Background/Purpose: Current management of lupus nephritis (LN) is guided by histopathological features on kidney biopsy and measurement of proteinuria. Urine proteomics is a non-invasive source of novel biomarkers which may better reflect the complex dynamic immunobiology of LN in real time. Two composite measures include CKD273, which can predict the risk of progression of chronic kidney disease in the general population, and LN120, which was designed to diagnose LN. Both are multidimensional urine proteomic classifiers consisting of 273 or 120 peptides, respectively, with major components including collagen fragments, abundant blood-derived proteins, and proteins involved in inflammation. We investigated the ability of these classifiers to predict traditional biopsy features and disease response in LN.

Methods: A total of 31 adults with biopsy-proven LN were included in this study. All participants met the SLICC and 2019 EULAR/ACR Classification Criteria for SLE based on a spot urine protein-to-creatinine ratio of >0.5 and class III, IV, and/or V LN on renal biopsy. Urine samples were collected at week 0 (at the time of renal biopsy) and week 12 and then subjected to peptidome analysis using a capillary electrophoresis-mass spectrometry (CE-MS) platform. This peptidome data was used to calculate CKD273 and LN120 classifiers at each time point. LN response status was determined at week 52 based on proteinuria, creatinine, and prednisone dosage (no more than 10 mg daily). Spearman’s rank correlation and t-tests were used to compare proteomic classifiers with renal biopsy characteristics and response.

Results: At week 0, both CKD273 and LN120, but not proteinuria, exhibited a moderate to strong correlation with histological activity index on renal biopsy (Figure 1; ρ = 0.65 with p = 0.00024 for CKD273; ρ = 0.47 with p = 0.013 for LN120). CKD273 also correlated with chronicity index (ρ = 0.54, p = 0.0037). Neither classifier significantly correlated with lupus nephritis ISN class. With respect to response, CKD273 and LN120 were not significantly different between groups at week 0. However, a reduction in LN120 was observed in 100% of complete responders, 60% of partial responders, and 50% of non-responders at week 12 (Figure 2). The magnitude of this change in LN120 in complete responders versus non-responders did not reach statistical significance (p = 0.13), though this is potentially because of the small number of responders with CE-MS data available at both time points (n = 4). CKD273 did not significantly change with time in any response group (Figure 3).

Conclusion: This work provides proof of concept that urine proteomic classifiers can noninvasively predict histological activity and chronicity in LN. Complete responders, but not partial responders or non-responders, exhibited an impressive numerical decrease in LN120 by week 12, suggesting that proteomic scores may track with and predict a durable treatment response. Larger studies are needed to validate these findings.

Figure 1. A) Correlation between CKD273 and activity index on renal biopsy at week 0. B) Analogous correlation between LN120 and activity index at week 0. R = Spearman’s rank correlation coefficient.

Figure 2. Change in LN120 from week 0 (W0) to week 12 (W12) by response group. Thin lines represent individual participants, and thicker lines represent the average LN120 per response group. Only individuals with urine peptidome data from both time points were included in this figure (4 complete responders, 5 partial responders, and 12 non-responders).

Figure 3. Change in CKD273 from week 0 (W0) to week 12 (W12) by response group. Thin lines represent individual participants, and thicker lines represent the average CKD273 per response group. Only individuals with urine peptidome data from both time points were included in this figure (4 complete responders, 5 partial responders, and 12 non-responders).

« Back to ACR Convergence 2020

ACR Meeting Abstracts - https://acrabstracts.org/abstract/urine-proteomic-classifiers-predict-renal-histological-activity-and-chronicity-indices-and-may-predict-treatment-response-in-lupus-nephritis/