, suggesting an increased thrombotic propensity, although this was not statistically significant. There were no significant differences in ESR, anti-double-stranded DNA antibodies, C3 and C4 levels (P values of 0.4, 0.8, 0.8 and 0.6, respectively). The elevated C-reactive protein of 17.1 g/dl in 1 lupus non-nephritis patient was due to a buttock abscess, which required admission to hospital for intravenous antibiotics.

Table 3 Analysis of the investigative disease activity in lupus patient groups (ESR erythrocyte sedimentation rate, CRP C-reactive protein, ANA anti-nuclear antibodies)
Full size table

Glomerular function

As expected, the lupus nephritis group had statistically significant evidence of more severe renal impairment by measurement of GFR (P =0.07) with 51Cr EDTA. Only 12 of 21 (57%) lupus patients had undergone a formal, measured GFR, so estimated GFR using the standard Schwartz formula [23] was used both at times of formal GFR measurement and at outpatient review. The results apply to all patients unless otherwise stated in Table 4. In some patients with chronic renal failure, there were clinically significant differences between the measured and estimated GFR, as shown by the differences in the median GFR in the lupus nephritis group (although Spearman’s rank correlation coefficient r was 0.825). Patients’ plasma Cr levels were also recorded for the study.

Table 4 Analysis of the glomerular function in lupus patient groups (GFR glomerular filtration rate)
Full size table

The lupus non-nephritis patient who had evidence of renal impairment with a measured GFR of 69 ml/min per 1.73 m2 had a serum Cr of 56 μmol/l, no evidence of proteinuria on urinalysis and a normal urinary albumin/Cr ratio of 0.02 mg/mg.

Patients had regular urine tests for urinary albumin/Cr ratios and the increased albuminuria of the lupus nephritis group was statistically significant (P <0.001), with normal levels of 0–0.1 mg/mg. Glomerular function assessed by urinary albumin/Cr ratio was a good discriminator between those with lupus nephritis and those without. All those with lupus nephritis had an elevated urinary albumin/Cr ratio. Only 1 lupus non-nephritis patient had an elevated urinary albumin/Cr ratio that appeared to be linked to menstruation.

Tubular function

Table 5 shows a comparison between the tubular markers in the two patient groups (lupus nephritis and lupus non-nephritis) and the normal controls. Figs. 1, 2 and 3, respectively, demonstrate the data on scatterplots of log NAG/Cr, log RBP/Cr and log NAG/Cr against log RBP/Cr in the two patient groups and the controls.

Table 5 Analysis of the tubular function in lupus patient groups (TRP tubular reabsorption of phosphate)
Full size table
Fig. 1
figure 1

Column scatterplot of log N -acetyl-β- d -glucosaminidase/creatinine (NAG/Cr) in normal and lupus patients groups

Full size image
Fig. 2
figure 2

Column scatterplot of log retinol-binding protein/creatinine (RBP/Cr) in normal and lupus patients groups

Full size image
Fig. 3
figure 3

Scatterplot demonstrating the relationship between log NAG/Cr and log RBP/Cr in normal and lupus patients groups

Full size image

Abnormalities of tubular function were present in 2 patients in the lupus non-nephritis group who had elevated urinary NAG levels. This evidence of tubular dysfunction in the face of no overt evidence of glomerular disease necessitates follow-up. The tubular function was not related to the use of non-steroidal anti-inflammatory drugs (NSAIDs), which are commonly prescribed for children with joint involvement in SLE, but are not advocated in children with evidence of renal involvement due to the side-effect of tubular damage. There were only 2 patients in each lupus group who were taking NSAIDs at the time of urine collection (Table 2). Of the 2 lupus non-nephritis patients who were taking NSAIDs, 1 had the lowest urinary NAG level (of all 21 patients) and the other had the lowest urinary RBP level. Although NSAIDs may influence renal tubular function, this was not marked in our patient population.

Of the lupus nephritis group, 60% had elevated urinary NAG levels, although the remaining 4 patients had levels at the upper end of the normal range. The urinary RBP was less discriminatory between the groups due to the very wide normal range, with only 2 lupus nephritis (and no lupus non-nephritis) patients with elevated levels. However, the 2 lupus non-nephritis patients with the highest urinary RBP levels later developed lupus nephritis.

Comparing the values of the tubular markers between the three groups (normal controls, lupus non-nephritis and lupus nephritis) using the Kruskal-Wallis test (a non-parametric test of one-way analysis of variance) shows that the differences are statistically significant. The chi-squared results for urinary RBP (log RBP) and urinary NAG (log NAG) were 16.45 (36.75) and 24.80 (51.39) with two degrees of freedom; so the P value for all variables was less than 0.0005.

The relationship between log NAG/Cr and log RBP/Cr in normal and lupus patient groups in Fig. 3 demonstrates that the differentiation between normal controls and lupus nephritis patients is greater with urinary NAG than RBP levels.

Analysing the data statistically, using Mann-Whitney non-parametric tests, gave statistically significant differences between the lupus non-nephritis and lupus nephritis groups, with a Z score of 2.18 (giving a two-sided P value of less than 0.05) for the log RBP and 3.10 (P <0.002) for the log NAG.

There was a statistically significant increase in the number of elevated urinary amino acids in the lupus nephritis group (P =0.03). The urinary TRP was not a good discriminator as the results were normal (with laboratory normal reference range above 80%) in all but 1 of the 21 patients (who was a patient with lupus nephritis and end-stage renal failure) with a Z score for urinary TRP of 0.67 (P =0.6).

