Changes in urinary metabolomic profile during relapsing renal vasculitis

Abstract

Current biomarkers of renal disease in systemic vasculitis lack predictive value and are insensitive to early damage. To identify novel biomarkers of renal vasculitis flare, we analysed the longitudinal urinary metabolomic profile of a rat model of anti-neutrophil cytoplasmic antibody (ANCA) vasculitis. Wistar-Kyoto (WKY) rats were immunised with human myeloperoxidase (MPO). Urine was obtained at regular intervals for 181 days, after which relapse was induced by re-challenge with MPO. Urinary metabolites were assessed in an unbiased fashion using nuclear magnetic resonance (NMR) spectroscopy, and analysed using partial least squares discriminant analysis (PLS-DA) and partial least squares regression (PLS-R). At 56 days post-immunisation, we found that rats with vasculitis had a significantly different urinary metabolite profile than control animals; the observed PLS-DA clusters dissipated between 56 and 181 days, and re-emerged with relapse. The metabolites most altered in rats with active or relapsing vasculitis were trimethylamine N-oxide (TMAO), citrate and 2-oxoglutarate. Myo-inositol was also moderately predictive. The key urine metabolites identified in rats were confirmed in a large cohort of patients using liquid chromatography-mass spectrometry (LC-MS). Hypocitraturia and elevated urinary myo-inositol remained associated with active disease, with the urine myo-inositol:citrate ratio being tightly correlated with active renal vasculitis.

Figure 1. Change in urine parameters and anti-MPO titre over time

. WKY rats were immunised with hMPO or HSA, and followed for 181 days. (A) Haematuria peaked at 56 days post immunisation and declined steadily thereafter. (B) Albuminuria, as assessed by albumin:creatinine ratio (ACR), also peaked at day 56, but remained elevated until day 181. Data are presented as mean +/− s.e.m. (C) Anti-MPO titres declined progressively over time and rats with EAV exhibited reduced excretory renal function, as estimated by measurement of (D) creatinine clearance. Data are presented as median and IQR.

Figure 2. Urinary metabolite profile of rats with EAV at the point of peak disease, during remission and following induction of relapse.

(A) PLS-DA on 2 Latent variables at several time points during the evolution of EAV (day 56, day 112, day 181 (pre relapse) and day 210 (post relapse)) discriminating between (●) MPO (n = 16) and (○) HSA (n = 11) groups, and between MPO re-stimulation (◓, n = 6) or control (saline) re-stimulation (○, n = 7) of EAV animals. MPO and HSA separate on 2 LVs following stimulation; this separation is largely lost by day 112 and 181, with re-emergence following induction of relapse. Haematuria data are presented as mean +/− s.e.m. (B) PLS-R of binned 1D NMR spectra of rat urine against histological glomerular damage score identifies key urine metabolite markers predictive of damage at day 56 (upper plot) and day 210 (lower plot). The correlation plot shows predicted vs. observed damage scores using N bins.

Figure 3. Effect of relapse induction.

WKY rats were immunised with hMPO or HSA and sacrificed at day 56 or day 210, the latter following treatment with either LPS + MPO (Relapse) or saline (Control) at day 181. (A) The degree of glomerulonephritis was scored blindly at 56 days and 210 days. (B) Lung haemorrhage was assessed by staining lung sections with Perl’s stain, which shows iron deposition as blue, and quantifying the degree of blue staining with image analysis. Data are presented as median and IQR. (C) Representative image of a renal section from a severely affected rat relapsed with MPO and LPS. There is a marked interstitial infiltrate, focal necrotising crescentic glomerulonephritis and tubular dilatation (H&E, x100). (D) Representative image depicting brown-staining haemosiderin-laden alveolar macrophages in a lung section from a relapsed animal (H&E, x400). (E,F) Representative images from Perl’s stained lung sections from (E) a control and (F) a relapsed animal (x10).

Figure 4. Urinary metabolite profile of patients with active and remission vasculitis.

(A) PLS-DA of binned LS-MS spectra of human urine samples from patients with active renal vasculitis (●, n = 28) against cases in remission (, n = 104). PLS-DA weightings identify key metabolites associated with active renal vasculitis. (B) Receiver operator characteristic curve showing the prediction accuracy of the binary logistic model based on citric acid and myo-Inositol with area under the curve (AUC) = 0.922 (95% CI 0.849–0.970, p = 0.001). (C) Variable importance in the project (VIP) scores summarising the predictive value of each metabolite in identifying active renal vasculitis. Values above 1 indicate a useful marker.

Figure 5. Utility of the urine myoinositol:citrate ratio in identifying active vasculitis.

(A) Myo-inositol:citrate ratio as determined by LC-MS according to diagnostic groupings. (B) Reverse confirmation of urine myo-inositol:citrate ratio using NMR, with inclusion of patients with urine infection (UTI), N = 3–4 per group. A value from a patient with remission vasculitis is shown for comparison. Bars depict median and 90^th centile. AAV (R) = AAV in remission. ****p < 0.001, *p < 0.05.