Thursday 14 march 2019
12:48 - 12:51h
Categories: Klinisch, Postersessie
Parallel session: Postersessies 2 - Clinical
H. Peters-Sengers1, A. Tomer2, S. Florquin3, J.J.T.H. Roelofs3, E.W. Steyerberg4, F.J. Bemelman5, D. Rizopoulos6, J. Kers3.
1Dept. of Infectious diseases, Amsterdam UMC, locatie AMC, Amsterdam. 2Dept. of Biostatistics, Erasmus University Medical Center, Rotterdam. 3Dept. of Pathology, Amsterdam UMC, locatie AMC, Amsterdam. 4Dept. of Public Health, Erasmus University Medical Center, Rotterdam. 5Dept. of Nephrology, Renal Transplantation Unit, Amsterdam UMC, locatie AMC, Amsterdam.6Dept. of Biostatistics, Erasmus University Medical Center, Rotterdam, The Netherlands.
Background: In the current study, we aimed to optimize the monitoring of kidney transplanted patients based on personalized risk estimates and compare such personalized screening with the one-size-fits-all protocol that is currently being used in our hospital. Such a personalized screening approach could potentially lead to a lower medical and financial burden in renal transplant patients with stable transplant function without losing important information on patients at risk for irreversible graft failure that would require timely intervention.
Methods: We included adult patients transplanted between 1996 and 2009 who had >1 additional SCr (umol/L) and spot or 24-hour urine collection to calculate the urine PCR (g/mol). A Joint Model was constructed to model the association between longitudinal markers of renal function and death-censored graft failure. As a proof of concept, we applied personalized screening intervals and compared this to the fixed screening approach in the number of screening intervals and graft failure offset, defined as the time difference between estimated intervention time and observed graft failure time.
Results: We included 238 renal transplanted patients with 13062 SCr measurements and 9616 PCR measurements. Majority were recipients of deceased donors (74.1%). Mean recipient age was 50.7 (SD 12.7) years, and majority firstly transplanted (84.5%). Death-censored graft survival was 83.9% (95%CI 78.2-89.6) at 5 years. A joint model that included both the SCr and the PCR trajectories did not reveal an increased time-dependent (t)AUC compared to a model that only included SCr trajectories ([t]AUC > 0.8 up till 2.5 years). The personalized screening approach resulted in obtaining less SCr measurements with a median (IQR) of 14 (6.0) versus 29 (8.5) visits. The time to intervene and overcome the risk for graft failure was comparable with the fixed schedule (14% versus 12% missed cases with graft failure).
Conclusions: A personalized screening could be applied for monitoring renal transplant patients. Patients who remain relatively stable may not require frequent measurement of SCr, and patients for whom the graft function deteriorates faster, a frequent schedule of SCr may be required to determine the best moment for intervention. Our findings have to be externally validated in other observational cohorts.