Factors associated with 25-year kidney allograft survival: a single-center retrospective analysis

One-year outcomes after kidney transplantation (KTx) have improved markedly; however, ultra-long graft survival remains rare. We analyzed an outcome-defined sample from a single-center, single-era cohort to identify early, potentially modifiable factors associated with ≥25-year graft survival and to translate these associations into pragmatic, hypothesis-generating, center-level process targets.

We retrospectively reviewed the first adult solitary KTx performed in 1980-1995. Two outcome-defined, unmatched groups were compared: ultra-long survivors (ULS; graft survival ≥25 years; n = 59) and early graft failure (EGF; ≤10 years; n = 61). We extracted pre- and peri-transplant variables, first-year rejection burden (0/1/≥2 treatment cycles), and longitudinal laboratory data. Unconditional logistic regression was used to estimate adjusted associations (odds ratios [ORs]) in a complete-case subset with a pre-specified, parsimonious adjustment (recipient and donor age).

Cold ischemia time (CIT) was shorter in ULS than in EGF (1281.8 ± 473.9 vs. 1764.8 ± 564.9 min; p = 0.016; mean difference 483 min) and was associated with higher odds of EGF (per 60 min, adjusted OR 1.29; 95% confidence interval (CI) 1.02-1.63; p = 0.032). ULS had a lower first-year rejection-treatment burden (≥1 antirejection treatment cycle, 40.7% vs. 63.9%; p = 0.017). Early kidney function profiles favored ULS (lower creatinine at 6 months, 1 year, and 5 years; all p ≤ 0.004; higher hemoglobin at 5 years; p < 0.001). Exploratory time-to-event analyses showed concordant directions for CIT and rejection-treatment burden. In an exploratory univariable landmark analysis, 6-month creatinine levels showed moderate within-sample discrimination for EGF versus ULS (apparent area under the curve [AUC], 0.739).

Overall, CIT showed the most consistent and potentially actionable association with long-term outcomes in this historical cohort; however, inferences are observational, and residual confounding cannot be excluded.