NDT Advance Access published April 7, 2014 Nephrol Dial Transplant (2014) 0: 1–10 doi: 10.1093/ndt/gfu062
Original Article Neutrophil gelatinase-associated lipocalin as a biomarker for lupus nephritis María Teresa Torres-Salido1,2,*, Josefina Cortés-Hernández1,2,*, Xavier Vidal1,2, Anna Pedrosa1,2, 1
Department of Medicine, Systemic Autoimmune Diseases Unit, Hospital Universitari Vall d´Hebron, Institut de Recerca (VHIR), Universitat
Autònoma de Barcelona, Barcelona, Spain and 2Department of Statistical, Systemic Autoimmune Diseases Unit, Hospital Universitari Vall d´Hebron, Institut de Recerca (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain
Correspondence and offprint requests to: Josep Ordi-Ros; E-mail: [email protected] *M.T.T.-S. and J.C.-H. equally contributed to the work.
A B S T R AC T Background. One of the challenges of treating patients with lupus nephritis (LN) is to accurately assess disease activity and predict its outcome. Since renal-biopsy cannot be performed routinely, new surrogate biomarkers are needed. Methods. We evaluated neutrophil gelatinase-associated lipocalin (NGAL), to predict renal outcome in LN. Serum and urinary NGAL levels, measured by the enzyme-linked immunosorbent assay, and the fractional excretion (FE) of NGAL relative to the FE of proteins (FE NGAL/FE protein ratio) were determined in a cross-sectional (n = 199) and longitudinal (n = 45) cohort of systemic lupus erythematosus (SLE) patients. Global and renal disease activity was assessed by the SLE disease activity indices, SLEDAI and rSLEDAI, respectively. Correlations between traditional biomarkers were established. Sensitivity, specificity and predictive values of NGAL for renal flare, response to therapy and progression to chronic kidney disease were calculated. Results. The FE NGAL/FE protein ratio exhibited the best sensitivity and specificity to discriminate patients with active LN from those with non-renal flare and inactive SLE. In the prospective study, this biomarker was found to be the best candidate to predict proteinuric flares with an 87% sensitivity and 62% specificity for ratios >14.56 and complete response with a 61% sensitivity and 78% specificity for ratios >26.54 in the presence of a simultaneous worsening or improving rSLEDAIs, respectively. In both conditions, the FE NGAL/FE protein ratio outperformed the anti-dsDNA antibody titres © The Author 2014. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
and C3 predictive value. Progression to chronic kidney disease was best predicted by estimated glomerular filtration rate levels, but persistently high levels of serum NGAL (>444.4 ng/mL, P = 0.0001 by Kaplan–Meier) predicted a faster progression. Conclusions. The FE NGAL/FE protein ratio is a reliable marker of disease activity in patients with SLE and could be used as an indicator of response to therapy, although further studies are required to confirm these results. Keywords: lupus nephritis, neutrophil gelatinase-associated lipocalin, urinary biomarkers
INTRODUCTION Kidney involvement is a major concern in systemic lupus erythematosus (SLE). Renal disease occurs in ∼40–75% of patients and is associated with an unpredictable course and significant short- and long-term morbidity [1]. Despite initial aggressive therapy, up to 25% of patients with lupus nephritis (LN) will progress to renal damage [2], partly as a result of the difficulty in recognizing the onset of the disease and relapses, early enough to establish a prompt treatment able to modify the course of the disease. At present, renal biopsy remains the ‘gold standard’ to diagnose and provide direct assessment of the degree of kidney damage, but due to its invasive nature cannot be performed serially. Traditional serological biomarkers, such as anti-dsDNA antibodies and complement levels, and laboratory tests for kidney function have been proved unreliable [3, 4]. For this 1
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
Miquel Vilardell-Tarrés1,2 and Josep Ordi-Ros1,2
ORIGINAL ARTICLE
M AT E R I A L S A N D M E T H O D S SLE patients from the Lupus Unit at Vall d’Hebron Hospital (Barcelona, Spain), fulfilling at least four of the American College of Rheumatology (ACR) revised classification criteria for SLE [21], were enrolled. Patients with urinary tract infection, diabetes mellitus, pregnancy, malignancy, non-lupusrelated renal failure and those undergoing haemodialysis or with renal transplantation history were excluded. The study was approved by the local Ethics Committee and all patients gave written informed consent. Disease activity was assessed by the SLE Disease Activity Index 2000 update (SLEDAI-2Ks; range 0–105) [22]. Renal activity was defined as the sum of the SLEDAI-2Ks accrued in the renal domain of the measure (rSLEDAIs; range 0–16). Extra-renal disease activity was calculated by subtracting the rSLEDAIs from the total SLEDAI score [SLEDAI-2K extrarenal score (eSLEDAIs); range 0–89]. Accumulated renal damage was assessed by the renal-related data of the SLICC/ ACR damage index (rSDI; range 0–47) [23]. Renal biopsies were performed within 1 week of starting treatment, unless contraindicated. Renal-biopsy specimens were examined by a light and immunofluorescence microscopy, and the findings were categorized according to the International Society of Nephrology/Renal Pathology Society classification (2003) [24] and rated for activity and chronicity [25]. Two different cross-sectional and prospective cohorts of patients were established to evaluate the association of NGAL levels with disease activity and its predictive value in clinical outcome, respectively.
