BackgroundGlucocorticosteroid is used for patients with primary nephrotic syndrome. This study aims to identify and validate that biomarkers can be used to predict steroid resistance.MethodsOur study contained two stages, discovery and validation stage. In discovery stage, we enrolled 51 minimal change disease (MCD) or focal segmental glomerulosclerosis (FSGS) patients treated with full dose steroid. Five urinary biomarkers including β2-microglobulin (β2-MG) and α1-microglobulin (α1-MG) were tested and candidates’ biomarkers were selected based on their associations with steroid response. In validation stage, candidates’ biomarkers were validated in two prospectively enrolled cohorts. Validation cohort A included 157 FSGS/MCD patients. Validation cohort B included 59 membranous nephropathy (MN) patients. Patients were classified into response group (RG) or non-response group (NRG) based on their responses to steroid treatment.ResultsIn discovery stage, higher urinary β2-MG was independently associated with response to corticosteroid treatment in MCD/FSGS patients [OR = 1.89, 95% CI 1.02–3.53] after adjusted by age and gender. In validation cohort A, patients in NRG had a significant higher urinary β2-MG [Ln (β2-MG/uCr): 4.6 ± 1.7 vs 3.2 ± 1.5] compared to patients in RG. We then developed a 3-variable risk score in predicting steroid resistance in FSGS/MCD patients based on the best predictive model including Ln(β2-MG/uCr) [OR = 1.76, 95% CI 1.30–2.37], age [OR = 1.005, 95% CI 0.98–1.03] and pathology [MCD vs FSGS, OR = 0.20, 95% CI 0.09–0.46]. The area under the ROC curves of the risk score in predicting steroid response was 0.80 (95% CI 0.65–0.85). However, no such association was found in MN patients.ConclusionsOur study identified a 3-variable risk score in predicting steroid resistance in patients with FSGS or MCD.