*Purpose: Organs discards continue to be high due to lack of accurate tools to assess donor quality and predict organ function.To identify characteristics associated with short-term outcome in deceased donor (DD) KT recipients, we compared predictive models that included KDPI only, molecular markers, and the combination of KDPI and molecular markers. Such a strategy may prove useful when evaluating composite scoring systems.

*Methods: Affymetrix gene expression data from pre-implant (PI) biopsies and creatinine reduction ratio were available for 185 KT subjects. Following Vilar et al (2010), short-term outcome was categorized as immediate graft function (IGF) defined as creatinine reduction ratio from post-transplant day 1 to 2 (CRR2) > 30, no dialysis-delayed graft function (ND-DGF, CRR2<=30) and dialysis-delayed graft function (D-DGF). We fit three different models to predict IGF versus D-DGF using (1) KDPI only, (2) gene expression data, and (3) KDPI and gene expression data. For each, we estimated the area under the receiver operating characteristic curve (AUC) and the net reclassification improvement for comparing the three models.

*Results: Among 185 DD KT recipients, there were 38 (19.9%) IGF, 76 (39.8%) ND-DGF, and 71 (37.2%) D-DGF subjects. GFR at one week post-KT differed significantly with mean GFRs of 64.92, 34.13, 17.15 in the IGF, ND-DGF, and D-DGF groups, respectively (P <0.001). Similarly, GFR at one month post-KT differed significantly with mean GFRs of 73.68, 52.8, 44.53 in the IGF, ND-DGF, and D-DGF groups, respectively (P <0.001). This indicates that Vilar’s short-term categorization, which is determined at post-KT day 2, may be a useful proxy for durable short-term response. Donor type (P=0.002), donor age (P=0.002), logarithm of last donor creatinine (mg/dL) (P<0.001), recipient age (P=0.005), recipient body weight (P<0.001), recipient body mass index (P=0.001), KDPI (P <0.001) and KDRI (P=0.001) were significantly different between the three groups. When considering these outcomes as ordered categories (IGF > ND-DGF > D-DGF) and analyzing the gene expression using ordinal response models, 33 probe sets were significant at the FDR<0.10 level and 10 were significant at the FDR<0.05 level. There were 17 probe sets included in the gene only model and 14 probe sets included in the KDPI + gene model, with 11 probe sets in common. The areas under the ROC curves for the three fitted models were KDPI only (AUC=0.724, 95% CI: 0.621, 0.827), gene only (AUC=0.935, 95% CI: 0.886, 0.983), and KDPI + gene (AUC=0.941, 95% CI: 0.900, 0.983). There was a significant difference between the KDPI only model and the gene only model (P=0.0004) as well as KDPI + gene model (P=0.0002). However, there was not a significant difference between the gene only and KDPI + gene models when comparing the AUCs (P=0.64).

*Conclusions: A panel of PI molecular markers may predict short-term outcomes more accurately than scoring systems currently in place.

« Back to 2019 American Transplant Congress