2
Patients in the cross-sectional study were categorized into: (i) SLE patients with LN. Those in turn were subclassified into three groups according to the renal activity status at inclusion: patients with active LN (ALN), defined by a new rSLEDAI ≥8 or rSLEDAI = 4 when proteinuria was the only renal-related criterion, and/or by biopsy-proven active renal disease (n = 38), and those with a previously treated LN that at the inclusion time achieved a partial response (PR) (n = 56) or complete response (CR) (n = 29), defined by a urinary protein-to-creatinine ratio (Upro : Ucre ratio) between 0.2–2.0 and 1.8 mg/mg after PR [27]. At each visit the SLEDAI-2Ks was recorded and laboratory tests were performed, including: full blood count, serum biochemistry, complement levels, anti-dsDNA antibody titres, complete urine analysis, 24-h proteinuria, Upro : Ucr ratio and the eGFR by using the Cockcroft–Gault equation [28]. All samples were obtained during the morning hours on the day of the clinic visit. The LN disease course was categorized based on the change in disease activity at a reference time point, either due to the development of flare (time A) or achievement of CR following treatment (time B), and progression to CKD. The respective disease activity and response to treatment scores were compared between two time points: the time of the event (time A or B) and the time of the preceding visits (–A1, –A2 and –B1, –B2). Urinary (uNGAL) and serum NGAL (sNGAL) measurements by the enzyme-linked immunosorbent assay (ELISA). For each patient, fresh blood and urine samples were collected. Urine samples were centrifuged at 2500 g (4°C) for 30 min to remove cellular debris before storing. Urine and serum samples were frozen within 2 h after collection and stored at −80°C until further analysis. NGAL levels were measured by ELISA using a commercially available kit (Bioporto, Denmark) following the manufacturer’s instructions. Plates were read at 450 nm. All the measurements were made in duplicate. The lower limit of detection was 0.5–4 ng/mL. Undetectable values were assigned a value of 0. Assay standardization uNGAL concentrations were normalized to urine-creatinine to correct for differences due to urine dilution in the
M. T. Torres-Salido et al.
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
reason, in recent years, there has been a growing interest in finding a non-invasive ‘surrogate marker’ of lupus renal disease that could be used to predict the onset of the disease and to monitor its progression. Urinary biomarkers are attractive candidates [5, 6] since not only are they relatively easy to measure, but they specifically reflect the local pathophysiological changes. Neutrophil gelatinase-associated lipocalin (NGAL) is a 25 kDa protein belonging to the lipocalin superfamily that has been extensively studied in acute kidney injury . It is one of the most robustly expressed proteins in the kidney following ischaemic or nephrotoxic injury in both animal models and humans [7–9]. In adult and childhood-onset SLE, it has been found to be a significant predictor of disease activity and flare [10–15], whereas in patients with an existing renal disease it is a predictor of disease progression [16–18]. In addition, NGAL also seems to be involved in renoprotection, in view of its implication in the development and de-differentiation of tubular epithelia following ischemia-reperfusion, it may predict graft function recovery, and administered into animal models, can mitigate renal injury [19–20]. Therefore, we hypothesized that NGAL, besides being a biomarker of active and/or progressive disease, can also be a biomarker of recovery. The goal of our study is to investigate the association of NGAL with renal disease activity, response to therapy and progression to chronic kidney disease (CKD).
same sample, with results being expressed as nanograms of uNGAL per milligram of urinary creatinine (ng/mg Cr). Fractional excretion (FE) indicators of NGAL (FE NGAL) were calculated using the formula: (urine/serum analyte concentration)/(urine/serum creatinine concentration) × 100. FE NGAL relative to the FE of total proteins (FE NGAL/FE protein) was calculated.
R E S U LT S Baseline patient characteristics Patients’ demographic characteristics, medication usage and laboratory measurements both in the cross-sectional and longitudinal studies are summarized in Tables 1 and 2. In both cohorts, there was a female (87–89%) and biopsy-proven type IV GMN (78–79%) predominance. In the cross-sectional cohort, renal patients had significantly higher SLEDAI-2Ks and rSLEDAIs and were taking mycophenolate mofetil (MMF) (P = 0.0006) and angiotensin-blocking agents (P = 0.0184) more frequently, whereas hydroxychloroquine was more prevalent in non-renal patients (P = 0.0087). As expected, patients with LN had significantly higher anti-dsDNA antibodies titres, creatinine levels and Upro : Ucre ratios along with lower eGFRs. All patients in the longitudinal study received MMF as continuous induction-maintenance therapy.
NGAL as a biomarker of LN
NGAL and global disease activity. In the cross-sectional study, uNGAL and FE NGAL/FE protein ratio had a significant association with the global SLEDAI-2Ks (r = 0.34 and r = −0.2336, P < 0.0001 and P = 0.0001, respectively) that was mainly dependent on the renal components of the score, since the correlation was lost when only extra-renal manifestations (eSLEDAIs) were considered (r = 0.04 and r = 0.03, ns, respectively). Globally, the association with sNGAL was also weaker (r = 0.1473, P = 0.0454). Similar results were observed in the longitudinal study (data not shown). No significant association was found between NGAL measurements and serum antidsDNA titres (measured by ELISA) and complement (C3 and C4) levels. Longitudinal NGAL changes and renal disease course LN course and changes in disease activity. A total of 278 observations of reference time points were available during the follow-up period. The median (Q1–Q3) number of visits for
3
ORIGINAL ARTICLE
Longitudinal study. Descriptive, inferential and correlation statistics were estimated with the same methods as in the cross-sectional study. Within-patient comparisons at different time points were carried out with generalized estimating equations in order to account for the repeated measurements design. Kaplan–Meier and Cox proportional hazards methods were used to analyse the relationship between the markers and the time to kidney disease progression. Statistical analyses were done with PASW Statistics 17.0 (SPSS, Inc., 2009, Chicago, IL, USA) and SAS 9.1.3 (SAS Institute, Inc., Cary, NC, USA).
