LAUSR

Introduction: Renal interstitial inflammation often presents in immunoglobulin A nephropathy (IgAN), but its predictive role in kidney disease progression remains controversial. Methods: This retrospective two-center cohort study included 1,420 adult IgAN patients between January 2003 and May 2018 followed for a median of approximately 7 years at two Chinese hospitals. The predictor was renal interstitial inflammation within the total cortical interstitium (none/mild [0–25%], moderate [26–50%], or severe [>50%]). For the further propensity score matching analyses, the participants with moderate and severe level of interstitial inflammation were pooled to match those with none/mild level of interstitial inflammation. The outcomes included the rate of kidney function decline, and the composite kidney endpoint event defined as a >40% reduction in the estimated glomerular filtration rate, end-stage kidney disease. Linear regression and Cox proportional hazards regression analyses were used to examine the association between interstitial inflammation and the outcomes. The predictive performance of the model also assessed using multivariate logistic regression analyses with the receiver operating characteristic curve analysis. Reclassification was assessed using the continuous net reclassification improvement and integrated discrimination improvement adapted for censoring for the assessment of the model with or without interstitial inflammation. Results: For the check of reproducibility, the kappa statistic was 0.71, and intraclass correlation coefficient was 0.77. After adjustment for relating covariates, a higher level of interstitial inflammation was associated with a faster rate of kidney function decline (eGFR slope [mL/min/1.73 m2] of 1.34 [95% CI: −2.56 to 5.23], 3.50 [95% CI: −0.40 to 7.40], and 7.52 [95% CI: 3.02 to 12.01]) in the patients with none/mild, moderate, and severe interstitial inflammation, respectively, in the multivariable linear regression models and with an increased risk of kidney disease progression (HR for moderate vs. none/mild, 1.85; 95% CI: 1.10–3.13; HR for severe vs. none/mild, 2.95; 95% CI: 1.52–5.73) in the multivariable Cox proportional hazards models. Analyses in the propensity score-matched cohort, subgroups, and the sensitive analyses yielded consistent results. The receiver operating curves indicated a higher area under the curve of 0.83 in the model with interstitial inflammation compared with 0.81 in that without interstitial inflammation. In addition, incorporating interstitial inflammation into the International IgAN Risk Prediction Tool improved the diagnostic power of the algorithm to predict risk of progression. Conclusion: Interstitial inflammation is a reproducible pathologic parameter that may be adopted as a predictor for kidney disease progression in patients with IgAN.