The present invention relates to the prediction of kidney injury. In this study, the inventors studied the transcriptomic landscape in AAV-GN to find biomarkers for refining diagnosis and/or prognosis, which would help stratify patient risk and tailor therapeutic management. The objectives of this study were to investigate the potential prognostic value and/or pathogenic role of the most relevant genes associated with kidney survival. They identified a 4-gene signature (XRCC6, PRKCD, TEK, and CLU) that predicted kidney survival better than histological-based classifications. Thus, the invention relates to a method for predicting the development of a kidney failure in a patient suffering from an ANCA-associated vasculitis-associated GN (AAV-GN).