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
Statistical analysis Cross-sectional study. Percentages and quartiles were used as descriptive statistics for categorical and continuous variables, respectively. Inferences on population mean values were done with 95% confidence intervals. Comparisons between groups were performed by means of the Chi-square test or Fisher’s exact test for categorical and the Mann–Whitney U test for continuous values. Spearman’s rank correlation coefficient was used to estimate the association between continuous variables. To perform multiple comparisons of various biochemical markers against the ALN group, a generalized linear model with a log link was fitted, and then Dunnett’s test was applied in order to adjust the P-values. Areas under ROC curves were calculated and the cut-off points were determined according to Youden’s index. Sensitivity, specificity and positive and negative predictive values, and their 95% confidence intervals, were also computed using generalized linear models with a logit link.
NGAL levels in relation to disease activity NGAL and renal disease activity. SLE patients had significantly higher uNGAL levels than healthy controls [52.11 (95% CI: 39.61–64.61) versus 20.41 (95% CI: 15.47–25.36])ng/mL, P = 0.0002]. Levels of uNGAL and normalized uNGAL were found to be strongly correlated (r = 0.8436, P < 0.0001). Therefore, all the results are presented as normalized to the urinecreatinine concentration. Patients with renal involvement, and mainly those with an active disease, exhibited the highest levels when compared with patients with non-renal flare [0.52 (0.34–0.70) versus 0.27 (0.15–0.40) ng/creatinine mg, P = 0.0028] or those in CR [0.33 (0.21–0.45), P = 0.0174], respectively (Figure 1A). uNGAL did not discriminate between patients with ALN and PR. In contrast, although sNGAL levels were not statistically different between SLE patients and controls [339.3 (95% CI: 302.2–376.4) versus 293.2 (95% CI: 241.4–344.9) pg/mL, P = 0.846], a significant difference was found between patients with ALN and those with non-active SLE (P = 0.008) (Figure 1B). The FE NGAL/FE Prot ratio was the best parameter to discriminate the groups, also allowing the differentiation between the ALN group, who exhibited the lowest rates (Figure 1C) and the group in PR. ROC curves for uNGAL, sNGAL and FE NGAL/FE protein ratio were also calculated to discriminate SLE patients with ALN from those with non-renal flare (Figure 1D), inactive SLE (Figure 1E), or previous renal involvement now in CR (Figure 1F). The FE NGAL/FE protein ratio also demonstrated the best ROC profile in all conditions. When we studied correlations of NGAL with other parameters, a weak correlation between sNGAL and uNGAL levels was found, as shown by the low degree of Spearman’s correlation (r = 0.2062, P = 0.0021). Levels of uNGAL also correlated with other renal parameters such as serum creatinine, Upro:Ucr ratio and rSLEDAIs and were inversely correlated with the FE NGAL/FE protein ratio (P < 0.0001) (data not shown). The level of uNGAL was the only parameter correlated with the activity score (r = 0.35, P < 0.015) at the histological level. No associations with the chronicity score or the different histological types of LN were observed.
Table 1. Demographics, disease activity scores and laboratory measures in patients with SLE SLE with renal involvement Total (n = 123)
Active LN (n = 38)
Previously treated LN (n = 85) PR (n = 56)
ORIGINAL ARTICLE
Total (n = 62)
Non-renal flare (n = 23)
Inactive SLE (n = 39)
CR (n = 29)
33 (28–42) 46 (82)
32 (27–42) 27 (96)
36 (28–45) 51 (82)
34 (29–48) 21 (91)
37 (28–43) 30 (77)
51 (90) 2 (4) 1 (2) 2 (4) 10 (4–14)
29 (100) 0 (0) 0 (0) 0 (0) 9 (4–15)
58 (94) 2 (3) 0 (0) 2 (3) 8 (5–14)
22 (96) 0 (0) 0 (0) 1 (4) 8 (1–11)
36 (91) 2 (5) 0 (0) 1 (3) 9 (5–14)
8 (6–11) 4 (4–8) 2 (0–4)
2 (1–4) 0 (0–0) 2 (1–4)
4 (2–8) 0 (0–0) 4 (2–8)
8 (7–10) 0 (0–0) 8 (7–10)
2 (2–4) 0 (0–0) 2 (2–4)
51 (91) 38 (68) 6 (11) 11 (20) 5 (9) 15 (27)
27 (93) 15 (52) 5 (17) 11 (38) 1 (3) 5 (17)
59 (95) 25 (40) 6 (10) 25 (40) 5 (8) 5 (5)
23 (100) 11 (48) 1 (4) 9 (39) 2 (9) 1 (4)
0.9 (0.7–1.5) 0.9 (0.8–1) 0.8 (0.7–0.9) 0.8 (0.7–0.8) 27 (11–57) 18 (4–89) 15 (3–90) 39 (2–188) 95 (76–115) 93 (82–113) 95 (82–111) 94 (80–129) 0.4 (0.3–0.8) 0.06 (0.04–0.1) 0.06 (0.04–0.1) 0.06 (0.04–0.1) 88 (63–111) 94 (74–110) 98 (82–116) 104 (87–120) 7 (13)
3 (10)
0 (0) 4 (8) 43 (88) 2 (4) 12 (9–12)
3 (12) 3 (12) 15 (58) 5 (19) 7 (3–12)
36 (92) 14 (36) 5 (13) 16 (41) 3 (8) 4 (10)
0.9 (0.8–1) 11 (4–51) 96 (84–111) 0.06 (0.04–0.1) 98 (79–112)
