Tacrolimus Exposure Before and After a Switch From Twice-Daily Immediate-Release to Once-Daily Prolonged Release Tacrolimus: The ENVARSWITCH Study

LCP-tacrolimus displays enhanced oral bioavailability compared to immediate-release (IR-) tacrolimus. The ENVARSWITCH study aimed to compare tacrolimus AUC0–24 h in stable kidney (KTR) and liver transplant recipients (LTR) on IR-tacrolimus converted to LCP-tacrolimus, in order to re-evaluate the 1:0.7 dose ratio recommended in the context of a switch and the efficiency of the subsequent dose adjustment. Tacrolimus AUC0–24 h was obtained by Bayesian estimation based on three concentrations measured in dried blood spots before (V2), after the switch (V3), and after LCP-tacrolimus dose adjustment intended to reach the pre-switch AUC0–24 h (V4). AUC0–24 h estimates and distributions were compared using the bioequivalence rule for narrow therapeutic range drugs (Westlake 90% CI within 0.90–1.11). Fifty-three KTR and 48 LTR completed the study with no major deviation. AUC0–24 h bioequivalence was met in the entire population and in KTR between V2 and V4 and between V2 and V3. In LTR, the Westlake 90% CI was close to the acceptance limits between V2 and V4 (90% CI = [0.96–1.14]) and between V2 and V3 (90% CI = [0.96–1.15]). The 1:0.7 dose ratio is convenient for KTR but may be adjusted individually for LTR. The combination of DBS and Bayesian estimation for tacrolimus dose adjustment may help with reaching appropriate exposure to tacrolimus rapidly after a switch.


INTRODUCTION
The pharmacokinetics of LCP-tacrolimus (Envarsus ® ) has been sparsely investigated [1], and clinical trials [2][3][4][5] have left some uncertainty on the exact starting dose, dose ratio with regards to other prolonged-release formulations, and blood levels to be expected in kidney (KTR) and liver transplant recipients (LTR). Previous experience with Advagraf ® showed that absorption could be almost nil in the first days post-transplantation, and that in stable patients, the 1:1 dose ratio resulted in lower C 0 but comparable AUC 0-24 h [6].
The relationship between tacrolimus exposure and effects renders individual dose adjustment essential to avoid underor overexposure [7]. The exposure index best associated with clinical effects is the area under the concentration-time curve (AUC) [7]. To overcome the inconveniences of collecting 10-12 blood samples over the dose interval, Bayesian estimators based on sparse sampling strategies have been developed for the AUC estimation of all tacrolimus formulations [1,[8][9][10][11][12] and are routinely used through the ISBA expert system 1 [13]. However, the collection of several blood samples by venipuncture in a medical environment induces costs and logistical constraints. Therefore, dried blood spot (DBS) sampling, which can easily be performed at home, has been proposed for the therapeutic drug monitoring (TDM) of tacrolimus [14][15][16][17][18][19][20]. After a fingerprick, blood is applied onto a special filter paper, which is subsequently mailed to the laboratory. Good acceptability by the patients [21] and reliability of measured drug levels [16][17][18][19][20] are arguments in favor of DBS for the TDM of tacrolimus in transplantation. Furthermore, DBS are particularly suited to LCP-tacrolimus for which the optimal sampling times for AUC 0-24 h estimation are 0, 8, and 12 h post-dose [1].
In this context, we hypothesized that implementing DBS home sampling for the Bayesian estimation of tacrolimus AUC 0-24 h before and after a conversion, and considering the pre-switch AUC 0-24 h as a reference for LCP-tacrolimus dose adjustment after the switch, would allow maintaining of tacrolimus AUC 0-24 h. Therefore, the aims of the ENVARSWITCH study were to verify, in KTR and LTR, the equivalence of the AUC 0-24 h values before and after a switch from IR-tacrolimus to LCP-tacrolimus at a 1:0.7 dose, followed by individual dose adjustment targeting the pre-switch AUC 0-24 h . The study also aimed to compare tacrolimus exposure indices (AUC 0-24 h , C max and C 0 ) before vs. after the switch, before and after dose adjustment.

