Association of Serum Nitrate and Nitrite Levels With Longitudinal Assessments of Disease Activity and Damage in Systemic Lupus Erythematosus and Lupus Nephritis

Abstract

Objective

Reactive intermediate production is an essential component of the innate immune response that is induced during disease activity in murine lupus. This study was undertaken to determine whether a marker of systemic nitric oxide (NO) production correlates with prospectively studied disease activity in human systemic lupus erythematosus (SLE) and lupus nephritis patients.

Methods

Eighty-three SLE patients and 40 control subjects were studied longitudinally. The SLE group included 23 patients with lupus nephritis documented by renal biopsy and 26 with a history of lupus nephritis. During each visit, following a 24-hour low-nitrate diet, traditional markers of disease activity and damage were determined. Serum nitrate plus nitrite (NOx) levels were determined by chemiluminescence detection.

Results

NOx levels were higher in SLE patients than in controls during the first visit. In univariate longitudinal analyses, NOx levels were associated with SLE Disease Activity Index scores. In multivariate analyses, NOx levels were associated with serum levels of C3 and creatinine and the urinary protein:creatinine ratio. Among patients with lupus nephritis, those with proliferative lesions had higher NOx levels, and higher NOx levels were associated with accumulation of renal damage and lack of response to therapy.

Conclusion

This is the first study to prospectively demonstrate longitudinal associations between serum NOx levels and markers of SLE and lupus nephritis disease activity. The more pronounced association with proliferative lupus nephritis and with longitudinal response to lupus nephritis therapy provides a rationale for the study of reactive intermediates as biomarkers of disease activity and therapeutic targets in proliferative lupus nephritis.

A common feature in the broad phenotype of systemic lupus erythematosus (SLE) is autoantibody production, a function primarily of the acquired immune system (1). However, an inappropriately active innate immune response is implicated in both the initiation and the pathogenic consequences of autoantibody production in SLE. An important antimicrobial factor in the innate immune response is the production of reactive oxygen and nitrogen intermediates (RONI), including superoxide and nitric oxide (NO) (2).

In murine models of lupus, NO production has been shown to increase with the progression of renal disease. Pharmacologic inhibition of inducible NO synthase (iNOS) in these models significantly reduced both NO and reactive oxygen intermediate (ROI) production and also reduced inflammatory glomerular lesions, without affecting the deposition of immunoglobulin or complement (2,3). These findings suggest that iNOS activity or products of iNOS such as superoxide and NO contribute to the proliferative glomerular lesions in murine lupus nephritis and that their production is distal to immune complex deposition and complement activation. Cross-sectional studies of NO production in human SLE have demonstrated increased levels of markers of systemic and local NO production in association with disease activity (4-12). These studies, while important, were limited by their retrospective cross-sectional design (4,8-14), lack of control for dietary sources of nitrates, and small study populations (4,6-12).

The present investigation was conducted to address these limitations by prospectively examining the longitudinal association between production of NO and lupus disease activity in individual patients. We hypothesized that NO production would be 1) increased in SLE patients compared with control subjects, 2) associated with SLE disease activity markers over time, and 3) associated with measures of proliferative lupus nephritis activity and damage as measured by simultaneous renal biopsy and renal outcome assessment.

PATIENTS AND METHODS

Study subjects

This study was approved by the Medical University of South Carolina (MUSC) Institutional Review Board. All SLE patients met at least 4 components of the American College of Rheumatology (ACR) 1997 revised criteria for SLE (15). Control subjects were free from autoimmune disease. Exclusion criteria were the presence of active infection and the presence of cancer, since both can affect serum nitrate plus nitrite (NOx) levels. Subjects who smoked cigarettes were asked not to smoke for 24 hours prior to each visit. All subjects consumed a low-NOx diet for 24 hours prior to each visit to reduce dietary contribution to serum NOx levels (16).

