Abstract: TH-OR036

Diffusion MRI for Assessment of Kidney Fibrosis

Session Information

Category: Chronic Kidney Disease (Non-Dialysis)

  • 305 CKD: Clinical Trials and Tubulointerstitial Disorders

Authors

  • Berchtold, Lena, Hôpitaux Universitaire de Genève, Geneva, Switzerland
  • Friedli, Iris, Hôpitaux Universitaire de Genève, Geneva, Switzerland
  • Hadaya, Karine, University Hospital of Geneva, Geneva, Switzerland
  • Martin, Pierre-Yves F., Hospital Universitaire de Geneve, Geneva City, Switzerland
  • Vallée, Jean-Paul, Hôpitaux Universitaire de Genève, Geneva, Switzerland
  • De Seigneux, Sophie M., Hôpitaux Universitaire de Genève, Geneva, Switzerland
Background

Renal interstitial fibrosis (IF) is a process common to all kidney diseases and is predictive of renal prognosis. IF can currently only be assessed by biopsy, an invasive procedure associated with complications and focal sampling. There is currently no clinically available noninvasive method to assess IF. Diffusion Magnetic Resonance Imaging (MRI) is emerging as a promising tool to evaluate kidney fibrosis non invasively. The aim of this study was to validate in a mixed CKD population a novel renal MRI diffusion sequence that we recently developed and to create a new non-invasive score for assessment of IF.

Methods

In this prospective study, we included 124 CKD patients having undergone a kidney biopsy (native or transplant). Optimized Diffusion-Weighted Imaging (DWI) and T1 sequences were compared to histological assessment of IF. Differences between cortical and medullary Apparent Diffusion Coefficient (ΔADC) and T1(ΔT1) values were assessed and compared to gold standard histopathology. We then combined routinely measured serum markers and ΔADC to create a new score for assessment of IF.

Results

In CKD patients, ΔADC correlated well with IF (r=-0.55, p<0.001). This good correlation was observed in both CKD and kidney transplant patients. ΔADC showed a better discrimination to IF than cortical ADC values, T1 values and ΔT1. To optimize fibrosis prediction, we combined ΔADC values to routinely obtained markers known to correlate to fibrosis (phosphate, hemoglobin, eGFR) to obtain a score of predicted fibrosis. We observed a strong correlation between our score and histological IF(r=0.8, p<0.001). We further built receiver operating characteristic curves and reported area under the curve (AUC) to discriminate between patients with high levels of fibrosis (≥40%). Analysis revealed that the new score was predictive of fibrosis ≥40% with an AUC was 0.99. The sensitivity and specificity to detect IF of more than 40% were 76.2% and 100% respectively, implying that our scoring system is able to identify patients with more than 40% without false positive.

Conclusion

In summary, we validated the use of the ΔADC to predict IF non invasively in CKD and kidney transplant patients. We further derive a scoring system from ΔADC and commonly obtained laboratory values and showed a high specificity to identify non invasively patient harboring extensive fibrosis (≥40%).