Study Design, Patients and Procedures
The ENVARSWITCH study ( The primary objective was to verify the absence of difference between pre-and post-switch tacrolimus AUC 0-24 h calculated by Bayesian estimation, in KTR and LTR switched from IRtacrolimus (Prograf ® ) to LCP-tacrolimus (Envarsus ® ) at a 1: 0.7 dose, possibly followed by individual dose adjustment targeting the pre-switch AUC 0-24 h .
We enrolled adult (≥18 year-old) kidney and liver transplant recipients, transplanted for between 2 weeks and 1 year, in whom a switch from IR-tacrolimus to LCP-tacrolimus had been decided, and in whom the IR-tacrolimus dose had been unchanged for at least 1 week or since the last two C 0 measurements. At the first protocol visit (V1), tacrolimus C 0 had to be between 4 and 12 μg/ L and hematocrit >0. 27.
After inclusion, real-time Bayesian estimation of AUC 0-24 h was performed ( Figure 1): on the day before the switch (V2), after the IR-tacrolimus morning and evening doses (two AUC 0-12 h estimations); 2-4 days after the switch (V3); 7-14 days after V3 (V4). Conversion from IR-tacrolimus to LCP-tacrolimus was done on a 1:0.7 (mg:mg) total daily dose basis. Further dose adjustment could be performed between days 7 and 9, according to the AUC 0-24h estimated at V3, to target the pre-switch AUC 0-24 h calculated by summing the morning and the evening tacrolimus AUC 0-12 h . AUC 0-24 h at V4 was compared to the individual target AUC 0-24 h (V2). No standardized AUC 0-24 h target was considered for the study.

Tacrolimus AUC Determination
AUC 0-24 h was obtained by Bayesian estimation and a limited sampling strategy (pre-dose then 1 h and 3 h post-dose for IRtacrolimus; pre-dose and 8 h then 12 h post-dose for LCPtacrolimus) [1,8,10]. DBS were collected on Whatman ™ 903 protein saver cards. At V2, the study nurses collected the DBS necessary for the determination of IR-tacrolimus morning AUC 0-12 h and trained the patients to collect DBS autonomously. Afterwards, DBS collection was performed at home by the patients. DBS were post-mailed within 24 h after sampling to Limoges University Hospital for centralized analysis. Tacrolimus concentrations were determined using a high performance liquid chromatography-tandem mass spectrometry method on a 4500 AB-Sciex system (Forster City, CA, United States) validated in accordance with the IATDMCT recommendations [22], covering a concentration range of 1-100 μg/L. AUC estimation and the recommended dose were transmitted to the clinicians via a dedicated website within 24 h following DBS reception (maximum 5 days).

Endpoints
The primary endpoint was the Westlake 90%-confidence interval (CI) of the ratio of the dose-adjusted LCP-tacrolimus steady-state AUC 0-24 h (V4) over the pre-switch IR-tacrolimus steady-state AUC 0-24 h (V2) after log-transformation, in the entire population.
Secondary endpoints were the Westlake 90%-CI of the ratio of AUC 0-24 h at V4 over AUC 0-24 h at V2 in KTR and in LTR patients and the differences in and ratios of AUC 0-24 h , C max and C 0 between V2 and V3 in each subgroup.
Renal function was assessed as serum creatinine (SCr) and glomerular filtration rate estimated using the CKD-EPI equation [23]. For regulatory reasons, whenever missing, the eGFR was estimated from SCr by applying the CKD-EPI equation and considering the individuals as "not Black," since there was a very high probability for patients to be of Caucasian or North-African ancestry.
Post hoc analyses were performed to examine, in the entire population and in each subgroup: 1) the correlation between the theoretical LCP-tacrolimus dose (calculated by applying the 1: 0.7 ratio) and the actual dose at V3; 2) the correlation between the LCP-tacrolimus dose proposed after V3 and the actual dose at V4. Doses and exposure indices were also compared between subgroups and periods.