Clinical evaluation

All subjects were examined prospectively at the MUSC General Clinical Research Center (GCRC) between May 1997 and September 2006. During the first visit for each SLE patient, it was confirmed that at least 4 components of the ACR SLE criteria were met. Almost all SLE patients were seen more than once (range 1–9 visits for SLE patients and 1–2 visits for controls). From 1997 to 2002, SLE patients were evaluated every 3 months at prescheduled visits for 2 years. In 2003, the protocol was changed to scheduled visits every 3 months for 1 year. History-taking, physical examination, phlebotomy, and urine collection were performed during each visit. SLE Disease Activity Index (SLEDAI) scores (17) were determined at each visit using SLEDAI-specific clinical laboratory measures and elements of the history and physical examination findings (18). The physicians scoring the SLEDAI (JCO and GSG) were trained both in-house and as part of multicenter trials that utilized this instrument. If the prednisone dosage was increased by the treating physician to >10 mg daily during the first visit, study blood collection was performed within 48 hours after this dosage increase.

Routine laboratory measures and processing of samples

All laboratory evaluations used to determine SLEDAI scores were performed in the clinical laboratory at the MUSC, using standard quality control measures. These laboratory measures included determination of serum levels of C3, C4, creatinine, and anti–double-stranded DNA (anti-dsDNA; measured by Crithidia luciliae assay before May 2002 and by enzyme-linked immunosorbent assay after May 2002), complete blood cell counts, erythrocyte sedimentation rate (ESR), and urine levels of protein and creatinine, and performance of microscopic analysis of urine. Before September 2004, urinary protein:creatinine ratios were calculated from 24-hour collections. After that date, most were calculated from spot collections. All samples for analysis were promptly processed in the GCRC laboratory. Serum samples for NOx measurements were centrifuged and immediately frozen in aliquots at −80°C.

Renal biopsy grading

As part of routine medical care during the study, renal biopsies were performed on 23 patients with active lupus nephritis. Active lupus nephritis was defined as hematuria, increasing proteinuria, or an increasing serum creatinine level necessitating renal biopsy to direct treatment decisions. All but 4 of the 23 patients with active nephritis had biopsies performed near the time of the first visit; in the remaining 4, biopsies were performed around the time of the second visit. A single, experienced renal pathologist (SES) reviewed the biopsy specimens and graded them according to World Health Organization (WHO) class (19) and National Institutes of Health (NIH) activity and chronicity indices (20). For the purposes of statistical analysis, specimens exhibiting combined proliferative and mesangial/membranous disease were classified by their proliferative lesion. Those with combined membranous and mesangial disease were classified as having WHO class V lesions. Only those with pure mesangial lesions were identified as WHO class II.

Measurement of NOx

Serum samples were analyzed by chemiluminescence detection according to the instructions of the manufacturer (NOA 280; Sievers, Boulder, CO) after reduction in vanadium chloride in hydrochloric acid (Sigma, St. Louis, MO) at 95°C. With each change of reducing reagent in the reaction vessel, a standard curve was created using dilutions of NO₃. Throughout the run, known standards were injected to determine the efficiency of the reduction reaction. Samples were analyzed in duplicate readings. Analyses were repeated when the relative standard deviation of duplicate measures exceeded 10% of the mean.

Statistical analysis

The significance of differences between SLE patients and control subjects was assessed using Pearson's chi-square test (with Fisher's 2-sided exact test when appropriate) for categorical variables, and Student's t-test (for normally distributed data) or Mann-Whitney U test for continuous variables. Bivariate correlations between variables in cross-sectional analyses were determined using either Pearson's or Spearman's correlation, depending on the distribution of the data.

To determine longitudinal associations between serum NOx levels and lupus disease activity, the following were used as input variables in a repeated-measures mixed model: C3, race, sex, SLE diagnosis (SLE or control), serum creatinine level, urine red blood cell count, peripheral white blood cell count, urinary protein:creatinine ratio, daily equivalent prednisone dose, and SLEDAI score. Serum NOx was the output variable. The repeated-measures mixed model accounts for associations between variables within individuals over time. This procedure was repeated selecting for lupus patients overall, and then again selecting only for patients with lupus nephritis. Due to the colinearity between SLEDAI variables and the SLEDAI score itself, another mixed model analysis was performed to determine the direct association between SLEDAI scores and serum NOx levels. The sum of the NOx area under the curve (AUC) between visits (AUC NOx) was calculated as a surrogate measure for the burden of exposure to NO production. The sum of the AUC for the SLEDAI (AUC SLEDAI) was calculated as a surrogate for disease burden over time. Due to the waxing and waning nature of SLE disease activity, this measure has been promoted as a marker of overall disease burden over time that can be used as an outcome measure in SLE treatment trials (21). Because not all visits fell within 3 weeks of the predetermined 12-week visit interval, a separate analysis was performed with AUC values normalized to time, by dividing by the time interval represented. The relative change in serum creatinine level (ΔCr) was calculated as follows: ΔCr = (Cr_last − Cr_first)/Cr_first. All data were expressed as the mean ± SEM. P values less than or equal to 0.05 were considered significant. SAS version 9.1 (SAS Institute, Cary, NC) was used for data analysis.

