Urinary Vascular Cell Adhesion Molecule, But Not Neutrophil Gelatinase-associated Lipocalin, Is Associated with Lupus Nephritis

Abstract

Objective

Vascular cell adhesion molecule-1 (VCAM-1), an adhesion molecule, is involved in the progression of glomerular and tubulointerstitial injury. Neutrophil gelatinase-associated lipocalin (NGAL), a member of the lipocalin superfamily, has been shown to rise in both acute and chronic kidney damage. Both VCAM-1 and NGAL have been found at high levels in the urine of patients with active lupus nephritis. We investigated both as potential biomarkers for lupus nephritis.

Methods

VCAM-1 and NGAL were measured by ELISA during 1 to 8 clinic visits in 107 patients with systemic lupus erythematosus (SLE; 91% women, 51% black, 36% white, 4% Asian, 4% Hispanic, and 5% others) for a total of 190 visits. Patients’ mean age was 41 years. We analyzed the relationship between these potential urine biomarkers and the urine protein/creatinine ratio (urine Pr/Cr), the Systemic Lupus International Collaborating Clinics (SLICC) renal activity score, SLE Disease Activity Index renal descriptors, and other clinical variables.

Results

VCAM-1 levels were strongly associated with the physician’s global estimate of disease activity (p = 0.0002), the renal visual analog scale (p < 0.0001), the urine Pr/Cr (p < 0.0001), and SLICC renal activity score (p < 0.0001). VCAM-1 levels were also associated with a urine Pr/Cr ≥ 0.5 (p < 0.0001). NGAL was not associated with any measure of disease activity or with lupus serologies.

Conclusion

Urine VCAM-1 had a strong association with measures of disease activity, including multiple renal activity descriptors. In contrast to previous SLE studies, NGAL failed to show any association with lupus nephritis.

Lupus nephritis (LN) is a common manifestation among patients with systemic lupus erythematosus (SLE), occurring in over 50%. One major determinant of poor prognosis among patients with SLE is renal involvement, LN being both more common and more severe in non-white patients^1,2,3. Despite new immunosuppressive agents available for treatment, no significant improvement has been seen in survival in patients with SLE or in the incidence of endstage renal disease from LN^4,5,6.

Renal biopsy remains the “gold standard” to determine SLE renal activity⁷, but carries its own risks. Neither clinical nor serological data can accurately predict the histopathological lesions of LN. There remains a need for noninvasive biomarkers that associate with renal activity and are able to predict renal flares.

Studies of murine lupus have shown the importance of adhesion molecules, cytokines, and chemokines in the pathogenesis of glomerular inflammation^8,9. However, to be clinically practical, biomarkers of LN would need to be measured in the urine, since urine collection is much easier and less invasive compared to repetitive renal biopsies.

Neutrophil gelatinase-associated lipocalin (NGAL; lipocalin-2) is a member of the lipocalin family and is extensively studied in both animals and humans^{10,11,12,13,14}. In a murine model for human SLE (chronic graft-vs-host disease), increased levels of NGAL were found after induction of disease¹¹. Increased levels of NGAL have also been found in other inflammatory conditions, including plasma of patients with vasculitis and Kawasaki syndrome^15,16. Cancers, including breast, colorectal, pancreatic, and ovarian tumors, have shown increased NGAL expression as well^17,18,19,20.

Brunner, et al have shown that urinary NGAL levels correlated with renal disease activity in pediatric lupus¹². Pitashny, et al found that urinary NGAL levels were higher in patients with LN, correlating with both the renal SLE Disease Activity Index (SLEDAI) descriptors and the urine protein to creatinine ratio (urine Pr/Cr)¹³. However, urinary NGAL is not specific for LN, being found in multiple types of acute renal injury^21,22,23,24.

