Validating the RAIL Algorithm in Adult Lupus Nephritis Patients

Gaurav Gulati¹, Khalid Abulaban², Jun Ying³, Xiaolan Zhang⁴, Huijuan Song⁵, Qing Ma⁶, Christopher Haffner⁶, Kasha Wiley⁷, Michael Bennett⁸, Brad H. Rovin⁵ and Hermine I. Brunner⁹, ¹Division of Immunology, Allergy and Rheumatology, University of Cincinnati College of Medicine, Cincinnati, OH, ²Pediatric Rheumatology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, ³University of Cincinnati, Cincinnati, OH, ⁴Ohio State University, Columbus, OH, ⁵Ohio State University Medical Center, Columbus, OH, ⁶Cincinnati Children's Hospital and Medical Center, Cincinnati, OH, ⁷Division of Rheumatology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, ⁸Division of Nephrology, Cincinnati Children`s Hospital Medical Center, Cincinnati, OH, ⁹Pediatric Rheumatology Collaborative Study Group, Cincinnati Children's Hospital Medical Center, Cincinnati, OH

Meeting: 2015 ACR/ARHP Annual Meeting

Date of first publication: September 29, 2015

Keywords: lupus nephritis and systemic lupus erythematosus (SLE), Urinary Biomarkers

Session Information

Date: Sunday, November 8, 2015

Title: Systemic Lupus Erythematosus - Clinical Aspects and Treatment Poster Session I

Session Type: ACR Poster Session A

Session Time: 9:00AM-11:00AM

Background/Purpose:

Lupus Nephritis (LN) significantly contributes to morbidity and mortality among Systemic Lupus Erythematosus (SLE) patients. We have established the Renal Activity Index for Lupus (RAIL) in pediatric LN with an accuracy of > 92% in predicting activity. The goal of this study was to validate the RAIL algorithm for adults with LN.

Methods:

Adult SLE patients who required a kidney biopsy as part of their standard care were enrolled. A urine sample was collected at the time of biopsy. Demographic information, SLE status (diagnostic criteria, laboratory and clinical data, renal disease activity (R-SLEDAI) and chronicity (SDI-R)), medications were collected. Kidney biopsies were read in a blinded fashion and the National Institute of Health Activity and Chronicity indices (NIH-AI, NIH-CI) were reported. Patients were dichotomized as high activity or not based on their NIH-AI scores. Fifteen urinary biomarkers (UBMs) were collected and the SIX UBMsused in pediatric RAIL data were used for the analysis (neutrophil gelatinase associated lipocalin, monocyte chemotactic protein 1, ceruloplasmin, adiponectin, hemopexin, kidney injury molecule 1). Urinary microalbumin and creatinine concentrations were measured to normalize for proteinuria in the analysis. Adult RAIL predictors were firstly assessed of associations to AI using univariate analyses, followed by prediction analyses to develop the RAIL algorithms using stepwise multiple logistical regression models. Predictive accuracy of the RAIL algorithm was calculated using the area under the Receiver Operating Characteristic (ROC) curve (AUC).

Results:

We studied 75 adults (mean age 32.4±10.1 years) and the majority were females (78.7%) with equal representation for Caucasian and African American ethnicities (46.7%). Their mean NIH-AI and NIH-CI scores were 4.57± 4.11 and 3.23 ± 2.59, respectively. 11 (15%) patients had high activity with NIH AI scores >10. Age, gender, ethnicity and LN stage were not significantly different between the two activity groups. Using the 6 UBMs used in the pediatric RAIL algorithm, an excellent prediction was obtained (AUC 0.87 (0.77, 0.97)). Using adult population based UBMs with total 10 UBMs there was notable improvement in diagnostic accuracy of the model (AUC 0.92 (0.84, 1.00)). The RAIL algorithms showed little noticeable improvement with adjustment for creatinine and microalbumin [Table 1].

Conclusion:

We have successfully validated the pediatric RAIL algorithm among adults to accurately and reliably measure LN disease activity. Further, in adults we found that adjustment for microalbumin or creatinine levels dose not cause a great change in diagnostic accuracy of RAIL. Using an adult model based RAIL with 10 UBMs does increase the accuracy of the algorithm. Further studies to independently validate and further study the relationship of these UBMs with specific biopsy findings are needed.

TABLE 1: Diagnostic Accuracy of RAILs using Pediatric Model and Adult Model
Statistics	Raw Score (UBMs = 6)	CREATININE adjusted Score (UBMs =6)	Raw Score (UBMs=10)	CREATININE adjusted Score (UBMs=10)	MICROALBUMIN adjusted Score (UBMs=10)
AUC	0.87 (0.77, 0.97)	0.90 (0.83, 0.98)	0.92 (0.84, 1.00)	0.94 (0.88, 0.99)	0.90 (0.83, 0.98)
Sensitivity	90.9%	90.9%	90.9%	90.9%	90.9%
Specificity	71.0%	77.4%	83.6%	83.6%	78.7%
Positive Likelihood Ratio	3.13	4.02	5.54	5.54	4.27
Negative Likelihood Ratio	0.13	0.12	0.11	0.11	0.12
Logit cut	-1.68	-1.41	-1.86	-1.72	-1.83

Disclosure: G. Gulati, None; K. Abulaban, None; J. Ying, None; X. Zhang, None; H. Song, None; Q. Ma, None; C. Haffner, None; K. Wiley, None; M. Bennett, None; B. H. Rovin, Lilly, 5; H. I. Brunner, None.

To cite this abstract in AMA style:

Gulati G, Abulaban K, Ying J, Zhang X, Song H, Ma Q, Haffner C, Wiley K, Bennett M, Rovin BH, Brunner HI. Validating the RAIL Algorithm in Adult Lupus Nephritis Patients [abstract]. Arthritis Rheumatol. 2015; 67 (suppl 10). https://acrabstracts.org/abstract/validating-the-rail-algorithm-in-adult-lupus-nephritis-patients/. Accessed .

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/validating-the-rail-algorithm-in-adult-lupus-nephritis-patients/