Background/Purpose:

The present study was aimed at developing an adult Renal Activity In Lupus (A-RAIL) algorithm for Lupus Nephritis (LN) activity assessment using urinary biomarkers (UBMs). A similar index for use in children with LN has been recently proposed.

Methods:

Adult Systemic Lupus Erythematosus (SLE) patients requiring a kidney biopsy as part of their standard care were enrolled. Patient demographics, SLE manifestations, laboratory parameters, renal activity/chronicity measures (R-SLEDAI, SDI-R) and a urine sample were collected at the time of biopsy. One expert read the biopsies in a blinded fashion to report National Institute of Health Activity and Chronicity indices (NIH-AI, NIH-CI). Patients were dichotomized as high activity or not based on the NIH-AI score of 10 as cut-off. UBMs measured included:  kidney injury molecule 1, monocyte chemotactic protein 1, neutrophil gelatinase associated lipocalin, ceruloplasmin, adiponectin, hemopexin, alpha-1-acid glycoprotein, hepcidin, lipocalin-like prostaglandin synthase (L-PGDS), transferrin, vitamin D binding protein, transforming growth factor beta, endothelial protein C receptor, and liver type fatty acid-binding protein 1 using standardized ELISA-based assays (except L-PDGS and transferrin, measured by immune nephelometry). Urinary microalbumin and creatinine levels were measured to normalize for proteinuria in the analysis. Candidate RAIL predictors were firstly selected using univariate analyses, followed by stepwise selection in multiple logistic regression models. An A-RAIL algorithm was developed using the selected UBMs, and predictive accuracy calculated using the area under the Receiver Operating Characteristic (ROC) curve (AUC).

Results:

Of the 75 patients studied (age 32.4 ± 10.1 years), 78.7% were females. Caucasian and African American ethnicities were equally represented (46.7%). Mean NIH-AI and CI scores were 4.57 ± 4.11 and 3.23 ± 2.59 respectively. Renal activity was high in 11 (15%). No significant differences were noted for age, gender, ethnicity, or LN stage or laboratory parameters between the two activity groups, except lower C3 in high activity (p = 0.029). A-RAIL algorithm was studied using 10 significant UBMs and all analyses controlled for NIH-CI. The predictive accuracy of analysis was reported “outstanding” at the optimal score threshold ([AUC = 0.9 (95% CI 0.84 – 1.00]), and was similar when normalized for urinary creatinine or microalbumin [Table 1].

Conclusion:

The A-RAIL algorithm, termed “liquid biopsy for LN” by some, provides an accurate non-invasive assessment of LN activity. Its robustness and clinical usefulness is confirmed by use of various standards of normalization of proteinuria with LN. These results are very similar to the pediatric RAIL developed using six UBMs. This sets foundation for further studies assessing disease progression and response to therapy. 

TABLE 1: Diagnostic Accuracy of A-RAIL Algorithm