Vascular cell adhesion molecule-1 (VCAM-1) is a member of the immunoglobulin superfamily that is highly expressed in kidneys, both in murine SLE models and in human LN^{25,26,27,28,29}. VCAM-1 is also expressed in normal kidneys, but levels increase many-fold in diseased states^29,30,31. High VCAM-1 expression was found in the endothelium of renal transplant rejection biopsy specimens. Antibodies directed against VCAM-1 prolonged allograft survival³². In a study of 30 SLE patients with LN, 20 SLE patients without LN, and 20 controls, high urinary VCAM-1 levels were seen in patients with Class III, IV, and V disease compared to those without LN and controls³³. In another study, elevated urinary levels of VCAM-1 were seen in 38 SLE patients with varying SLEDAI scores (0 to 18) versus patients with rheumatoid arthritis and healthy controls. These levels correlated with SLEDAI scores, especially renal SLEDAI descriptors, and the urine Pr/Cr ratio, and were higher in patients with active LN versus those with no renal disease²⁵. Serum levels of VCAM-1 have also been shown to be associated with SLE disease activity^26,34.

Our primary goal was to relate the levels of VCAM-1 and NGAL with levels of proteinuria. We also evaluated these 2 promising urinary biomarkers for LN prospectively in a longitudinal study of SLE patients with and without LN and correlated them with renal pathology activity and chronicity indices.

MATERIALS AND METHODS

Our study was approved by the Johns Hopkins University School of Medicine Institutional Review Board. All patients gave informed consent. The patients were part of the Hopkins Lupus Cohort, a prospective longitudinal study of lupus activity and outcomes. Consecutive patients with biopsy-proven LN or proteinuria were invited to participate, as were a subset of consecutive patients without these conditions. The cohort protocol required at least quarterly visits, but patients were seen more often depending on disease activity. At each visit, disease activity indices — the physician’s global assessment, Safety of Estrogens in Lupus Erythematosus: National Assessment (SELE-NA)-SLEDAI³⁵, and renal activity on a 0 to 3 visual analog scale (VAS)³⁶ —were calculated by one rheumatologist. Vital signs and laboratory tests were recorded: weight, blood pressure, complete blood count, platelets, erythrocyte sedimentation rate, creatinine, cholesterol, C3, C4, anti-dsDNA, anticardiolipin, lupus anticoagulant by Russell’s viper venom time test confirmation, urinalysis, urine Pr/Cr ratio; as well as treatment [prednisone, immunosuppressives, hydroxychloroquine, angiotensin-converting enzyme (ACE) inhibitor and dose, angiotensin receptor blocker and dose]. Urine samples were collected at each visit.

Quantifying renal disease activity

We used the Systemic Lupus International Collaborating Clinics (SLICC) Renal Activity Score to quantify renal disease activity³⁷. This score was derived from a regression analysis using the physician rating of renal activity as the “gold standard.” It is calculated as follows: proteinuria 0.5 to 1 g/day (3 points), proteinuria 1 to 3 g/day (5 points), proteinuria > 3 g/day (11 points), urine red blood cells ≥ 5/high power field (hpf; 3 points), and urine white blood cells (WBC) ≥ 5/hpf (1 point).

Urine collection, preparation, and biomarker measurement

Urine “clean-catch midstream” samples of 25–50 ml were collected into appropriate sterile containers and a protease inhibitor was added. Samples were placed on ice or refrigerated at 4°C within 1 h of collection. Aliquots and the remaining urine sample in the centrifuge tube were frozen at −80°C and batched for shipment to the University of Texas Southwestern Medical Center laboratory (Dallas, TX, USA). VCAM-1 and NGAL were measured by ELISA (R&D Systems, Minneapolis, MN, USA). The results were reported as pg/ml (VCAM-1) and ng/ml (NGAL).