RESULTS

Demographic and clinical characteristics of the subjects

Eighty-three SLE patients and 40 control subjects were enrolled in the study and followed up prospectively at prescheduled visits every 3 months for 1–2 years. Forty-nine of the 83 lupus patients either had a history of nephritis or had active nephritis for which renal biopsy was necessitated during the study. Twenty-three of the patients with nephritis had active nephritis (hematuria, increasing proteinuria, or increasing serum creatinine level) necessitating biopsy. For each of the patients with active nephritis, one of the study visits occurred during the biopsy admission or biopsy outpatient observation period and before treatment had been initiated or within 48 hours after dosage increase (prednisone equivalent >10 mg daily or steroid-sparing therapy). Making determinations prior to institution of therapy was a unique feature of this study and reduced the confounding effects of treatments such as corticosteroids and mycophenolate mofetil, both of which can reduce iNOS transcription (22,23). Demographic and clinical characteristics of the SLE patients and control subjects are presented in Table 1. The SLE group and the control group did not differ significantly in terms of any demographic feature except for a higher proportion of men in the control group. Demographic characteristics of the patients with lupus nephritis were reflective of those of the SLE group as a whole.

Table 1.

Characteristics of the SLE patients and control subjects^*

		SLE patients (n = 83)	Control (n = 40)	P
Age, years†		37 ± 1	41 ± 2	0.09
Female, %		95	82	0.03
Race, %
	African American	81	65	0.1‡
	White	19	30
	Hispanic	0	2.5
	Asian	0	2.5
Study visits completed, no.
	1	83	40	<0.001‡
	2	62	13
	3	52	0
	4	45	0
	5	40	0
	6	19	0
	7	15	0
	8	13	0
	9	10	0
Nephritis, no. (mean no. of visits)		49 (3)	0	NA
Active nephritis, no. (mean no. of visits)		23 (2.5)	0	NA
SLEDAI score†		7.6 ± 0.7	0	NA
Daily equivalent prednisone dosage, mg†		11 ± 2	0	NA
Serum C3, mg/dl†		88 ± 4	116 ± 3	<0.001

^*Except where indicated otherwise, values are the mean ± SEM. SLE = systemic lupus erythematosus; NA = not applicable; SLEDAI = SLE Disease Activity Index.

^†Data from the first visit.

^‡P for distribution.

Serum NOx levels in the lupus, lupus nephritis, and control groups

Serum NOx levels were determined in all subjects during the first visit, as a surrogate measure of systemic NO production at entry into the study. Serum NOx levels were compared between controls and the following SLE groups: 1) all SLE patients, 2) SLE patients without a history of nephritis, 3) SLE patients with a history of nephritis but without active nephritis during the period of the study, and 4) SLE patients with active nephritis necessitating biopsy during the study. As shown in Figure 1, serum NOx levels were significantly higher in all lupus groups than in controls. Furthermore, serum NOx levels were significantly higher in the group with active lupus nephritis than in all other groups except those with a history of lupus nephritis (Figure 1).

Figure 1.

Serum nitrate plus nitrite (NOx) levels at the first visit in control subjects (n = 40), in the overall group of systemic lupus erythematosus (SLE) patients (n = 83), and in the subgroups of SLE patients with no history of nephritis (sine nephritis; n = 34), with a history of nephritis but without active disease necessitating renal biopsy (history nephritis; n = 26), and with active nephritis (increasing proteinuria, increasing serum creatinine level, or hematuria) necessitating renal biopsy at the time of a study visit (n = 23). Values are the mean and SEM. * = P < 0.05 versus controls; ** = P < 0.05 versus the active nephritis group.