The results of the urinary RBP and NAG in the control group showed a wide range of distribution as previously reported [18]. There were 5 of the 94 controls whose urinary RBP/Cr was greater than 12 μg/mmol. Although this could be considered within the normal range, it is not clear whether, in spite of attempts to eliminate confounding factors, some controls may have had minor viral infections and/or were taking medications, which may affect tubular function. However, the urinary NAG concentrations had a much tighter normal range.

There were 2 lupus nephritis patients whose urinary RBP was greater than the others within the lupus nephritis group and 2 patients within the lupus non-nephritis group whose urinary NAG was much greater than the rest of their group, both of whom had normal glomerular function.

Discussion

Some controversy exists on the various methods of assessing disease severity in children with lupus. Some believe that clinical assessment should take precedence over laboratory investigation. However, it is recognised that laboratory investigation may suggest a flare in disease activity, although the patient is well clinically, whereas at other times clinically the disease might appear active without commensurate laboratory findings.

The glomerular pathology in SLE has been considered the main determinant of outcome in lupus nephritis, although evidence is emerging of the major role of tubulointerstitial disease in prognosis [24, 25, 26, 27, 28, 29, 30, 31]. There is some evidence indicating that proximal renal tubular dysfunction as measured by urinary RBP may be clinically useful to differentiate patients with active lupus nephritis from those with stable or absent renal disease [17]. This investigation may have a role in the early diagnosis and subsequent monitoring of renal disease in those patients with subclinical nephropathy [32].

Tubular function is not routinely assessed in lupus patients, despite the fact that SLE may be associated with a variety of tubular defects [16]. Studies of tubular function in adult-onset lupus patients have identified defects in the handling of potassium, sodium and hydrogen ions with distal renal tubular acidosis (due to an isolated proton secretory defect, gradient or acid back-leak type and responsive or unresponsive voltage-dependent distal renal tubular acidosis) [16]. There has been some work showing that tubular dysfunction occurs during active proliferative glomerulonephritis in lupus and can be influenced by immunosuppressive treatment. A prospective study has shown no correlation between the different tubular function studies and the activity index of tubulointerstitial abnormalities in the renal biopsy. This suggests that tubular function studies are probably a more sensitive indicator of tubulointerstitial disease [33].

There has been published work demonstrating that tubular markers define renal disease earlier than albuminuria. There is evidence that some adult patients with SLE have tubulopathies during exacerbations of disease activity [34]. The tubulopathy may be secondary to iatrogenic factors resulting from the use of NSAIDs such as ibuprofen or diclofenac sodium [35]. Acrolein, a urinary metabolite of cyclophosphamide, is known to cause haemorrhagic cystitis that might ultimately lead to bladder wall fibrosis, but it remains unclear whether tubulointerstitial damage can result. Occasional anecdotal cases suggest that this might occur and hence could also be a factor [36].

One of the questions that this study posed was whether patients with abnormal tubular markers in childhood lupus have active disease as opposed to renal scarring. It is known that disease activity may be manifest predominantly in one or more organs, as opposed to being equally active systemically. The activity within the kidney may be part of a systemic flare (which is often the case), but might ultimately cause subacute damage, with possibly the tubulointerstitium being the target organ affected. This might cause different abnormalities (such as proximal or distal tubular dysfunction) in different patients.

There is evidence that tubulointerstitial disease may not only be present but may carry a worse prognosis for the long-term outcome of renal parenchymal disease [24, 37, 38, 39, 40, 41]. The final common pathway is fibrosis in the interstitium and, as with other renal parenchymal diseases, much effort is being made to minimise this with drugs such as angiotensin-converting enzyme inhibitors [42].

Figure 1 shows that the log NAG/Cr ratio is significantly higher in the lupus nephritis group [compared with both the control group (P =0.001) and the lupus non-nephritis group (P =0.001)]. There were 2 patients within the lupus non-nephritis group whose urinary NAG was much greater than the rest of their group. The question is what should be done about them. They both had normal urinary albumin/Cr ratios, but may have the potential to develop overt nephropathy. It would seem prudent that they should have careful long-term follow-up with close monitoring. In particular, these patients may warrant further investigation, including percutaneous renal biopsy and perhaps further treatment with increased immunosuppression.

Figure 2 shows that the log RBP/Cr ratio is significantly higher in the lupus nephritis group [compared with both the control group (P =0.001) and the lupus non-nephritis group (P =0.03)]. The 2 lupus non-nephritis patients with the highest urinary RBP levels (although still within the normal range of the controls) both later went on to develop biopsy-proven lupus nephritis (after they subsequently developed significant proteinuria).

The finding of tubular dysfunction may provide further information for patients with SLE, although the differences between normal controls and lupus nephritis patients are more noticeable with urinary NAG than RBP levels (Fig. 3). However, due to the wide normal ranges of results (more pronounced with urinary RBP levels) and overlap of groups, there may be some limitations in the clinical use of urinary NAG and RBP levels.

It is interesting that there was a correlation between the urinary albumin/Cr ratios and the urinary RBP and NAG in the 4 lupus nephritis patients who had the lowest, but still abnormal, urinary albumin/Cr ratios. They also had lower log RBP and log NAG than the other lupus nephritis patients. Serial measurements of tubular function in these patients may reveal that the specimens were taken during periods of quiescence, although determining disease quiescence, mild tubulointerstitial disease and interstitial fibrosis may prove difficult clinically, without biopsies specifically looking for tubulointerstitial disease.