The UK kidney donor risk index poorly predicts long-term transplant survival in paediatric kidney transplant recipients.

Abstract

The UK kidney offering scheme introduced a kidney donor risk index (UK-KDRI) to improve the utility of deceased-donor kidney allocations. The UK-KDRI was derived using adult donor and recipient data. We assessed this in a paediatric cohort from the UK transplant registry. We performed Cox survival analysis on first kidney-only deceased brain-dead transplants in paediatric (<18 years) recipients from 2000-2014. The primary outcome was death-censored allograft survival >30 days post-transplant. The main study variable was UK-KDRI derived from seven donor risk-factors, categorised into four groups (D1-low risk, D2, D3 and D4-highest risk). Follow-up ended on 31-December-2021. 319/908 patients experienced transplant loss with rejection as the main cause (55%). The majority of paediatric patients received donors from D1 donors (64%). There was an increase in D2-4 donors during the study period, whilst the level of HLA mismatching improved. The KDRI was not associated with allograft failure. In multi-variate analysis, increasing recipient age [adjusted HR and 95%CI: 1.05(1.03-1.08) per-year, p<0.001], recipient minority ethnic group [1.28(1.01-1.63), p<0.05), dialysis before transplant [1.38(1.04-1.81), p<0.005], donor height [0.99 (0.98-1.00) per centimetre, p<0.05] and level of HLA mismatch [Level 3: 1.92(1.19-3.11); Level 4: 2.40(1.26-4.58) versus Level 1, p<0.01] were associated with worse outcomes. Patients with Level 1 and 2 HLA mismatches (0 DR +0/1 B mismatch) had median graft survival >17 years regardless of UK-KDRI groups. Increasing donor age was marginally associated with worse allograft survival [1.01 (1.00-1.01) per year, p=0.05]. Adult donor risk scores were not associated with long-term allograft survival in paediatric patients. The level of HLA mismatch had the most profound effect on survival. Risk models based on adult data alone may not have the same validity for paediatric patients and therefore all age-groups should be included in future risk prediction models.