Validation of the International IgA risk prediction tool in American Indians and Hispanics

Background The International Immunoglobulin A nephropathy (IgAN) risk prediction assesses the risk of kidney failure in patients with IgAN. The performance of this risk prediction tool has not been studied in American Indians and Hispanics. We conducted a single-center study to assess the equation performance in this population. Methods We calculated the 5-year risk of developing kidney failure with the IgAN risk prediction equation without race and assessed the equation performance using the metrics of calibration, discrimination, and overall prediction error. Results Thirty-four patients were included, most of whom identified as of Hispanic race/ethnicity (44.1%), or as American Indians (26.5%). At biopsy, the median (IQR) age, serum creatinine, and spot urine protein to creatinine ratio were 38 years (27-45), 2.15 mg/dl (1.51-3.04), and 2.7 g/g (1.5-5.8), respectively. The equation identified patients at high risk of developing kidney failure early with a concordance statistic of 0.79 (95% CI 0.68 – 0.89). The agreement between observed and predicted outcomes at 5 years was marginal, with over-estimation of risk for patients with low observed risk and vice versa. Overall prediction error was suboptimal in this cohort [index of prediction accuracy 0.34 (0.03 – 0.51)]. Conclusions The International IgAN risk prediction equation without race accurately identified patients at elevated risk of developing kidney failure. At 5 years, the agreement between the observed and predicted outcomes was sub-optimal, possibly due to advanced kidney disease in this cohort. A diverse development population may improve the risk prediction.