G IRISH1,2,3, S CHADBAN1,4,5, N BOUDVILLE6,7, S CAMPBELL8,9, J KANELLIS10,11, P A CLAYTON1,2,3
1Australia and New Zealand Dialysis and Transplant (ANZDATA) Registry, South Australian Health and Medical Research Institute (SAHMRI), , Adelaide, Australia, 2Central and Northern Adelaide Renal and Transplantation Service. Royal Adelaide Hospital, Adelaide, Australia, 3Department of Medicine, University of Adelaide, Adelaide, Australia, 4Renal Medicine, Royal Prince Alfred Hospital, , Camperdown, Australia, 5Kidney Node, Charles Perkins Centre, University of Sydney, Sydney, Australia, 6Medical School, University of Western Austalia, Perth, Australia, 7Department of Renal Medicine, Sir Charles Gairdner Hospital, Perth, Australia, 8Department of Nephrology, Princess Alexandra Hospital, Brisbane, Australia, 9School of Medicine, The University of Queensland, Brisbane, Australia, 10Department of Nephrology, Monash Health, Melbourne, Australia, 11Centre for Inflammatory Diseases, Department of Medicine, Monash University, Melbourne, Australia
Aim: We developed a risk prediction score for overall graft survival in adult recipients of a living kidney donor transplant in an Australian and New Zealand population.
Background: Risk scores may aid risk quantification and decision-making in kidney transplantation. The Living Kidney Donor Profile Index (LKDPI) was developed to choose between living donors. The original LKDPI is not discriminatory (C=0.59, 95% CI 0.55-0.62) or well calibrated in Australia/New Zealand.
Methods: Using data from the Australia and New Zealand Dialysis and Transplant Registry, we included adult recipients of living kidney donor transplants over 2004-2018. We constructed Cox models for overall graft survival. We refit the original USA variables and constructed a new model considering all available potential covariates. Model performance was examined by partial validation, bootstrapping, Harrell’s C-statistic (C) and Akaike information criterion (AIC).
Results: 4049 living donors were included; 58.2% were female, 10.4% had hypertension and 18.7% were obese. For the original USA LKDPI the variables of both male and donor:recipient weight ratio were mis-specified. The C-statistic for the re-fit model was 0.57 (95%CI0.54-0.59). When remodelling the score, the new variables included were history of hypertension and HLA-A mismatches. Variables excluded were donor:recipient weight ratio, HLA-B, both male, and ABO incompatibility. The Aus/NZ LKDPI C-statistic was 0.56 (95%CI 0.54-0.58). Calibration was similar but the Aus/NZ LKDPI had better model fit (AIC:10402 vs. 11023).
Conclusions: The Aus/NZ LKDPI had similar discrimination to the original LKDPI and the LKDPI with adjusted coefficients. The discrimination was low for both scores and so should be used with caution to decide between donors. The new score is better calibrated for our population so could be used to predict graft prognosis.
Biography:
Georgina L Irish is an Australian Nephrologist who works at the Royal Adelaide Hospital. She undertook her training at The Royal Adelaide Hospital, Flinders Medical centre and Oxford University Trust Hospital. She is undertaking her Masters of Clinical Epidemiology through the University of Sydney. Georgina’s commenced her PhD in 2020 on Decision making in Kidney transplantation. She will be using epidemiological registry data to help inform decisions around kidney transplantation, and create decision aids to make these data available to the patient and clinician at the point of care