Statistical methods

We normalized our measures of urinary VCAM-1 and NGAL by dividing by the urine creatinine concentration to adjust for the variable urine concentration. Then we transformed each variable by taking the log_e(variable + 1) prior to calculating means and performing inference. This was done because distributions of NGAL and VCAM-1 were skewed. We used a mixed-effects model to measure the associations between biomarkers and clinical variables, including a random effect for the patient, to account for the correlation between repeated observations from the same patient. For difference between subgroups (low, medium, high) with respect to the proportion of visits with urine Pr/Cr ratio ≥ 0.5, a General Estimating Equations method was used to calculate p values accounting for repeated observations from the same patient. To assess the association between biomarkers and biopsy results, we chose the closest visit date to the biopsy date when the US National Institutes of Health (NIH) renal activity index (0–24) and chronicity indices (0–12) were calculated. We then used linear regression models to examine associations between these indices and clinical predictors.

RESULTS

Data were obtained from 107 patients with SLE (91% women). VCAM-1 and NGAL were measured at 1 to 8 visits per patient over the course of a 23-month period for a total of 190 visits. Twenty-four percent had 2 visits, 21% had ≥ 3, and 1% had 5, 6, and 8 visits each. Of 45% of patients with followup visits, 11% had disease flares.

Patients were white (36%), black (51%), Hispanic (4%), Asian (4%), and other ethnicity (5%). The mean age was 41 years. Cumulative clinical manifestations included malar rash (51%), discoid rash (26%), photosensitivity (47%), oral ulcers (38%), arthritis (73%), serositis (54%), renal disorder [83%; 78% had LN based on International Society of Nephrology (ISN) biopsy class, 77% based on proteinuria, and 75% based on both biopsy and proteinuria], neurological disorder (11%), immunologic disorder (93%), and positivity for antinuclear antibodies (99%).

Table 1 shows the association between the biomarkers and various clinical conditions observed at each visit. VCAM-1 levels were strongly associated with the physician’s global estimate of disease activity and the renal VAS. VCAM-1 was also associated with the urine Pr/Cr ratio (p < 0.0001) and with the SLICC Renal Activity Score (p < 0.0001). SELENA-SLEDAI results (Table 1) showed no association of NGAL or VCAM-1 with SLEDAI when analyzed categorically. However, when we analyzed SLEDAI as a continuous variable, only VCAM-1 had a borderline association (p = 0.0544). NGAL was not associated with any measure of disease activity or with lupus serologies (Table 1). There was no strong relationship between the biomarkers and diabetes mellitus, current use of ACE inhibitor, or use of an angiotensin receptor blocker (Table 1). When we redid the analyses after excluding patients with diabetes mellitus and those receiving ACE inhibitor or angiotensin receptor blocker, all the results were qualitatively similar to those reported above.

Table 1.

Mean log-transformed and normalized (by urine creatinine) VCAM-1 and NGAL, by clinical variables at each visit.