Adverse Events (AEs)
All AEs occurring between enrollment and the end of the trial were recorded on an ongoing basis, regardless of whether they were related or not to IR-tacrolimus or LCP-tacrolimus. Seriousness was assessed according to ICH E2A [24] and severity (mild, moderate, severe) according to its impact on activities of daily life. The causality to the investigational drug was independently assessed by the investigator and the sponsor (worst causality) at the time of the event. All AEs were coded using the MedDRA dictionary (version 23.0).

Statistical Analyses
Statistical analyses were performed using R version 4.0 (R Project for Statistical Computing: 2 ). Categorical data are reported as frequencies and percentages, continuous data as means ± standard deviations (SD). Continuous variables were compared between periods using Student paired-t test.
Data were analyzed for the intent-to-treat population (Full Analysis Set, FAS; KTR and LTR, referred to as "the entire population") and for the per-protocol set (PPS). The FAS comprised included patients who complied with all study visits, while the PPS was restricted to patients of the FAS with no critical protocol deviation. Unless stated, all results are based on the FAS. Safety analyses were based on all included patients.
The comparison of AUC 0-24 h between V2 and subsequent visits was based on the mean ratios between log-transformed AUC 0-24 h and their Westlake 90%-CI. AUC 0-24 h between visits were deemed bioequivalent if the Westlake 90%-CI fell within the 0.90-1.11 range defined by the European Medicine Agency for the bioequivalence of drugs with a narrow therapeutic index [25][26][27][28].
The comparison of the exposure indices between the three periods was done by computing Pearson's coefficient tests, and calculating the mean relative difference and root mean square error (RMSE) of exposure indices at V3 and V4 with respect to those measured at V2.

Sample Size
It was estimated that 96 patients would demonstrate a mean ratio of 1 [90%CI within 0.90-1.11] between V4 and V2 logtransformed AUC 0-24 h , with an expected coefficient of variation = 25% for tacrolimus AUC 0-24 h and 80% power. Anticipating that 10% patients may not meet the requirements of tacrolimus C 0 between 4 and 12 μg/L and hematocrit >0.27, and that 20% may drop out (including missing or poor DBS collection or analysis), the total number of patients to enroll was set to 134.

Patients
Overall, 134 patients (70 KTR and 64 LTR) were enrolled. Three patients did not meet the inclusion criteria at V1 and 30 either discontinued study participation or displayed unexploitable AUC 0-24 h at V2 ( Figure 2). Thus, the FAS comprised 101 patients, of whom 75 constituted the PPS. The KTR and LTR subgroups ( Table 1) were comparable in terms of sex ratio, weight, body mass index and haematocrit, but LTR were characterized by a significantly older age (p = 0.001), later post-transplantation period (p = 0.002), and better kidney function (p = 0.022 for SCr and <0.001 for eGFR).

Tacrolimus Dose and Exposure Indices
AUC 0-24 were distributed normally (Shapiro-Wilk test p between 0.067 and 0.2195). At V2, the mean IR-tacrolimus daily dose was significantly higher in KTR than in LTR (p < 0.001; Table 2), and so were C 0 and AUC 0-24 h (p < 0.001) ( Table 1). The difference on daily dose and AUC 0-24 h of LCP-tacrolimus between KTR and LTR persisted at V4 (p = 0.001 and p < 0.001, respectively).

Compliance With the Recommended Dose
Correlations between the IR-tacrolimus dose at V2 × 0.7 and the LCP-tacrolimus dose at V3, and between the LCP-tacrolimus dose proposed at V3 and the administered dose at V4 were strong in both subgroups (r > 0.9, Figure 5), showing overall good compliance of the clinicians with the doses recommended at all steps.

Impact of Dose Adjustment on AUC 0-24 h
The impact of dose adjustment on AUC 0-24h was evaluated by comparing AUC 0-24 h V4 vs. V2 depending on whether the patients needed dose adjustment after V3 (88 patients) or not (13 patients) and whether dose adjustment was done (53 patients) or not (35 patients) ( Figure 6). No patient benefited from a dose adjustment if no dose adjustment had been proposed. The AUC 0-24 h at V4 and V2 were well correlated in KTR who did not require and did not have a dose adjustment (r = 0.982), but not in LTR in the same situation (r = 0.225). In contrast, among patients for whom we proposed dose adjustment, the correlation was poor in KTR, whether dose adjustment had been applied or not (r = 0.458 and 0.356, respectively) and fair in LTR recipients (r = 0.581 and 0.794, respectively).

