Development and validation of an optimized integrative model using urinary chemokines for noninvasive diagnosis of acute allograft rejection. Issue 12 (27th June 2020)

• Record Type: Journal Article
• Authors: Tinel, Claire; Devresse, Arnaud; Vermorel, Agathe; Sauvaget, Virginia; Marx, David; Avettand‐Fenoel, Véronique; Amrouche, Lucile; Timsit, Marc-Olivier; Snanoudj, Renaud; Caillard, Sophie; Moulin, Bruno; Olagne, Jérome; Essig, Marie; Gwinner, Wilfried; Naesens, Maarten; Marquet, Pierre; Legendre, Christophe; Terzi, Fabiola; Rabant, Marion; Anglicheau, Dany
• Is Part Of: American journal of transplantation. Volume 20:Issue 12(2020)
• Page Start: 3462
• Page End: 3476
• Publication Date: 2020-06-27
• Subjects: biomarker -- clinical decision‐making -- clinical research/practice -- kidney transplantation/nephrology -- rejection: acute; Transplantation of organs, tissues, etc -- Periodicals; 617.95
• DOI: 10.1111/ajt.15959
• Languages: English
• ISSNs: 1600-6135

Abstract

Abstract : The urinary chemokines CXCL9 and CXCL10 are promising noninvasive diagnostic markers of acute rejection (AR) in kidney recipients, but their levels might be confounded by urinary tract infection (UTI) and BK virus (BKV) reactivation. Multiparametric model development and validation addressed these confounding factors in a training set of 391 samples, optimizing the diagnostic performance of urinary chemokines. CXCL9/creatinine increased in UTI and BKV viremia with or without nephropathy (BKVN) (no UTI/leukocyturia/UTI: −0.10/1.61/2.09, P = .0001 and no BKV/viremia/BKVN: −0.10/1.90/2.29, P < .001) as well as CXCL10/creatinine (1.17/2.09/1.98, P < .0001 and 1.13/2.21/2.51, P < .001, respectively). An optimized 8‐parameter model (recipient age, sex, estimated glomerular filtration rate, donor specific antibodies, UTI, BKV blood viral load, CXCL9, and CXCL10) diagnosed AR with high accuracy (area under the curve [AUC]: 0.85, 95% confidence interval [CI]: 0.80‐0.89) and remained highly accurate at the time of screening (AUC: 0.81, 95% CI: 0.48‐1) or indication biopsies (AUC: 0.85, 95% CI: 0.81‐0.90) and within the first year (AUC: 0.86, 95% CI: 0.80‐0.91) or later (AUC: 0.90, 95% CI: 0.84‐0.96), achieving AR diagnosis with an AUC of 0.85 and 0.92 ( P < .0001) in 2 external validation cohorts. Decision curve analyses demonstrated the clinical utility of the model. Considering confounding factors rather than excluding them, we optimized a noninvasive multiparametric diagnostic model for AR of kidney allografts with unprecedented accuracy. Abstract : Inclusion of potential confounding factors of increased urinary chemokine levels led to the development of a robust, noninvasive, 8‐parameter model of acute renal allograft rejection.