Clinical Variables at Each Visit	VCAM-1		NGAL
	Mean (SD)	p*	Mean (SD)	p*
Age, yrs
21–44 (n = 115)	0.21 (0.40)	0.07	0.17 (0.40)	0.43
45–70 (n = 75)	0.11 (0.17)		0.12 (0.17)
Sex
Female (n = 175)	0.17 (0.34)	0.79	0.16 (0.31)	0.31
Male (n = 15)	0.14 (0.15)		0.07 (0.15)
Ethnicity
White (n = 62)	0.15 (0.20)	0.66	0.11 (0.16)	0.46
African American (n = 108)	0.17 (0.39)		0.18 (0.37)
Other (n = 20)	0.23 (0.34)		0.11 (0.13)
Hematuria
Present (n = 20)	0.20 (0.14)	0.73	0.13 (0.12)	0.80
Absent (n = 170)	0.17 (0.35)		0.15 (0.31)
Proteinuria
Present (n = 28)	0.27 (0.29)	0.15	0.13 (0.14)	0.48
Absent (n = 162)	0.15 (0.33)		0.15 (0.32)
Pyuria
Present (n = 8)	0.20 (0.13)	0.76	0.18 (0.14)	0.74
Absent (n = 182)	0.17 (0.34)		0.15 (0.30)
Anti-dsDNA
Present (n = 64)	0.18 (0.20)	0.75	0.13 (0.12)	0.44
Absent (n = 126)	0.17 (0.38)		0.16 (0.36)
Low C3 or C4
Present (n = 84)	0.20 (0.25)	0.46	0.12 (0.13)	0.21
Absent (n = 105)	0.15 (0.38)		0.17 (0.38)
Leukopenia
Present (n = 5)	0.22 (0.22)	0.64	0.18 (0.10)	0.82
Absent (n = 185)	0.17 (0.33)		0.15 (0.30)
Urine protein/creatinine ratio
≥ 0.5 (n = 76)	0.26 (0.24)	< 0.0001	0.14 (0.15)	0.34
< 0.5 (n = 106)	0.07 (0.10)		0.13 (0.17)
Renal failure
Ever (n = 15)	0.11 (0.15)	0.57	0.15 (0.17)	0.95
Never (n = 175)	0.18 (0.34)		0.15 (0.31)
Hydroxychloroquine
Yes (n = 146)	0.17 (0.36)	0.97	0.16 (0.33)	0.48
No (n = 44)	0.18 (0.23)		0.11 (0.12)
Prednisone
Yes (n = 155)	0.19 (0.36)	0.068	0.15 (0.32)	0.78
No (n = 35)	0.07 (0.07)		0.13 (0.17)
Mycophenolate mofetil
Yes (n = 99)	0.16 (0.17)	0.72	0.12 (0.14)	0.24
No (n = 91)	0.18 (0.45)		0.18 (0.41)
Azathioprine
Yes (n = 16)	0.45 (0.93)	0.0017	0.36 (0.89)	0.014
No (n = 174)	0.15 (0.20)		0.13 (0.16)
Diabetes mellitus
Present (n = 20)	0.08 (0.08)	0.30	0.17 (0.22)	0.69
Absent (n = 170)	0.18 (0.34)		0.15 (0.31)
Use of ACE/ARB inhibitor
Yes (n = 132)	0.14 (0.15)	0.0488	0.13 (0.16)	0.16
No (n = 58)	0.24 (0.55)		0.20 (0.48)
SLICC Renal Activity Score
≥ 4 (n = 64)	0.26 (0.25)	< 0.0001	0.14 (0.14)	0.17
< 4 (n = 107)	0.09 (0.13)		0.12 (0.15)
SLEDAI
≥ 4 (n = 53)	0.21 (0.24)	0.30	0.13 (0.13)	0.60
< 4 (n = 137)	0.16 (0.36)		0.16 (0.34)
Physician’s Global Assessment
≥ 1.5 (n = 89)	0.27 (0.45)	0.0002	0.18 (0.39)	0.23
< 1.5 (n = 101)	0.08 (0.11)		0.12 (0.17)
Renal activity (VAS)
≥ 1.5 (n = 75)	0.31 (0.47)	< 0.0001	0.19 (0.42)	0.19
< 1.5 (n = 115)	0.08 (0.10)		0.12 (0.16)

^*p values are based on a mixed-effects model because some patients contributed multiple observations.

VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin; ACE: angiotensin-converting enzyme; ARB: angiotensin receptor blocker; SLICC: Systemic Lupus International Collaborating Clinics; SLEDAI: Systemic Lupus Erythematosus Disease Activity Index; VAS: visual analog scale.

Table 2 shows relationships between the biomarkers and urine Pr/Cr > 0.5. Among those with medium or high urinary VCAM-1, 63% and 87%, respectively, had urine Pr/Cr ≥ 0.5, compared to only 15% of those with low VCAM-1 (p < 0.0001).

Table 2.

Proportion (%) of patient visits with urine protein/creatinine ratio ≥ 0.5 in subgroups defined by level of VCAM-1 and NGAL (normalized to urine creatinine).