0.2060 0.2323 1.0000
0.7315
26.54 [ROC AUC 0.692 (95% CI: 0.566–0.818)] exhibited the highest specificity to predict CR (Table 4). The specificity was similar to the anti-dsDNA antibodies measurement, but the FE NGAL/FE protein ratio showed the highest sensitivity and negative predictive value (NPV) to predict CR (Table 4). When results of FE NGAL/FE protein ratios were combined with the improvement of rSLEDAIs over time, the specificity rose to 91%. For patients with sustained partial remission, as it occurred in the cross-sectional study, the FE NGAL/FE protein ratio remained low through the follow-up [19.62 ± 23.64 (mean ± SD)]. A value 14.56 [ROC AUC of 0.6995 (95% CI: 0.508–0.890)] displayed the best predictive profile when compared with traditional biomarkers such as anti-dsDNA antibodies and C3 levels (Table 4).
ORIGINAL ARTICLE 6 Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
M. T. Torres-Salido et al.
F I G U R E 1 : Urinary and serum NGAL levels, FE NGAL and FE NGAL relative to the fractional excretion of total proteins (FE NGAL/ FE protein ratio) in SLE patients and controls. Values are mean and 95% CI. The histograms show urinary (a), urinary normalized (b) and serum (c), fractional excretion (FE) of NGAL relative to the fractional excretion of total proteins (FE NGAL/ FE protein). Specificity/sensitivity ROC curves pertaining to the use of urinary normalized ( ) and serum ( ) NGAL levels, and fractional excretion (FE) of uNGAL relative to the fractional excretion of total proteins (FE NGAL/FE proteins) ( ) for discriminating active lupus nephritis for non-renal flare (d), inactive SLE (e), and CR (f ) groups. PPV, positive predictive value, NPV, negative predictive value.
7
−3 (−5 to (−1)) 0.15 (−0.09 to 0.4) −15 (−115 to 86) 20 (3.3 to 37) −9 (−11 to (−7)) 0.13 (−0.1 to 0.4) 49 (−122 to 221) 134 (38 to 233) −6 (−8 to (−4)) −0.01 (−0.18 to 0.16) 64 (−49 to 178) 115 (24 to 207) 0 (0 to 0) 0.5 (0.2 to 0.7) 357 (244 to 471) 157 (73 to 241) 6 (4 to 8) 0.5 (0.3 to 0.7) 293 (232 to 354) 41 (23 to 59) 9 (7 to 11) 0.3 (0.2 to 0.7) 308 (225 to 391) 21 (10 to 32) rSLEDAI-2Kb uNGAL(ng/Cr mg) sNGAL (ng/mL) FE NGAL/FE protein ratio
rSLEDAI-2K, renal SLEDAI-2K; uNGAL, urinary neutrophil gelatinase-associated lipocalin; uNGAL/Cr, urinary neutrophil gelatinase-associated lipocalin normalized to the urinary creatinine levels; sNGAL, serum neutrophil gelatinase-associated lipocalin; FE NGAL (%), fractional excretion (FE) of NGAL; FE NGAL/FE protein ratio, fractional excretion (FE) of NGAL relative to fractional Excretion of total proteins; 95% CI = 95% confidence interval; NS = not significant. a Time −A2 = time point 2 visits prior to the reference time point; time −A1 = time point 1 visit prior to the reference time point; time A (Flare) = reference time point at which the disease course was defined. b Renal SLEDAI was scored on a urinalysis that indicates the presence of proteinuria, haematuria, pyuria and urinary cast, each receiving a score of up four points for a total possible ranging from 0 to 16. c Time −B2 = time point 2 visits prior to the reference time point; time −B1 = time point 1 visit prior to the reference time point; time B (complete response) = reference time point at which the disease course was defined.
0.007 NS NS 0.012
−B1 versus −B2
B versus −B1 −B2 0.421 Sensitivity 44 (17–75) Specificity 66 (56–75) Positive predictive value 5 (2–13) Negative predictive value 96 (92–98) Variable Predicted probability of complete response (%, 95% CI) Cut-off >0.1046 Sensitivity 100 (82–100) Specificity 17 (10–26) Positive predictive value 9 (6–13) Negative predictive value 100 (91–100)
sNGAL (ng/mL)
FE NGAL/FE Prot ratio
anti-dsDNA titres (IU/mL)
C3 (mg/dL)
>1090 11 (1.5–51) 98 (94–99) 14 (2–58) 96 (93–98)
>14.56 87 (44–98) 62 (52–72) 8 (4–15) 99 (95–100)
>34.5 79 (44–95) 37 (25–49) 5 (3–10) 97 (90–99)
Original Article Neutrophil gelatinase-associated lipocalin as a biomarker for lupus nephritis María Teresa Torres-Salido1,2,*, Josefina Cortés-Hernández1,2,*, Xavier Vidal1,2, Anna Pedrosa1,2, 1
Department of Medicine, Systemic Autoimmune Diseases Unit, Hospital Universitari Vall d´Hebron, Institut de Recerca (VHIR), Universitat
Autònoma de Barcelona, Barcelona, Spain and 2Department of Statistical, Systemic Autoimmune Diseases Unit, Hospital Universitari Vall d´Hebron, Institut de Recerca (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain
Correspondence and offprint requests to: Josep Ordi-Ros; E-mail: [email protected] *M.T.T.-S. and J.C.-H. equally contributed to the work.
