TY - JOUR
T1 - External Validation of a Pretransplant Biomarker Model (REVERSE) Predictive of Renal Recovery After Liver Transplantation
AU - Levitsky, Josh
AU - Asrani, Sumeet K.
AU - Abecassis, Michael
AU - Ruiz, Richard
AU - Jennings, Linda W.
AU - Klintmalm, Goran
PY - 2019/10/1
Y1 - 2019/10/1
N2 - In patients with end-stage liver disease, the ability to predict recovery of renal function following liver transplantation (LT) remains elusive. However, several important clinical decisions depend on whether renal dysfunction is recoverable after LT. We used a cohort of patients undergoing LT to independently validate a published pre-LT model predictive of post-transplant renal recovery (Renal Recovery Assessment at Liver Transplant [REVERSE]: high osteopontin [OPN] and tissue inhibitor of metalloproteinases-1 [TIMP-1] levels, age < 57, no diabetes). Serum samples pre-LT and 4-12 weeks post-LT (n = 117) were analyzed for kidney injury proteins from three groups of recipients: (1) estimated glomerular filtration rate (eGFR) < 30 mL/minute/1.73 m2 prior to and after LT (irreversible acute kidney injury [AKI]), (2) eGFR < 30 mL/minute/1.73 m2 prior to LT and >50 mL/minute/1.73 m2 after LT (reversible AKI [rAKI]) (3) eGFR > 50 mL/minute/1.73 m2 prior to and after LT (no AKI). In patients with elevated pre-LT serum levels of OPN and TIMP-1, recovery of renal function correlated with decreases in the level of both proteins. At 4 weeks post-LT (n = 77 subset), the largest decline in OPN and TIMP-1 was seen in the rAKI group. Validation of the REVERSE model in this independent data set had high area under the curve (0.78) in predicting full post-LT renal recovery (sensitivity 0.86, specificity 0.6, positive predictive value 0.81, negative predictive value 0.69). Our eGFR findings were confirmed using measured GFR. Conclusion: The REVERSE model, derived from an initial training set combining plasma biomarkers and clinical characteristics, demonstrated excellent external validation performance characteristics in an independent patient cohort using serum samples. Among patients with kidney injury pre-LT, the predictive ability of this model may prove beneficial in clinical decision-making both prior to and following transplantation.
AB - In patients with end-stage liver disease, the ability to predict recovery of renal function following liver transplantation (LT) remains elusive. However, several important clinical decisions depend on whether renal dysfunction is recoverable after LT. We used a cohort of patients undergoing LT to independently validate a published pre-LT model predictive of post-transplant renal recovery (Renal Recovery Assessment at Liver Transplant [REVERSE]: high osteopontin [OPN] and tissue inhibitor of metalloproteinases-1 [TIMP-1] levels, age < 57, no diabetes). Serum samples pre-LT and 4-12 weeks post-LT (n = 117) were analyzed for kidney injury proteins from three groups of recipients: (1) estimated glomerular filtration rate (eGFR) < 30 mL/minute/1.73 m2 prior to and after LT (irreversible acute kidney injury [AKI]), (2) eGFR < 30 mL/minute/1.73 m2 prior to LT and >50 mL/minute/1.73 m2 after LT (reversible AKI [rAKI]) (3) eGFR > 50 mL/minute/1.73 m2 prior to and after LT (no AKI). In patients with elevated pre-LT serum levels of OPN and TIMP-1, recovery of renal function correlated with decreases in the level of both proteins. At 4 weeks post-LT (n = 77 subset), the largest decline in OPN and TIMP-1 was seen in the rAKI group. Validation of the REVERSE model in this independent data set had high area under the curve (0.78) in predicting full post-LT renal recovery (sensitivity 0.86, specificity 0.6, positive predictive value 0.81, negative predictive value 0.69). Our eGFR findings were confirmed using measured GFR. Conclusion: The REVERSE model, derived from an initial training set combining plasma biomarkers and clinical characteristics, demonstrated excellent external validation performance characteristics in an independent patient cohort using serum samples. Among patients with kidney injury pre-LT, the predictive ability of this model may prove beneficial in clinical decision-making both prior to and following transplantation.
UR - http://www.scopus.com/inward/record.url?scp=85072993007&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85072993007&partnerID=8YFLogxK
U2 - 10.1002/hep.30667
DO - 10.1002/hep.30667
M3 - Article
C2 - 31002431
AN - SCOPUS:85072993007
VL - 70
SP - 1349
EP - 1359
JO - Hepatology
JF - Hepatology
SN - 0270-9139
IS - 4
ER -