Cross-sectional analysis of the association between serum NOx levels and lupus activity variables

To determine the relationship between serum NOx levels and clinical disease activity variables in a cross-sectional analysis, bivariate correlations between variables at the first visit were determined for all SLE patients. Serum NOx level was the input variable, while serum creatinine, C3, C4, and anti-dsDNA levels, urinary protein: creatinine ratio, ESR, platelet count, white blood cell count, and daily equivalent prednisone dose were the output variables. Of these variables, serum NOx levels correlated positively (P ≤ 0.05) with serum creatinine level, anti-dsDNA level, and urine protein:creatinine ratio and negatively with serum C3 level (Table 2). These findings indicate a significant association between serum NOx levels and standard laboratory measures of disease activity and confirm findings in prior studies performed by our group and others.

Table 2.

P values for the correlations between disease activity variables in SLE patients at the first visit^*

	Serum Cr	SLEDAI	C3	Anti-dsDNA	C4	ESR	Urinary protein/Cr	Platelet count	Daily prednisone dosage
Serum NOx	<0.001	NS	0.01	0.04	NS	NS	<0.001	NS	NS
Serum Cr	NS	NS	0.02	NS	NS	0.05	<0.001	NS	NS
SLEDAI	NS	NS	<0.001	<0.001	<0.001	<0.001	<0.001	NS	0.03
Anti-dsDNA	NS	NS	NS	NS	<0.001	NS	<0.001	NS	0.02
C4	NS	NS	<0.001	NS	NS	NS	NS	NS	NS
Urinary protein/Cr	NS	NS	NS	NS	<0.001	<0.001	NS	NS	NS

^*SLE = systemic lupus erythematosus; Cr = creatinine; SLEDAI = SLE Disease Activity Index; anti-dsDNA = anti-double–stranded DNA; ESR = erythrocyte sedimentation rate; NOx = nitrate plus nitrite; NS = not significant.

Association between serum NOx levels and SLEDAI scores over time

Due to the waxing and waning nature of SLE disease activity, there is no single universally accepted consensus method for measuring disease activity or response to therapy. Therefore, we used 2 different approaches for comparing SLEDAI scores and serum NOx levels. First, SLEDAI scores and serum NOx levels at each visit were compared using a mixed model analysis. This method tested for the ability of a measure of systemic NO production to reflect simultaneous disease activity. By this method, SLEDAI scores were found to be significantly associated with serum NOx levels (β estimate 0.96, P = 0.0005). Using another approach, the AUC SLEDAI was shown to be correlated with the AUC NOx. This approach examined the relationship between global NO production and SLE disease activity burden over time and was designed to detect associations that may have been temporal but not simultaneous. AUC SLEDAI and AUC NOx were significantly associated with one another in this analysis as well (r = 0.52, P < 0.001) (Figure 2).

Figure 2.

Association between serum nitrate plus nitrite (NOx) levels and Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) scores over all visits, in SLE patients with >1 study visit. From the serum NOx levels and SLEDAI scores at inidividual visits, the sum of the area under the curve for NOx (AUC NOx) and the sum of the AUC for the SLEDAI (AUC SLEDAI) were calculated for all visits.

Longitudinal multivariate analysis of associations between serum NOx levels and lupus disease activity

To determine whether the above associations observed in a cross-sectional analysis were also present in individual lupus patients over time, a mixed model was created, using all of the continuous SLEDAI elements as input variables (17). This analysis was designed to determine which SLE disease activity measures were more strongly associated with NO production. In the first model, all patients were included. Variables significantly associated with NOx levels in this model were the diagnosis of SLE (P = 0.004), serum C3 level (P = 0.014), serum creatinine level (P < 0.001), and urinary protein:creatinine ratio (P = 0.001). Among SLE patients overall and among the subgroup of patients with lupus nephritis, the association was positive for serum creatinine level (P < 0.0001 for both groups) and urinary protein:creatinine ratio (P = 0.002 and P = 0.03, respectively) and negative for serum C3 level (P = 0.003 and P = 0.04, respectively).

Association between serum NOx levels and renal biopsy WHO class, activity index, and chronicity index

To explore a possible association between systemic NO production and renal inflammation, serum NOx levels were determined during the visit corresponding to renal biopsy in 23 patients. These patients were categorized according to WHO classification of renal biopsy. Serum NOx levels were greatest in patients with class IV disease, followed by class III, class II, and class V disease (Figure 3A). Levels among those with proliferative disease (classes III and IV; n = 11 and n = 6, respectively) were significantly greater than levels among those with mesangial and membranous disease (classes II and V; n = 2 and n = 4, respectively) (P = 0.04). This measure of systemic NO production appeared to reflect the extent of proliferative or inflammatory disease that the WHO classifications represent. To further explore this notion, NIH activity and chronicity indices were compared with serum NOx levels (Figure 3B). Activity and chronicity indices were both found to be associated with serum NOx levels (r = 0.34, P = 0.1 and r = 0.63, P = 0.001, respectively), although statistical significance was achieved only with chronicity scores. The stronger association with chronicity scores may reflect the longitudinal association between NO production and renal damage (see below).