A B S T R AC T Background. One of the challenges of treating patients with lupus nephritis (LN) is to accurately assess disease activity and predict its outcome. Since renal-biopsy cannot be performed routinely, new surrogate biomarkers are needed. Methods. We evaluated neutrophil gelatinase-associated lipocalin (NGAL), to predict renal outcome in LN. Serum and urinary NGAL levels, measured by the enzyme-linked immunosorbent assay, and the fractional excretion (FE) of NGAL relative to the FE of proteins (FE NGAL/FE protein ratio) were determined in a cross-sectional (n = 199) and longitudinal (n = 45) cohort of systemic lupus erythematosus (SLE) patients. Global and renal disease activity was assessed by the SLE disease activity indices, SLEDAI and rSLEDAI, respectively. Correlations between traditional biomarkers were established. Sensitivity, specificity and predictive values of NGAL for renal flare, response to therapy and progression to chronic kidney disease were calculated. Results. The FE NGAL/FE protein ratio exhibited the best sensitivity and specificity to discriminate patients with active LN from those with non-renal flare and inactive SLE. In the prospective study, this biomarker was found to be the best candidate to predict proteinuric flares with an 87% sensitivity and 62% specificity for ratios >14.56 and complete response with a 61% sensitivity and 78% specificity for ratios >26.54 in the presence of a simultaneous worsening or improving rSLEDAIs, respectively. In both conditions, the FE NGAL/FE protein ratio outperformed the anti-dsDNA antibody titres © The Author 2014. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
and C3 predictive value. Progression to chronic kidney disease was best predicted by estimated glomerular filtration rate levels, but persistently high levels of serum NGAL (>444.4 ng/mL, P = 0.0001 by Kaplan–Meier) predicted a faster progression. Conclusions. The FE NGAL/FE protein ratio is a reliable marker of disease activity in patients with SLE and could be used as an indicator of response to therapy, although further studies are required to confirm these results. Keywords: lupus nephritis, neutrophil gelatinase-associated lipocalin, urinary biomarkers
INTRODUCTION Kidney involvement is a major concern in systemic lupus erythematosus (SLE). Renal disease occurs in ∼40–75% of patients and is associated with an unpredictable course and significant short- and long-term morbidity [1]. Despite initial aggressive therapy, up to 25% of patients with lupus nephritis (LN) will progress to renal damage [2], partly as a result of the difficulty in recognizing the onset of the disease and relapses, early enough to establish a prompt treatment able to modify the course of the disease. At present, renal biopsy remains the ‘gold standard’ to diagnose and provide direct assessment of the degree of kidney damage, but due to its invasive nature cannot be performed serially. Traditional serological biomarkers, such as anti-dsDNA antibodies and complement levels, and laboratory tests for kidney function have been proved unreliable [3, 4]. For this 1
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
Miquel Vilardell-Tarrés1,2 and Josep Ordi-Ros1,2
ORIGINAL ARTICLE
M AT E R I A L S A N D M E T H O D S SLE patients from the Lupus Unit at Vall d’Hebron Hospital (Barcelona, Spain), fulfilling at least four of the American College of Rheumatology (ACR) revised classification criteria for SLE [21], were enrolled. Patients with urinary tract infection, diabetes mellitus, pregnancy, malignancy, non-lupusrelated renal failure and those undergoing haemodialysis or with renal transplantation history were excluded. The study was approved by the local Ethics Committee and all patients gave written informed consent. Disease activity was assessed by the SLE Disease Activity Index 2000 update (SLEDAI-2Ks; range 0–105) [22]. Renal activity was defined as the sum of the SLEDAI-2Ks accrued in the renal domain of the measure (rSLEDAIs; range 0–16). Extra-renal disease activity was calculated by subtracting the rSLEDAIs from the total SLEDAI score [SLEDAI-2K extrarenal score (eSLEDAIs); range 0–89]. Accumulated renal damage was assessed by the renal-related data of the SLICC/ ACR damage index (rSDI; range 0–47) [23]. Renal biopsies were performed within 1 week of starting treatment, unless contraindicated. Renal-biopsy specimens were examined by a light and immunofluorescence microscopy, and the findings were categorized according to the International Society of Nephrology/Renal Pathology Society classification (2003) [24] and rated for activity and chronicity [25]. Two different cross-sectional and prospective cohorts of patients were established to evaluate the association of NGAL levels with disease activity and its predictive value in clinical outcome, respectively.
