Background: There are inconsistent findings regarding the relationship between trough whole blood tacrolimus concentration (TAC C0) and acute kidney rejection in recipients undergoing TAC therapeutic drug monitoring (TDM). However, studies have not always assessed TAC C0 at the time of rejection or accounted for variability in hematocrit. Therefore, this study aimed to investigate the temporal relationship between TAC C0 and acute rejection, including when accounting for variation in hematocrit.
Methods: For 38 recipients who developed biopsy-proven acute rejection (BPAR) in the first 14 days after kidney transplantation, daily TAC C0 from TDM and hematocrit was collected from case notes. Differences in log10-transformed TAC C0 between the day of BPAR (log Cr), 1 day before BPAR (log Cr-1), and 2 days before BPAR (log Cr-2) and the combined median concentrations for the days preceding these (log Cprior) were examined by repeated-measures analysis of variance with Dunnett post hoc testing. Generalized linear mixed-effects regression (glmer) examined the ability of TAC C0 to predict acute rejection episodes with and without controlling for hematocrit.
Results: Log Cr-1 [mean difference (95% confidence interval) = -0.13 (-0.21 to -0.048), post hoc P = 0.002] and log Cr [-0.13 (-0.24 to -0.025), post hoc P = 0.013] were significantly lower than log Cprior. TAC C0 was a significant (P = 0.0078) predictor of rejection episodes (area under the receiver operating characteristic curve = 0.79) only in glmer models accounting for variability in hematocrit.
Conclusions: In recipients who developed BPAR, there was a significant temporal relationship between TAC C0 and BPAR incidence under TAC TDM that may not be detected in cross-sectional studies, especially if variability in hematocrit is not addressed. This supports a TAC C0-rejection relationship, which differs between recipients, and may explain why some recipients do or do not experience rejection within or below the TDM range, respectively. However, studies with larger sample sizes are needed to confirm this finding.