Increased renal function decline in fast metabolizers using extended-release tacrolimus after kidney transplantation

Fast metabolism of immediate-release tacrolimus (IR-Tac) is associated with decreased kidney function after renal transplantation (RTx) compared to slow metabolizers. We hypothesized, by analogy, that fast metabolism of extended-release tacrolimus (ER-Tac) is associated with worse renal function. We analyzed data from patients who underwent RTx at three different transplant centers between 2007 and 2016 and received an initial immunosuppressive regimen with ER-Tac, mycophenolate, and a corticosteroid. Three months after RTx, a Tac concentration to dose ratio (C/D ratio) < 1.0 ng/ml · 1/mL defined fast ER-Tac metabolism and ≥ 1.0 ng/ml · 1/mL slow metabolism. Renal function (estimated glomerular filtration rate, eGFR), first acute rejection (AR), conversion from ER-Tac, graft and patient survival were observed up to 60-months. 610 RTx patients were divided into 192 fast and 418 slow ER-Tac metabolizers. Fast metabolizers showed a decreased eGFR at all time points compared to slow metabolizers. The fast metabolizer group included more patients who were switched from ER-Tac (p < 0.001). First AR occurred more frequently (p = 0.008) in fast metabolizers, while graft and patient survival rates did not differ between groups (p = 0.529 and p = 0.366, respectively). Calculation of the ER-Tac C/D ratio early after RTx may facilitate individualization of immunosuppression and help identify patients at risk for an unfavorable outcome.

. Study recruitment. 610 patients were included on the basis of taking extended-release tacrolimus (ER-Tac) in the first week after transplantation. Three months after RTx, patients were divided in fast and slow metabolizers with regards to their ER-Tac concentration to dose ratio (C/D ratio). RTx recipients were observed up to 5-years after RTx. www.nature.com/scientificreports/ 10 (D10) and months 1, 2, 3, 6, 12, 24, 36, 48, 60 (M1-M60) after RTx. Creatinine was analyzed in a whole blood sample (enzymatic assay; Creatinine-Pap, Roche Diagnostics, Mannheim, Germany). Calculation of the eGFR was performed using the Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI). In a first approach, eGFR values were compared between patients with a C/D ratio < 1 ng/mL · 1/mg vs. ≥ 1.0 ng/ mL · 1/mg. In the next step, the influence of the M3 C/D ratio on eGFR was evaluated. Therefore, the eGFR changes from each time point to M3 were compared within each connected metabolizer group (eGFR slope) and between both groups. The time to the occurrence of the first event of "switch of immunosuppression", "graft failure", and "death", whichever occurred first, was determined as further subject of the study. "Switch of immunosuppression" was defined as the conversion from ER-Tac to any other immunosuppression. In a first step, the occurrence of "first AR" in the period RTx until M3 was analyzed of patients who received a Tac C/D ratio at M3 and were subsequently characterized as fast or slow metabolizers. In a second step, the impact of the M3 C/D ratio on further "first AR" events was assessed in a 5-year follow-up, when the time to "first AR" was investigated. Switching of immunosuppression, graft failure or death ended follow-up, so patients who switched, restarted dialysis or died without prior AR were censored at the respective date. Any "first AR" was recorded in case of AR treatment or biopsy-proven acute rejection (BPAR) and subsequent treatment. Histologic results on rejections in the transplant centers Cologne and Münster were obtained from indication biopsies only. The transplant center Leuven performed protocol biopsies 3, 12 and 24 months after RTx. The BANFF 2019 criteria were used to define BPAR. CNIT was assess by the local pathologists according to the histological patterns described in detail by Naesens et al. 27 . "Graft failure" was defined as irreversible deterioration of kidney function requiring permanent renal replacement therapy.
Statistical analysis. Statistical analyses were performed using IBM SPSS ® Statistics 27 for Windows (IBM Corporation, Somers, NY, USA) and SAS software, Version 9.4 TS1M5 of the SAS System for Windows (Copyright © 2021 SAS Institute Inc., Cary, NC, USA). All p-values and confidence limits were two-sided and were intended to be exploratory, not confirmatory. Therefore, no adjustment for multiplicity was made. Exploratory p-values ≤ 0.05 were considered to be statistically noticeable.
In descriptive analysis, normally-distributed continuous variables are reported as mean ± standard deviation and not normally-distributed continuous variables as median (25% quantile-75% quantile, IQR). Absolute and relative frequencies are given for categorical variables. Metabolism groups were compared using Welch's t-tests for normally-distributed data, Mann-Whitney U-tests for skewed-distributed data, and Fisher's exact tests for categorical variables. Comparison of the eGFR changes within each metabolism group was performed using Wilcoxon's signed-rank tests. Boxplots were used for graphical representation.
In order to model renal function (eGFR) over time adjusted for co-variables and dropouts over time, a multivariable linear mixed model was fitted. Recipient's age, European Senior Program (ESP) (yes/no), the main effects of time (3,6,12,24,36,48, 60 months) and metabolizer group (fast/slow), and the interaction between time and group were included as influencing variables. Repeated measurements of each patient were modeled using SAS PROC MIXED by fitting a marginal linear mixed model with an unstructured variance-covariance matrix for the residuals with patient as subject and the order given by time. The empirical sandwich estimator was applied. Missing values were treated as missing at random. Results are reported as least square estimates with corresponding 95% confidence interval (CI), and p-values from the Wald test.
Event rates of the "first AR", of the component endpoint "switch of the immunosuppression", "graft failure" and "death as first event" (whatever occurred first), as well as of overall survival were estimated using the Kaplan-Meier method with 95% CI using log-transformation. Patients without an event were censored at their last visit date. In the analysis of "first AR", patients who switched immunosuppression, showed a graft failure or died without prior AR were additionally censored at the respective date. All time-to-event endpoints started at 3 months after RTx. Consequently, AR and graft failures which occurred until 3 months were not considered in the analysis and patients were excluded from these analyses. Hazard ratios (HR) and 95% CI are given to quantify the effect of fast versus slow metabolizers. A competing risk approach was used to estimate the effect on each component of the combined endpoint. The impact of fast vs. slow metabolizer on the components was thus estimated using Fine and Gray's model leading to subdistribution hazard ratios (sub-HR). Cumulative incidence was estimated using the Aalen-Johansen estimator. Gray's k-sample test was applied to compare the cumulative incidence of the corresponding event type. Additionally, the cause-specific hazard (csH) for each component of the competing event type was compared between the metabolizer groups via cause-specific hazard ratios (csHR) using the methods by Prentice (results not shown).
The RTx cohort showed a skew distribution of the ER-Tac C/D ratio at 3 months with a median of 1.3 (IQR 0.86-1.88) ng/mL · 1/mg. Table 1 shows the baseline characteristics of the patients. Fast ER-Tac metabolizers were younger (p = 0.003) although more patients within the ESP (p = 0.006) received a graft, albeit with low absolute ESP numbers. There were no other noticeable differences in patient characteristics.
Immunosuppression. Compared to slow metabolizers, the group with a C/D ratio < 1.0 ng/mL · 1/mg received higher ER-Tac doses at 1, 3 and 6 months after RTx (all p < 0.001), and even had slightly lower Tac www.nature.com/scientificreports/ trough levels at month 1 and 3 (both p < 0.001; Table 2). Median Tac blood concentrations were within the target trough level at all three time points. In the Leuven study center, fast metabolizers received slightly more methylprednisolone than slow metabolizers at month 1 after RTx (12 (IQR 9.5-16.0) mg vs. 12 (IQR 8.0-12.0) mg; p = 0.006).