2
Patients in the cross-sectional study were categorized into: (i) SLE patients with LN. Those in turn were subclassified into three groups according to the renal activity status at inclusion: patients with active LN (ALN), defined by a new rSLEDAI ≥8 or rSLEDAI = 4 when proteinuria was the only renal-related criterion, and/or by biopsy-proven active renal disease (n = 38), and those with a previously treated LN that at the inclusion time achieved a partial response (PR) (n = 56) or complete response (CR) (n = 29), defined by a urinary protein-to-creatinine ratio (Upro : Ucre ratio) between 0.2–2.0 and 1.8 mg/mg after PR [27]. At each visit the SLEDAI-2Ks was recorded and laboratory tests were performed, including: full blood count, serum biochemistry, complement levels, anti-dsDNA antibody titres, complete urine analysis, 24-h proteinuria, Upro : Ucr ratio and the eGFR by using the Cockcroft–Gault equation [28]. All samples were obtained during the morning hours on the day of the clinic visit. The LN disease course was categorized based on the change in disease activity at a reference time point, either due to the development of flare (time A) or achievement of CR following treatment (time B), and progression to CKD. The respective disease activity and response to treatment scores were compared between two time points: the time of the event (time A or B) and the time of the preceding visits (–A1, –A2 and –B1, –B2). Urinary (uNGAL) and serum NGAL (sNGAL) measurements by the enzyme-linked immunosorbent assay (ELISA). For each patient, fresh blood and urine samples were collected. Urine samples were centrifuged at 2500 g (4°C) for 30 min to remove cellular debris before storing. Urine and serum samples were frozen within 2 h after collection and stored at −80°C until further analysis. NGAL levels were measured by ELISA using a commercially available kit (Bioporto, Denmark) following the manufacturer’s instructions. Plates were read at 450 nm. All the measurements were made in duplicate. The lower limit of detection was 0.5–4 ng/mL. Undetectable values were assigned a value of 0. Assay standardization uNGAL concentrations were normalized to urine-creatinine to correct for differences due to urine dilution in the
M. T. Torres-Salido et al.
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
reason, in recent years, there has been a growing interest in finding a non-invasive ‘surrogate marker’ of lupus renal disease that could be used to predict the onset of the disease and to monitor its progression. Urinary biomarkers are attractive candidates [5, 6] since not only are they relatively easy to measure, but they specifically reflect the local pathophysiological changes. Neutrophil gelatinase-associated lipocalin (NGAL) is a 25 kDa protein belonging to the lipocalin superfamily that has been extensively studied in acute kidney injury . It is one of the most robustly expressed proteins in the kidney following ischaemic or nephrotoxic injury in both animal models and humans [7–9]. In adult and childhood-onset SLE, it has been found to be a significant predictor of disease activity and flare [10–15], whereas in patients with an existing renal disease it is a predictor of disease progression [16–18]. In addition, NGAL also seems to be involved in renoprotection, in view of its implication in the development and de-differentiation of tubular epithelia following ischemia-reperfusion, it may predict graft function recovery, and administered into animal models, can mitigate renal injury [19–20]. Therefore, we hypothesized that NGAL, besides being a biomarker of active and/or progressive disease, can also be a biomarker of recovery. The goal of our study is to investigate the association of NGAL with renal disease activity, response to therapy and progression to chronic kidney disease (CKD).
same sample, with results being expressed as nanograms of uNGAL per milligram of urinary creatinine (ng/mg Cr). Fractional excretion (FE) indicators of NGAL (FE NGAL) were calculated using the formula: (urine/serum analyte concentration)/(urine/serum creatinine concentration) × 100. FE NGAL relative to the FE of total proteins (FE NGAL/FE protein) was calculated.
R E S U LT S Baseline patient characteristics Patients’ demographic characteristics, medication usage and laboratory measurements both in the cross-sectional and longitudinal studies are summarized in Tables 1 and 2. In both cohorts, there was a female (87–89%) and biopsy-proven type IV GMN (78–79%) predominance. In the cross-sectional cohort, renal patients had significantly higher SLEDAI-2Ks and rSLEDAIs and were taking mycophenolate mofetil (MMF) (P = 0.0006) and angiotensin-blocking agents (P = 0.0184) more frequently, whereas hydroxychloroquine was more prevalent in non-renal patients (P = 0.0087). As expected, patients with LN had significantly higher anti-dsDNA antibodies titres, creatinine levels and Upro : Ucre ratios along with lower eGFRs. All patients in the longitudinal study received MMF as continuous induction-maintenance therapy.
NGAL as a biomarker of LN
NGAL and global disease activity. In the cross-sectional study, uNGAL and FE NGAL/FE protein ratio had a significant association with the global SLEDAI-2Ks (r = 0.34 and r = −0.2336, P < 0.0001 and P = 0.0001, respectively) that was mainly dependent on the renal components of the score, since the correlation was lost when only extra-renal manifestations (eSLEDAIs) were considered (r = 0.04 and r = 0.03, ns, respectively). Globally, the association with sNGAL was also weaker (r = 0.1473, P = 0.0454). Similar results were observed in the longitudinal study (data not shown). No significant association was found between NGAL measurements and serum antidsDNA titres (measured by ELISA) and complement (C3 and C4) levels. Longitudinal NGAL changes and renal disease course LN course and changes in disease activity. A total of 278 observations of reference time points were available during the follow-up period. The median (Q1–Q3) number of visits for
3
ORIGINAL ARTICLE
Longitudinal study. Descriptive, inferential and correlation statistics were estimated with the same methods as in the cross-sectional study. Within-patient comparisons at different time points were carried out with generalized estimating equations in order to account for the repeated measurements design. Kaplan–Meier and Cox proportional hazards methods were used to analyse the relationship between the markers and the time to kidney disease progression. Statistical analyses were done with PASW Statistics 17.0 (SPSS, Inc., 2009, Chicago, IL, USA) and SAS 9.1.3 (SAS Institute, Inc., Cary, NC, USA).
Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
Statistical analysis Cross-sectional study. Percentages and quartiles were used as descriptive statistics for categorical and continuous variables, respectively. Inferences on population mean values were done with 95% confidence intervals. Comparisons between groups were performed by means of the Chi-square test or Fisher’s exact test for categorical and the Mann–Whitney U test for continuous values. Spearman’s rank correlation coefficient was used to estimate the association between continuous variables. To perform multiple comparisons of various biochemical markers against the ALN group, a generalized linear model with a log link was fitted, and then Dunnett’s test was applied in order to adjust the P-values. Areas under ROC curves were calculated and the cut-off points were determined according to Youden’s index. Sensitivity, specificity and positive and negative predictive values, and their 95% confidence intervals, were also computed using generalized linear models with a logit link.
NGAL levels in relation to disease activity NGAL and renal disease activity. SLE patients had significantly higher uNGAL levels than healthy controls [52.11 (95% CI: 39.61–64.61) versus 20.41 (95% CI: 15.47–25.36])ng/mL, P = 0.0002]. Levels of uNGAL and normalized uNGAL were found to be strongly correlated (r = 0.8436, P < 0.0001). Therefore, all the results are presented as normalized to the urinecreatinine concentration. Patients with renal involvement, and mainly those with an active disease, exhibited the highest levels when compared with patients with non-renal flare [0.52 (0.34–0.70) versus 0.27 (0.15–0.40) ng/creatinine mg, P = 0.0028] or those in CR [0.33 (0.21–0.45), P = 0.0174], respectively (Figure 1A). uNGAL did not discriminate between patients with ALN and PR. In contrast, although sNGAL levels were not statistically different between SLE patients and controls [339.3 (95% CI: 302.2–376.4) versus 293.2 (95% CI: 241.4–344.9) pg/mL, P = 0.846], a significant difference was found between patients with ALN and those with non-active SLE (P = 0.008) (Figure 1B). The FE NGAL/FE Prot ratio was the best parameter to discriminate the groups, also allowing the differentiation between the ALN group, who exhibited the lowest rates (Figure 1C) and the group in PR. ROC curves for uNGAL, sNGAL and FE NGAL/FE protein ratio were also calculated to discriminate SLE patients with ALN from those with non-renal flare (Figure 1D), inactive SLE (Figure 1E), or previous renal involvement now in CR (Figure 1F). The FE NGAL/FE protein ratio also demonstrated the best ROC profile in all conditions. When we studied correlations of NGAL with other parameters, a weak correlation between sNGAL and uNGAL levels was found, as shown by the low degree of Spearman’s correlation (r = 0.2062, P = 0.0021). Levels of uNGAL also correlated with other renal parameters such as serum creatinine, Upro:Ucr ratio and rSLEDAIs and were inversely correlated with the FE NGAL/FE protein ratio (P < 0.0001) (data not shown). The level of uNGAL was the only parameter correlated with the activity score (r = 0.35, P < 0.015) at the histological level. No associations with the chronicity score or the different histological types of LN were observed.
Table 1. Demographics, disease activity scores and laboratory measures in patients with SLE SLE with renal involvement Total (n = 123)
Active LN (n = 38)
Previously treated LN (n = 85) PR (n = 56)
ORIGINAL ARTICLE
Total (n = 62)
Non-renal flare (n = 23)
Inactive SLE (n = 39)
CR (n = 29)
33 (28–42) 46 (82)
32 (27–42) 27 (96)
36 (28–45) 51 (82)
34 (29–48) 21 (91)
37 (28–43) 30 (77)
51 (90) 2 (4) 1 (2) 2 (4) 10 (4–14)
29 (100) 0 (0) 0 (0) 0 (0) 9 (4–15)
58 (94) 2 (3) 0 (0) 2 (3) 8 (5–14)
22 (96) 0 (0) 0 (0) 1 (4) 8 (1–11)
36 (91) 2 (5) 0 (0) 1 (3) 9 (5–14)
8 (6–11) 4 (4–8) 2 (0–4)
2 (1–4) 0 (0–0) 2 (1–4)
4 (2–8) 0 (0–0) 4 (2–8)
8 (7–10) 0 (0–0) 8 (7–10)
2 (2–4) 0 (0–0) 2 (2–4)
51 (91) 38 (68) 6 (11) 11 (20) 5 (9) 15 (27)
27 (93) 15 (52) 5 (17) 11 (38) 1 (3) 5 (17)
59 (95) 25 (40) 6 (10) 25 (40) 5 (8) 5 (5)
23 (100) 11 (48) 1 (4) 9 (39) 2 (9) 1 (4)
0.9 (0.7–1.5) 0.9 (0.8–1) 0.8 (0.7–0.9) 0.8 (0.7–0.8) 27 (11–57) 18 (4–89) 15 (3–90) 39 (2–188) 95 (76–115) 93 (82–113) 95 (82–111) 94 (80–129) 0.4 (0.3–0.8) 0.06 (0.04–0.1) 0.06 (0.04–0.1) 0.06 (0.04–0.1) 88 (63–111) 94 (74–110) 98 (82–116) 104 (87–120) 7 (13)
3 (10)
0 (0) 4 (8) 43 (88) 2 (4) 12 (9–12)
3 (12) 3 (12) 15 (58) 5 (19) 7 (3–12)
36 (92) 14 (36) 5 (13) 16 (41) 3 (8) 4 (10)
0.9 (0.8–1) 11 (4–51) 96 (84–111) 0.06 (0.04–0.1) 98 (79–112)
0.2060 0.2323 1.0000
0.7315
26.54 [ROC AUC 0.692 (95% CI: 0.566–0.818)] exhibited the highest specificity to predict CR (Table 4). The specificity was similar to the anti-dsDNA antibodies measurement, but the FE NGAL/FE protein ratio showed the highest sensitivity and negative predictive value (NPV) to predict CR (Table 4). When results of FE NGAL/FE protein ratios were combined with the improvement of rSLEDAIs over time, the specificity rose to 91%. For patients with sustained partial remission, as it occurred in the cross-sectional study, the FE NGAL/FE protein ratio remained low through the follow-up [19.62 ± 23.64 (mean ± SD)]. A value 14.56 [ROC AUC of 0.6995 (95% CI: 0.508–0.890)] displayed the best predictive profile when compared with traditional biomarkers such as anti-dsDNA antibodies and C3 levels (Table 4).