Figure 3.

Association between serum nitrate plus nitrite (NOx) levels and proliferative glomerulonephritis seen on renal biopsy. A, Serum NOx levels in patients with active lupus nephritis grouped by World Health Organization (WHO) classification of the findings on renal biopsy performed at the time of the serum collection. Values are the mean and SEM (n = 2 patients in WHO class II, 11 patients in class III, 6 patients in class IV, and 4 patients in class V). * = P = 0.04. Differences between individual groups were not statistically significant. B, Correlations between serum NOx levels and renal biopsy activity and chronicity index scores in the 23 patients with active lupus nephritis. Open circles depict activity index scores (r = 0.34, P = 0.1); solid circles depict chronicity index scores (r = 0.63, P = 0.001).

Association between measures of longitudinal exposure to NO and renal damage in lupus nephritis patients

The presence of elevated serum NOx levels among patients with proliferative glomerulonephritis and low complement levels raises the possibility that increased glomerular iNOS activity may lead to renal damage, as seen in murine models of lupus (24). To explore this notion to the extent possible with observational data, we analyzed the association between markers of systemic NO production and progression of renal damage over time. Selecting only for data from lupus nephritis patients, AUC NOx was found to be correlated with the relative change in creatinine level from the first to the last visit among those with >1 visit (n = 33). AUC NOx correlated significantly with ΔCr (r = 0.43, P = 0.01). Similar results were obtained when changes in the calculated glomerular filtration rate were substituted for ΔCr.

Because both AUC NOx and ΔCr were affected by the duration of followup, a backward regression analysis was performed with AUC NOx and time of followup as input variables and ΔCr as the output variable. AUC NOx still showed a significant association with ΔCr (P = 0.05), while followup time had a weaker association that did not reach statistical significance (P = 0.08). However, because there was a weak association between ΔCr and followup time and because variations in the duration of followup could bias the results (compliance with presentation for study visits every 90 days was 60%), a third analysis was performed to remove any confounding effects of followup time. In this analysis, AUC NOx and ΔCr were both normalized to time of followup, and the correlation analysis was repeated. The association between AUC NOx/followup time and ΔCr/followup time was found to be even greater than that observed with data that were not normalized to followup time (r = 0.64, P < 0.001).

To determine the association between serum NOx levels and response to therapy for lupus nephritis, patients with active lupus nephritis were divided into responders (ΔCr <25%) and nonresponders (ΔCr ≥25%). Serum NOx values were calculated for each visit and plotted (Figure 4). Among nonresponders, NOx levels were significantly higher than those in responders at the second visit (P = 0.016). This finding provides further support for the notion that in order to be effective in preventing renal damage, therapy for active lupus nephritis must reduce RONI production. When specific medications in the 2 groups were used as variables in a chi-square analysis, no significant differences due to differences in the treatment regimens between responders and nonresponders were found (responders mycophenolate mofetil n = 5, cyclophosphamide n = 2, prednisone alone n = 0, azathioprine n = 4; nonresponders mycophenolate mofetil n = 1, cyclophosphamide n = 2, prednisone alone n = 2, azathioprine n = 0) (P = 0.16).

Figure 4.

Serum nitrate plus nitrite (NOx) levels over time in patients with active lupus nephritis, according to response to therapy. Responders were defined as those with a change in serum creatinine level of <25% (n = 13); nonresponders were defined as those with a change in creatinine level of ≥25% (n = 4). Values are the mean ± SEM. * = P = 0.02 versus responders.