Renal function.
As early as D10, renal function of the fast ER-Tac metabolizers was slightly decreased compared to the slow metabolizers and reached a noticeable difference at M2 (p = 0.004). This difference persisted over the entire time period. (Fig. 3a). In a further step, we analyzed whether the M3 C/D ratio had an influence on the subsequent eGFR change. In contrast to the different ΔeGFR development from D10 to M3 after RTx (p = 0.0362), the subsequent ΔeGFR changes from the M3 eGFR did not differ between both groups (Fig. 3b). However, 5 years after RTx the eGFR in patients with a C/D ratio < 1 ng/mL · 1/mg was lower than in patients with ≥ 1 ng/mL · 1/mg (45.6 ± 19.4 vs. 51.5 ± 19.6 mL/min/1.73 m 2 ; p = 0.039).
The results in the linear mixed model were similar to the univariate analyses ( Table 3). The mean eGFR in fast metabolizers was 7.5 (95% CI 4.3-10.6, p < 0.001) mL/min/1.73 m 2 lower pooled over all time points (M3-M60). eGFR changed in the whole collective from M3 (p < 0.001) and the changes between both metabolism groups were slightly different (interaction term, p = 0.037). Age at RTx (p < 0.001) had a further potential influence on the renal function. Despite the small p-value the additional effect of age was quite low. Per year of age older the eGFR value declined by -0.4 (95% CI − 0.5 to − 0.3) mL/min/1.73 m 2 . The influence of transplantation in the ESP indicated only a trend (p = 0.084). In these patients, the eGFR was − 5.7 (95% CI − 12.2 to 0.8) mL/min/1.73 m 2 lower than in patients who received an organ in the regular allocation system.