ORIGINAL ARTICLE 6 Downloaded from http://ndt.oxfordjournals.org/ at Rutgers University on June 16, 2014
M. T. Torres-Salido et al.
F I G U R E 1 : Urinary and serum NGAL levels, FE NGAL and FE NGAL relative to the fractional excretion of total proteins (FE NGAL/ FE protein ratio) in SLE patients and controls. Values are mean and 95% CI. The histograms show urinary (a), urinary normalized (b) and serum (c), fractional excretion (FE) of NGAL relative to the fractional excretion of total proteins (FE NGAL/ FE protein). Specificity/sensitivity ROC curves pertaining to the use of urinary normalized ( ) and serum ( ) NGAL levels, and fractional excretion (FE) of uNGAL relative to the fractional excretion of total proteins (FE NGAL/FE proteins) ( ) for discriminating active lupus nephritis for non-renal flare (d), inactive SLE (e), and CR (f ) groups. PPV, positive predictive value, NPV, negative predictive value.
7
−3 (−5 to (−1)) 0.15 (−0.09 to 0.4) −15 (−115 to 86) 20 (3.3 to 37) −9 (−11 to (−7)) 0.13 (−0.1 to 0.4) 49 (−122 to 221) 134 (38 to 233) −6 (−8 to (−4)) −0.01 (−0.18 to 0.16) 64 (−49 to 178) 115 (24 to 207) 0 (0 to 0) 0.5 (0.2 to 0.7) 357 (244 to 471) 157 (73 to 241) 6 (4 to 8) 0.5 (0.3 to 0.7) 293 (232 to 354) 41 (23 to 59) 9 (7 to 11) 0.3 (0.2 to 0.7) 308 (225 to 391) 21 (10 to 32) rSLEDAI-2Kb uNGAL(ng/Cr mg) sNGAL (ng/mL) FE NGAL/FE protein ratio
rSLEDAI-2K, renal SLEDAI-2K; uNGAL, urinary neutrophil gelatinase-associated lipocalin; uNGAL/Cr, urinary neutrophil gelatinase-associated lipocalin normalized to the urinary creatinine levels; sNGAL, serum neutrophil gelatinase-associated lipocalin; FE NGAL (%), fractional excretion (FE) of NGAL; FE NGAL/FE protein ratio, fractional excretion (FE) of NGAL relative to fractional Excretion of total proteins; 95% CI = 95% confidence interval; NS = not significant. a Time −A2 = time point 2 visits prior to the reference time point; time −A1 = time point 1 visit prior to the reference time point; time A (Flare) = reference time point at which the disease course was defined. b Renal SLEDAI was scored on a urinalysis that indicates the presence of proteinuria, haematuria, pyuria and urinary cast, each receiving a score of up four points for a total possible ranging from 0 to 16. c Time −B2 = time point 2 visits prior to the reference time point; time −B1 = time point 1 visit prior to the reference time point; time B (complete response) = reference time point at which the disease course was defined.
0.007 NS NS 0.012
−B1 versus −B2
B versus −B1 −B2 0.421 Sensitivity 44 (17–75) Specificity 66 (56–75) Positive predictive value 5 (2–13) Negative predictive value 96 (92–98) Variable Predicted probability of complete response (%, 95% CI) Cut-off >0.1046 Sensitivity 100 (82–100) Specificity 17 (10–26) Positive predictive value 9 (6–13) Negative predictive value 100 (91–100)
sNGAL (ng/mL)
FE NGAL/FE Prot ratio
anti-dsDNA titres (IU/mL)
C3 (mg/dL)
>1090 11 (1.5–51) 98 (94–99) 14 (2–58) 96 (93–98)
>14.56 87 (44–98) 62 (52–72) 8 (4–15) 99 (95–100)
>34.5 79 (44–95) 37 (25–49) 5 (3–10) 97 (90–99)
