Abstract
BACKGROUND: Differentiating between glomerular and tubulointerstitial diseases can guide selection of appropriate patients for kidney biopsy. The aim of this study is to identify urine tests that can differentiate between these histological diagnoses. METHODS: In this sub-study of a prospectively enrolled cohort of participants with urine samples concurrent with their kidney biopsy, we tested the association of 24 features on urinalysis, urine sediment microscopy, and biomarkers of glomerular and tubular injury and inflammation with histological diagnosis of glomerular or tubulointerstitial disease. We selected a combination of features associated with glomerular disease using stepwise forward and backward regression, and LASSO algorithm after dividing the cohort into training (70%) and test (30%) sets. RESULTS: Of 359 participants, 121 had glomerular, 89 had tubulointerstitial diseases, and 149 were classified as mixed. Compared to patients with tubulointerstitial diseases, those with glomerular diseases had more dipstick hematuria (3+ vs. 1+, P < 0.001) and urine albumin (1.25 vs. 0.09 mg/mg, P < 0.001). Patients with glomerular diseases had higher levels of tubular health biomarkers (Uromodulin, 1.22 vs. 0.92, P = 0.03). In a multivariable model, higher urine albumin, dipstick blood, and urine uromodulin were independently associated with higher odds of glomerular diseases (test set AUC, 0.81 (0.69, 0.93)). CONCLUSION: Urine tests, including urine albumin, dipstick blood, and urine uromodulin, were associated with the histological diagnosis of glomerular disease. These findings can help clinicians differentiate between glomerular and tubulointerstitial diseases and guide clinical decisions regarding a kidney biopsy.