Validation and calibration of the prognostic Kidney Donor Risk Index (KDRI) scoring system of deceased donors for renal transplantation in The Netherlands

H. Peters-Sengers, M.B.A. Heemskerk, J.J. Homan van der Heide, S.P. Berger, F.J. Bemelman

Chair(s): dr. D.E. Braat & dr. N.C. van der Weerd

Thursday 9 march 2017

10:12 - 10:24h at Johan de Meesterzaal

Categories: Parallel - Klinisch

Parallel session: Parallelsessie XII – Klinisch 4 - Pretransplantatie en donatie

Background:
 The prognostic Kidney Donor Risk Index (KDRI)—developed and internally validated (c-statistic 0.62) in the US—is a widely used tool to predict transplant outcome of a deceased donor kidney.

Methods:
We aimed to externally validate and calibrate the KDRI as proposed by Rao et al. (2009), containing 10 donor (KDRIdonor-only) and 4 transplant factors (KDRIoriginal), with stratification on recipient age. We used the Dutch Organ Transplantation Registry to include 3147 recipients (≥18y) from all Dutch centers, transplanted from 2002 to 2012 with a first brain-death or circulatory-death donor (≥18y) kidney, and followed them till September 2015.

Results:
The median Dutch KDRI was increased to 1.21 compared with reported by Rao et al. in 2009, and comparable with the year 2012 in US (median of 1.24). Kidneys in the highest KDRIoriginal quintile (1.45+) had an adjusted 5-year graft survival of 68.9%, whereas the lowest quintile (<0.79) showed survival rate of 84.7%. Discriminative ability (Harrell’s C) of the KDRIoriginal was 0.63 (95%CI 0.62-0.64), and slightly lower for the KDRIdonor-only 0.62 (95%CI 0.61-0.63). Worse KDRI donors (>1.45) show less discrimination in elderly (65+) recipients than younger recipients, suggesting that in the elderly population other risk factors than the KDRI contribute to graft loss. The calibration-slope of the KDRIoriginal was 0.98 (SE 0.13) and KDRIdonor-only was 0.97 (SE 0.14), both not significant from 1 (p=.850; p=.844, respectively), indicating that discrimination of the KDRI was almost identical in the Dutch cohort. A joint test of all KDRI factors in cox regression—including the KDRI as offset term—indicated overall evidence of lack of fit (χ2[13]=46.5, p<.001). We found misspecification of the following donor factors: hypertension (p=.064), length (p=.003), weight (p<.001), and cold ischemic time (p<.001). The following donor factors delivered no added value in terms of model fit to the KDRI: atherosclerosis, smoking, HLA-A mismatches, and inotropes prior to donation.

Conclusion:
The KDRI scoring system for deceased donors shows equivalent discrimination and accuracy as compared with the US. The C statistic is too low to consider the KDRI as a useful decision-making tool for the individual recipient. However, the KDRI could be used to assist allocation for longevity matching between cohorts of donors and recipients.

More information