Time-to-event endpoints.
Consideration of the combined endpoint of switch/graft failure/death showed that fast metabolizers had more and earlier events than slow metabolizers over the 5-year follow-up (hazard ratio 1.51 (95% CI 1.07-2.14), Fig. 4a). Competing risk analysis indicated that crucial to these differences was a higher cumulative incidence of fast metabolizers who were switched from ER-Tac to other immunosuppression (p < 0.0001, Table 4, Fig. 4d). In detail, more fast metabolizers were switched from ER-Tac to IR-Tac (p = 0.002) or everolimus (p = 0.021). The main reasons for the change were CNIT (< 0.001) and large variations in Tac trough concentrations (p = 0.036) compared to slow metabolizers. The cumulative incidence of the component "graft failure" of the combined endpoint switch/graft failure/death was not different between the groups (p = 0.562, Fig. 4c). One slow metabolizer died without prior switch or graft failure. A difference in overall survival was not observed (p = 0.320, Fig. 4b). However, analysis of "first AR" events (after M3) showed that fast ER-Tac metabolizers developed more events compared to slow metabolizers (p = 0.0077, Fig. 5, Table 5). In a further analysis, no differences were found in the histological AR subtypes. Considering the period from RTx to M3, fast metabolizers also suffered more frequently from "first AR" than slow metabolizers (24.5% vs. 16.5%; p = 0.026, Table 5).

Discussion
Because IR-Tac and ER-Tac have comparable safety and efficacy, we hypothesized that a C/D ratio-guided analysis would be able to stratify patients treated with ER-Tac analogously to patients treated with IR-Tac with respect to their Tac metabolism.

Figure 2.
Empirical distribution of the patients in terms of their concentration to dose ratio (C/D ratio) three months after RTx. Fast ER-Tac metabolizers were defined by a C/D ratio < 1 ng/mL · 1/mg, and slow metabolizers had a C/D ratio ≥ 1 ng/mL · 1/mg. www.nature.com/scientificreports/ Previous analysis of IR-Tac metabolism groups after RTx and liver transplantation (LTx) defined by a C/D ratio cut-off showed repeatedly and consistently worse renal outcomes in fast metabolizers compared to slow metabolizers 16,17,21,22,28 . Other studies without C/D ratio calculation, but dose and trough data pointed into the same direction 29,30 . Interestingly, even studies who included patients with IR-Tac and ER-Tac observed comparable outcomes to pure IR-Tac studies 18,19 . An analysis of the Korean Organ Transplantation Registry did not provide information regarding the used Tac formulation but observed the same findings 31 . Further evidence comes from PK profile analyses and assumptions. When analyzing the PK profiles of RTx patients treated with ER-Tac, one can observe the same mixed PK pattern with different profile types (probably related to the Tac metabolic type of the patients) as in IR-treated RTx recipients 32,33 . Interestingly, the C max was also not significantly different or only slightly lower than that of patients treated with IR-Tac and was comparable in healthy subjects 34,35 . This is important, as we found high C max associated with a higher degree of CNIT 22 . Thus, there was evidence from the literature suggesting a similar C/D ratio-dependent effect in IR-Tac and ER-treated patients.
We have calculated a C/D ratio cutoff of 1.0 ng/mL · 1/mg for the differentiation of metabolism types in ER-Tac treated patients. This fits well with the previously calculated cut-offs 1.05 ng/mL · 1/mg for IR-treated RTx and 1.09 ng/mL · 1/mg for LTx patients 16,17,21 . In addition, the TOMATO study, which included a hybrid cohort of 1029 IR-and ER-Tac patients, and which also used a C/D cutoff of 1.05 ng/mL · 1/mg, found that, on the one hand, the C/D ratio was stable between months 6-12 after RTx. On the other hand, a hazard ratio of 2.25 could be calculated for death-censored graft loss in patients with a C/D ratio < 1.05 ng/mL · 1/mg in contrast to the CYP3A5 genotyping which was not predictive 18 . Unfortunately, the authors had no data on eGFR in the www.nature.com/scientificreports/ follow-up. A reduced survival as well as increased incidence of graft failure in fast metabolizers is consistent to our IR-Tac data 21 , in contrast, we could not observe the same differences in the present study (Fig. 4b,c). This may be related to the fact that during follow-up, 17.0% of patients in the fast ER-Tac metabolizer group were switched to an alternative immunosuppressive regimen, compared to only 6.6% in the slow metabolizer group (p < 0.0001, Table 4, Fig. 4d). The main reason for switching from ER-Tac was CNIT, which was observed more frequently in the group with the C/D ratio < 1.0 ng/mL · 1/mg. This is in congruence with our previous studies on IR-Tac and the observation that C max could be critical for its development 16,21,22,36 . IR-and ER-Tac show similar PK in terms of C max and their narrow therapeutic window. Accordingly, adverse effects such as CNIT are common and may occur in similar frequency, especially in fast metabolizers who have higher C max than slow metabolizers-at least in IR-Tac treated patients 22 . One limitation of our present study is that we do not have C max data. Interestingly, the major difference in eGFR between both groups developed within 3 months after RTx. We already knew from former studies that CNIT can occur early after RTx, especially in patients treated with high Tac dosages 29,37 . Recently, in de novo ER-Tac-treated patients, it was shown that a C/D ratio < 1.05 ng/mL · 1/mg and the presence of interstitial fibrosis and tubular atrophy (IF/TA) in a biopsy 3 months after RTx were associated with future IF/TA progression 38 . Of note, the authors observed that unlike Tac trough levels, intrapatient variability, or time below the therapeutic range, just a low C/D ratio was associated with IF/TA progression. This is congruent with data from Egeland et al. who identified fast Tac metabolism as a risk factor for the development of IF/TA 39 . It is important to note in this context that both Tac formulations have high intra-and inter-individual coefficients of variation for AUC, C min and C max 40,41 . Recently, this high intra-patient variability (IPV) in Tac trough levels especially early after RTx has been associated with rejection, de novo donor-specific antibodies (DSA), progressive IF/TA and reduced graft survival 42 . One explanation could be that patients with high IPV are outside the therapeutic range for extended periods of time since e.g. a single missed Tac dose can greatly affect exposure especially in recipients with fast metabolism 43 . It has been argued that switching patients from IR-Tac to ER-Tac may be beneficial in terms of IPV, but the data remain inconclusive and clinically limited 44,45 . Consistent with these observations, patients with a C/D ratio < 1.0 ng/mL · 1/mg suffered more AR events than patients with a higher C/D ratio (Table 5, Fig. 5). Similar observations have been made in patients with IR-Tac-treated patients 21,23 .
Our study has limitations. Since this is a retrospective study, it can only generate new hypotheses. Moreover, we did not assess C max and coefficient of variation/IPV in our patients. Therefore, we can only assume effects of fast ER-Tac metabolism (low C/D ratio) of renal outcomes. Further, more patients in the group with a C/D ratio < 1.0 ng/mL · 1/mg were switched mainly due to side effects to an alternative immunosuppression. Thus, we cannot rule out completely that this might have affected the outcomes. However, since we have previously shown that a conversion from Tac to everolimus can be safely done in selected patients with e.g. CNIT this could  www.nature.com/scientificreports/ have stabilized some fast metabolizers and reduced the difference in eGFR loss 36 . Donor factors may influence graft outcome. Unfortunately, donor eGFR data are not available to us. However, since the donors did not differ significantly between the groups with respect to various factors such as age, BMI, and KDPI, and the rate of DGF was comparable, we do not expect major differences in the eGFR of the donors.