Lack of relationship between renal NOx clearance and serum NOx levels

The above data strongly suggested that systemic NO production is affected by lupus disease activity, particularly, complement activation, particularly among patients with lupus nephritis. The association between serum NOx levels and renal damage over time could be due to a pathogenic effect of reactive intermediate production and/or to reduced renal clearance of NOx. To address the second possibility, NOx levels were determined in urine samples obtained at a subset of visits from SLE patients with active nephritis (n = 22) (randomly selected for low, intermediate, and high serum NOx levels). Urine creatinine and simultaneous serum creatinine levels were also determined. The fractional excretion of NOx was calculated as (urine NOx/urine creatinine)/(serum NOx/serum creatinine). A backward multiple regression analysis was performed with serum C3 and fractional excretion of NOx as input variables and serum NOx as the output variable. In the first model, there was a significant association between C3 and serum NOx levels (r = −0.46, P = 0.04) but not between fractional excretion of NOx and serum NOx levels (r = −0.016, P = 0.6). Similar results were obtained when the relationship between serum NOx levels and urine NOx/urine creatinine was investigated. Thus, elevations in serum NOx levels were associated with markers of complement activation but not with the degree of renal excretion of NOx. This indicated that reduced renal clearance of NOx was not a dominant factor in the association between serum NOx levels and renal damage.

DISCUSSION

The results presented herein demonstrate significant associations between a marker of systemic NO production and measures of disease activity in SLE. There were significant increases in serum NOx levels observed during complement activation, particularly in the setting of proliferative glomerulonephritis. Global NO burden over time (AUC NOx) was associated with progression of renal damage, and serum NOx levels mirrored response to therapy. These findings suggest a pathogenic role of RONI production in human proliferative glomerulonephritis, similar to findings in our prior study of murine lupus nephritis (24,25).

The novelty of this study lies in: 1) its prospective and longitudinal design, 2) reduction of the confounding effects of exogenous dietary NOx through the use of a low-NOx diet prior to sample collection, 3) study of a large number of patients, a significant proportion of whom were African American, and 4) reduction of the confounding effect of high-dose corticosteroids on iNOS expression during renal biopsy visits.

A previous prospective study did not demonstrate an association between lupus nephritis and serum NOx levels. However, in that study, higher corticosteroid doses were allowed, and simultaneous renal biopsy results were not available (5). Other groups have demonstrated in cross-sectional analyses that patients with active SLE, low complement levels, and high anti-dsDNA levels had elevated levels of serum NOx (10,12). The longitudinal design of the present study enabled the analysis of within-patient associations between disease activity markers and serum NOx levels, thus avoiding between-patient variability that is inherent in any lupus cohort and that may have reduced the power of previous studies. This is the first study to demonstrate an association between serum NOx levels and proteinuria in humans. This correlation is consistent with findings in studies of murine lupus nephritis, in which overproduction of NO has been shown to parallel, and even precede, the onset of proteinuria (24,25).

The longitudinal association of systemic NO production with response to therapy and with biopsy-proven proliferative glomerulonephritis is a novel finding that has implications with regard to therapy. Our study population was predominantly African American, while the study cohorts in prior investigations of NO measures in SLE were mostly white or Hispanic. Thus, it is intriguing to postulate that one potential mechanism for the higher rate of renal failure and poorer response to therapy in African Americans with proliferative glomerulonephritis compared with whites (26) is increased RONI production.

In an earlier investigation of the same study cohort, we observed increases in markers of production of systemic peroxynitrite (ONOO, a product of iNOS activity), which were more closely associated with disease activity among African Americans than among whites (27). We have also described an increased prevalence of 2 iNOS gene (NOS2) promoter polymorphisms in African American women with lupus compared with sex- and race-matched controls (28). These polymorphisms were not present in patients of European descent and have been reported to be frequent in malaria-endemic areas (29,30). One has been associated with increased ex vivo leukocyte NO production (31). Together, these findings suggest the possibility of differential genetic control of NOS2 transcription under inflammatory conditions among African Americans. One confounder in the association between serum NOx levels and treatment response is that the treatment was random and not controlled. Future studies investigating this association should involve subjects participating in large clinical trials, to reduce the number of treatment variables.