Subgroups Defined by Biomarker Levels at Each Visit	Proportion (%) of Visits with Urine Protein/Creatinine Ratio ≥ 0.5	p*
VCAM-1**
Low	15/97 (15)	< 0.0001
Medium	34/54 (63)
High	27/31 (87)
NGAL**
Low	43/104 (41)	0.2810
Medium	18/49 (37)
High	15/26 (57)

^*For difference between subgroups (low, medium, high) with respect to the proportion of visits with urine protein/creatinine ratio ≥ 0.5, calculated using a GEE approach that accounts for repeated measures from the same patients.

^**Low: ≤ 0.1; medium: 0.1–0.3; high: ≥ 0.3 for VCAM-1 and NGAL.

VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin.

Table 3 shows the sensitivity, specificity, and positive and negative predictive values of various rules, based on the biomarkers, for identifying patients with high levels of renal activity (i.e., Renal Activity Score ≥ 4).

Table 3.

Sensitivity, specificity, and positive and negative predictive values of various rules for identifying patients with a SLICC renal activity score of 4 points or more based on VCAM-1 and NGAL.

Definition of a “Positive” Test	Sensitivity, %	Specificity, %	Positive Predictive Value, %	Negative Predictive Value, %
VCAM-1/creatinine ratio
> 0.1	79	73	66	84
> 0.2	54	88	75	74
> 0.3	34	93	75	68
NGAL/creatinine ratio
> 0.1	45	57	43	61
> 0.2	28	75	45	61
> 0.3	20	86	54	62

VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin; SLICC: Systemic Lupus International Collaborating Clinics.

Twenty-six patients had a renal biopsy within 3 months of the biomarker measurement. Tables 4 and 5 show the relationship between biomarker values and biopsy results within the last 3 and 6 months, respectively. Our results showed there was a higher percentage of patients with high VCAM-1 levels in Class V disease versus other ISN classes, although the results were not statistically significant.

Table 4.

Analysis of renal biopsies by International Society of Nephrology class within 3 months.

Urine Biomarker	Class II, n (%)	Class III, n (%)	Class IV, n (%)	Class V, n (%)
VCAM-1*
Low (n = 8)	0 (0)	3 (38)	2 (25)	3 (38)
Medium (n = 7)	2 (29)	3 (43)	2 (29)	4 (57)
High (n = 11)	0 (0)	3 (27)	3 (27)	9 (82)
NGAL*
Low (n = 15)	2 (13)	5 (33)	4 (27)	9 (60)
Medium (n = 8)	0 (0)	3 (38)	2 (25)	5 (63)
High (n = 3)	0 (0)	1 (33)	1 (33)	2 (67)

^*Low: ≤ 0.1; medium: 0.1–0.3; high: ≥ 0.3 for VCAM-1 and NGAL.

VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin.

Table 5.

Analysis of renal biopsies by International Society of Nephrology (ISN) class within 6 months.

Urine Biomarker	Class II, III, IV n (%)	p	Class V, n (%)	p
VCAM-1*
Low (n = 12)	8 (67)	> 0.99	6 (50)	0.12
High (n = 21)	14 (67)		16 (76)
NGAL*
Low (n = 16)	12 (75)	0.45	10 (63)	0.71
High (n = 16)	10 (63)		11 (69)

^*Low: ≤ 0.1; high: ≥ 0.1 for VCAM-1 and NGAL.

VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin.

Table 6 shows the association between the renal biopsy NIH activity and chronicity indices and clinical renal predictors. The NIH activity index was associated with the SLICC renal activity score (p = 0.005), 24-hour urine protein (p < 0.0001), and urine WBC count (p < 0.0006), whereas the NIH chronicity index score was associated with serum creatinine (p < 0.0001). There was no correlation between NGAL and VCAM-1 and the renal biopsy activity index (p = 0.50 and p = 0.25) or chronicity index (p = 0.14 and p = 0.23).

Table 6.

Association between renal biopsy National Institutes of Health (NIH) activity and chronicity indices and clinical renal predictors.