Conclusion
We conclude from our data that patients treated with ER-Tac who have a C/D ratio < 1.0 ng/mL · 1/mg showed a decreased eGFR, especially early after RTx. This was mainly associated with CNIT and a higher rate of first AR, leading to more frequent changes in immunosuppression in these patients. Thus, this group of patients has an increased risk of developing a worse outcome after RTx and therefore deserves special attention. Because it is very simple and without cost, we propose to calculate the C/D ratio early after RTx in patients treated with ER-Tac. Switching rapid ER-Tac metabolizers to other Tac formulations, such as LCP-Tac, or other immunosuppressive agents, such as belatacept or everolimus, is safe and may be beneficial, taking into account the immunologic and metabolic profile of the individual. www.nature.com/scientificreports/ Figure 4. Kaplan-Meier curves of the composite endpoint "switch from ER-Tac", "graft failure", or "death" as first event by metabolism group starting from three months after RTx (a). Overall survival is shown in (b). Cumulative incidence of the "graft failure" component (c) and "switch from ER-Tac" (c) of the composite endpoint. Since only one death without prior switch or graft failure occurred in the slow metabolism group, the curves were not drawn. The impact of fast vs. slow metabolizer was thus estimated using Fine and Gray's model leading to subdistribution hazard ratios (sub-HR) on the components of the composite endpoint. Cumulative incidence was estimated using the Aalen-Johansen estimator. Gray's k-sample test was applied to compare the cumulative incidence of the corresponding event type. The combined endpoint "switch/graft failure/death" showed more events in the fast metabolizer group. Competing risk analysis revealed that "switch from ER-Tac" occurred more frequently in fast than in slow metabolizers, but no differences were found in regards to graft failure or overall survival.   Figure 5. Time to "first acute rejection" (AR) from 3 months after transplantation. Patients who switched immunosuppression, showed a graft failure or died without prior AR were censored at the respective date. Patients with an AR between transplantation and 3 months were excluded (n = 116). Fast ER-Tac metabolizers showed more first rejections compared to slow metabolizers within 5-years after transplantation.