While there was a very strong association between serum NOx levels and SLEDAI scores in the bivariate longitudinal analysis, the weaker (though still significant) association in multivariate longitudinal analysis may be due to several factors. First, the clinical variables used in the mixed models were in fact elements from the SLEDAI. Inclusion of colinear variables into multivariate models often results in loss of significance of one or more of those variables. The strong association between SLEDAI scores and serum NOx levels in the bivariate analysis supports this interpretation. Second, there is a clear association between reduced complement levels and NO production, yet not all of the elements of the SLEDAI are associated with significant complement activation. The association with proliferative renal disease, which is dependent on complement activation, is consistent with this notion. However, the association with renal disease may not have been captured by the SLEDAI as well as it would have been with, for instance, the British Isles Lupus Assessment Group index (BILAG), due to lack of sensitivity of the SLEDAI to changes in renal disease activity. However, the BILAG measure was not available at the initiation of this trial and therefore could not be used for longitudinal followup (32).

The association between progression of renal insufficiency and elevated NOx levels over time may result from two general mechanisms. First, reduced renal excretion of NOx may lead to increased serum levels. The demonstration of a lack of association between renal excretion of NOx and serum NOx levels, with a clear association between serum NOx and C3 levels in the same patients, makes it unlikely that decreased NOx clearance is the dominant cause of elevated serum NOx levels in this population. Second, the association between global burden of NOx production over time (AUC NOx) and future renal damage (ΔCr) suggests that glomerular RONI production is a pathogenic mediator in glomerular proliferation and sclerosis. This is further supported by data demonstrating that serum NOx levels did not decrease as quickly in patients whose lupus nephritis did not respond to treatment, compared with responders. Because not all study visits occurred within 4 weeks of the predetermined 90-day visit interval, the AUC analyses could be subject to a time bias. While this bias should be considered, the more significant association between AUC NOx and ΔCr when data were normalized to time suggest that the results obtained were not predominantly a reflection of time bias.

Based on the results of this and previous studies, we propose the following model for potential causes and consequences of pathogenic RONI production in complement-mediated disease manifestations of SLE: In the setting of immune complex deposition and complement activation, complement split products C3a and C5a bind to their receptors. Engagement of C5a and C3a receptors results in elevated transcription of NOS2, increased expression of iNOS, and logarithmic increases in production of RONI such as superoxide and ONOO (33). Evidence of this mechanism has been provided in models of anti–Thy-1–induced nephritis, in which soluble CR1 reduces NOS2 messenger RNA and iNOS protein levels, and in intestinal ischemia-reperfusion models, in which an anti-C5 monoclonal antibody reduces RONI production (34,35). RONI are known mediators of apoptosis (36). Markers of increased NO production have been observed in cells activated by immune complex deposition and complement activation in lupus, i.e., vascular endothelial cells (8) and glomerular and tubulointerstitial cells (7).

In studies of lupus patients, increased apoptosis was noted among those with proliferative glomerulonephritis in association with increased iNOS expression (7,37), and in circulating endothelial cells without evidence of increased iNOS expression (38). However, findings of other studies have suggested that iNOS expression induces apoptosis in endothelial cells. In human umbilical vein endothelial cells, ROI induce NF-κB–mediated increases in iNOS expression and apoptosis (39). From these observations, one could hypothesize that RONI play a role in SLE endothelial cell apoptosis as well. ROI can depolymerize glomerular basement membrane heparan sulfate moieties, resulting in loss of the glomerular basement membrane charge barrier and increased excretion of protein into the urine (40). ONOO is capable of altering the activity of multiple enzymes thought to play regulatory roles in glomerular function, such as cyclooxygenase (41), prostacyclin synthase (42), and endothelial NOS (43). ROI are capable of increasing DNA binding of the redox-sensitive transcriptional regulator NF-κB (44). This can result in an increased inflammatory response that, in an adoptive transfer model of autoimmune diabetes, can be inhibited by a superoxide dismutase mimetic (45). Finally, ONOO can increase the antigenicity of dsDNA (46), resulting in a feed-forward loop that can accelerate the initiation and progression of proliferative glomerulonephritis.

The results presented herein suggest that increased RONI production is a pathogenic mediator in complement-mediated disease processes in SLE, particularly proliferative lupus nephritis. This has potential implications regarding the management of and/or monitoring of response to therapy of aggressive proliferative glomerulonephritis. The full effect of reactive intermediates on glomerular pathology cannot be determined without specifically targeting their production in humans. Therefore, future work will evaluate the effect of available reactive intermediate–reducing therapies on markers of iNOS activity relative to treatment response. In addition, novel therapies for inhibiting iNOS activity or scavenging RONI should be studied as adjunctive induction treatment in proliferative lupus nephritis.