Variable	Estimated Effect on Mean Renal Biopsy NIH Activity Index per 1 Unit of Variable	p
Urinalysis protein	1.366 ± 0.768	0.08
Urinalysis red blood cells/hpf	0.144 ± 0.075	0.06
Urinalysis white blood cells/hpf	0.191 ± 0.056	0.0006
Urine creatinine level	0.016 ± 0.012	0.18
Urine protein/creatinine ratio	0.238 ± 0.355	0.50
24-h protein	4.020 ± 0.345	< 0.0001
Serum creatinine	−0.231 ± 1.697	0.47
SLICC renal activity score	0.437 ± 0.156	0.005
Log-normalized VCAM	−1.329 ± 1.154	0.25
Log-normalized NGAL	−0.878 ± 1.316	0.50
	Estimated Effect on Mean Renal Biopsy NIH Chronicity Index per 1 Unit of Variable	p
Urinalysis protein	−0.361 ± 0.357	0.31
Urinalysis red blood cells/hpf	0.017 ± 0.038	0.66
Urinalysis white blood cells/hpf	0.009 ± 0.026	0.71
Urine creatinine level	−0.003 ± 0.005	0.60
Urineprotein/creatinine ratio	−0.172 ± 0.157	0.28
24-h protein	−1.043 ± 0.603	0.084
Serum creatinine	2.511 ± 0.549	< 0.0001
SLICC renal activity score	−0.036 ± 0.083	0.66
Log-normalized VCAM	−0.627 ± 0.518	0.23
Log-normalized NGAL	−0.845 ± 0.577	0.14

SLICC: Systemic Lupus International Collaborating Clinics; VCAM-1: vascular cell adhesion molecule-1; NGAL: neutrophil gelatinase-associated lipocalin; hpf: high power field.

DISCUSSION

We analyzed urinary levels of VCAM-1 and NGAL and their associations with lupus serologies and markers of disease activity. VCAM-1 is an adhesion molecule that, although weakly expressed in normal kidneys, is increased many-fold in immune-mediated renal disease^29,38. Our study found that an increased urinary VCAM-1 level correlated with the urine Pr/Cr ratio and the SLICC Renal Activity Score, as well as with other measures of disease activity. Similar findings have been reported in other studies^26,34,39. However, our study included a larger patient population with serial sample collection over a longer period, in contrast to previous studies that had smaller patient populations, short followup, and serum sampling as opposed to urine sampling^26,34. In contrast with 2 studies^26,34, we found no association of urinary VCAM-1 with low complement. We did find urinary VCAM-1 to be higher in Class V LN.

Activity and chronicity indices have provided useful information on renal survival in patients with SLE^40,41. In our study the renal biopsy NIH activity index correlated with renal predictors, including the SLICC renal activity score, 24-hour urine protein, and urine WBC counts. As expected, the renal biopsy NIH chronicity index was associated with the serum creatinine findings. Serum creatinine and a high NIH renal biopsy chronicity index have been shown to have independently increased the risk for the development of chronic renal disease or death⁴². In our study there was a strong association between the 2. In contrast, urine levels of VCAM-1 and NGAL did not correlate with renal pathology activity or chronicity indices. However, in a separate unpublished study by our group, we noted that urine VCAM-1 levels correlate significantly with renal pathology activity index when the kidneys are biopsied on the same day the urine is collected.

In our study, in contrast to previous studies, urinary NGAL was not associated with either lupus serologies or any markers of renal disease activity. Two studies that did find an association of urinary NGAL with LN were cross-sectional^12,13. Our study included a larger number of patients with multiple visits. It is possible that the different measures used to assess renal function/injury and the difficulty in reproducibly differentiating cumulative renal damage from acute active disease in the absence of a renal biopsy could explain different results reported from previous studies.

Urinary VCAM-1, but not urinary NGAL, appears to be a promising biomarker for lupus nephritis. Validation of urinary VCAM-1 in combination with other biomarkers, including concurrent renal biopsies, might allow this marker to be used in routine SLE clinical care.