Renal scintigraphy to predict persistent renal failure after acute kidney injury: an observational study.

INTRODUCTION: Renal scintigraphy (RS) is occasionally performed to assess the risk of persistent renal failure (PRF) in patients with acute kidney disease (AKD). However, its diagnostic performance has never been assessed. METHODS: We identified all patients with AKD for whom RS was performed in our institution between 2010 and 2017. PRF was defined as persistently low (< 33% of baseline) estimated glomerular filtration rates (eGFR), 1 year after RS. Nuclear medicine specialists reviewed RS data and rated, for each patient, the likelihood of PRF ("PRF score"). We evaluated the performance to predict PRF (area under the ROC curve (AUC)) of RS-derived parameters such as renal accumulation index, accumulation slope, and new parameters derived from serial kidney activity counts. We tested the ability of those parameters to improve a clinical model including hypertension, diabetes, AKI severity and baseline eGFR. Finally, we conducted sensitivity analyses using alternate PRF definitions. RESULTS: Among 97 patients included, 57 (59%) fulfilled the criteria for PRF. The PRF score was able to predict PRF with an AUC of 0.63. Similarly, the accumulation index and accumulation slope respective AUCs were 0.64 and 0.63. None of these parameters were able to improve the performance of the clinical model. Among new parameters, the 3rd/2nd minute activity ratio and 3rd/2nd minute activity slope had fair diagnostic performance (AUC 0.72 and 0.74, respectively) and improved the performance of the clinical model. Results were confirmed in sensitivity analyses. CONCLUSION: Conventional renal scintigraphy can identify patients at high risk of PRF with a high specificity but a low sensitivity. New parameters, with comparable diagnostic abilities can be obtained within three minutes of injection.