Uraemic burden index: a novel predictor of pre-emptive kidney transplant outcome

Abstract

Timing of pre-emptive kidney transplantation (PKT) and the role of estimated glomerular filtration rate (eGFR) change in outcome prediction remains a subject of debate. This study aimed to assess potential factors, with special attention to uraemic burden, which may be associated with 5-year outcomes. In our retrospective observational cohort study, first PKT adults registered in the CRISTAL database between 2013 and 2019 were analysed to elucidate the role of eGFR and other associating factors with death and graft loss. Recipient-, donor- and transplantation-related features were analysed by using multivariable logistic regression analysis. A conditional inference tree was applied for risk stratification. A total of 2327 first PKT [52.8 years (interquartile range 43-64), 38% female) were included. The mean percentage of PKT over time was 14%. Primary kidney disease (congenital anomalies, glomerulonephritis and other causes versus autosomal dominant polycystic kidney disease), donor age and number of DR mismatches associated with combined 5-year outcomes [odds ratio 2.64 (95% confidence interval 1.42-4.93); 1.94 (1.1-4.93); 1.76 (1.06-2.92); 1.03 (1.02-1.05); 1.67 (1.1-2.53); P < .05], whereas donor type was not associated with outcomes. By supervised decision-tree analysis, >30% risk of failure in PKT was attributed to high recipient risk, higher donor age, uraemic burden index (UBI)-a novel parameter defined by the product of eGFR change and the logarithmic time on the waiting list-and two DR mismatches. In conclusion, eGFR and donor type were not associated with death or graft failure in PKT. UBI can potentially be a novel parameter of uraemic burden and contribute to predict 5-year risk of failure. Clinical decisions based on objective risk estimations might be crucial to approach the 'PKT in due course' concept.

Keywords: biomarker; eGFR; kidney transplantation; outcome; pre-emptive.

Graphical Abstract

Figure 1:

Flow chart of patient selection from the CRISTAL database. *Start of follow-up (T0) is at the time of PKT, the censoring date is the post-transplant fifth year or lost to follow-up. The primary event is the overall graft loss (graft loss or death, whichever comes first).

Figure 2:

(a) Number of transplantations between 2013 and 2018 (P < .05), (b) proportion of PKT per donor type and (c) if PKT versus non-PKT by year and (d) proportion of deceased and living donor PKTs within all transplantations between 2013 and 2019. (e) The number of living and deceased donor PKTs by the year of transplantation (P < .0001)). D: deceased; L: living.

Figure 3:

(a) eGFR at the time of waitlisting and at time of PKT by transplant year. (b) Least square mean of eGFR by years 15.5 at waitlisting versus 11.8 ml/min/1.73 m² at PKT (P < .05) and 2013 versus 2016, 2017 and 2018 (P < .05) and by CKD baseline disease categories [(c) waitlisting: 2 versus 1, 5 or 7 years; P < .05; PKT: P = not significant). Baseline CKD categories: 1- ADPKD, 2- CAKUT, 3- diabetes mellitus, 4- hypertension/RVD, 5- glomerulonephritis, 6- ‘other progressive renal disease’ (e.g. tubulointerstitial diseases, medullar cystic disease, pyelonephritis, tubular dysgenesis, trauma, tumour, Balkan nephropathy, tuberculosis, goutte, nephrocalcinosis, myeloma, amyloidosis, systemic disease, toxic nephropathy, other), 7- unknown or missing.

Figure 4:

UBI, i.e. ΔeGFR * log(time on WL), where ΔeGFR = eGFR at PKT − eGFR at WL. WL: waitlisting.

Figure 5:

eGFR at the time of waitlisting and at the time of PKT by UBI tertiles: (a) eGFR at waitlisting versus PKT (P < .0001) [tertile 1 versus 2 and 3 (P < .001), but P = .053 tertile 2 versus 3] and the time on the waitlist classified by donor type and grouped by UBI tertiles [(b) both P < .0001)]. WL: waitlisting; ms: months.

Figure 6:

Decision tree for overall graft loss (death and failure) in the (a) combined, (b) deceased and (c) living donor PKT group. WL: waitlisting; Recipient: recipient score; MM DR: DR mismatch (0, 1, 2).