Renal Function
No difference in renal function was found between V2 and V4 in the entire population (Scr at V4 = 119 ± 50.2 μmol/L, p = 0.826; eGFR = 62.5 ± 22.3 mL/min, p = 0.974), nor in subgroups separately. The average SCr and eGFR at V4 were, respectively

Safety
The safety analysis set comprised the 134 patients enrolled ( Table 3).
Patients were on IR-tacrolimus for a maximum of 7 days and on LCP-tacrolimus for a maximum of 3 weeks during their participation in the study. Nineteen and fifty AEs occurred respectively while on IR-tacrolimus and LCP-tacrolimus. Eleven patients (8.2%) experienced at least one AE while on IR-tacrolimus and 33 (25.8%) while on LCP-tacrolimus. One patient (0.7%) while on IR-tacrolimus and twelve (9.4%) while on LCP-tacrolimus experienced at least one AE considered as possibly related to tacrolimus. The majority of AEs were of mild-to-moderate severity (100% on IR-tacrolimus and 94% on LCP-tacrolimus). The incidence of tremor, diarrhea, and hyperglycemia on LCPtacrolimus was respectively 3.1%, 1.6%, and 0.8%. Three serious AEs occurred in three patients, among which two were related to tacrolimus: one pneumopathy and one subtherapeutic dosage. This latter AE occurred in a patient who had subtotal colectomy, resulting in a decreased AUC 0-24 at V3 confirmed by a low C 0 value, explained by the lower absorption of tacrolimus in its extended-release formulation. The patient was switched back to IR-tacrolimus and excluded from the study.

DISCUSSION
ENVARSWITCH confirms bioequivalent exposure to tacrolimus in terms of AUC 0-24 h in 101 stable KTR or LTR converted from IR-tacrolimus to LCP-tacrolimus using a 1:0.7 dose ratio. It is the first clinical study proposing the combination of DBS and Bayesian estimation for tacrolimus AUC 0-24 h determination and dose adjustment. The Westlake interval in the entire population fell within the bioequivalence criteria for narrow therapeutic index drugs [26][27][28], and no significant difference was found between the mean AUC 0-24 h before the switch and after the switch followed by individual dose adjustment. Despite the removal of 30/131 (23%) patients from the FAS, the study remained sufficiently powered to validate its primary objective (N > 96 patients). The higher than expected proportion of dropouts was compensated by the lower than expected proportion of patients not meeting the inclusion criteria at V2 (hence not eligible for formulation switching). Also, although the analysis was less powered, the Westlake interval calculated from data of the PPS still fulfilled the bioequivalence criteria. These results suggest that the 1:0.7 dose conversion ratio combined with individual dose adjustment is overall adapted. Importantly, as patients' ethnicity was not collected for regulatory reasons, and because the 1:0.7 conversion factor is not recommended for patients of African origin, we hypothesize that the patients to whom this study was proposed by their treating physician were of other origins, and mostly white Europeans.
The Westlake interval between the AUC 0-24 h measured before and right after the switch also fell within the bioequivalence criteria in the entire population. This suggests that before any individual dose adjustment, the 1:0.7 dose conversion ratio is adapted, as proposed from the conversion studies [2][3][4]. Nevertheless, while subgroup analyses found no difference between AUC 0-24 h at V2 and at V3 and V4 in KTR recipients, the Westlake interval was close to, but did not fall   within, the bioequivalence criteria in LTR. This might partly be due by a lack of power in the subgroup analysis. Additionally, the correlation between the theoretical and actual doses in both contexts of the conversion and dose adjustment proposal was better in KTR than in LTR ( Figure 5). More precisely, the LTR group tended to receive lower doses than those they were supposed to receive, especially for theoretical doses above 5 mg/day, while KTR overall received the theoretical doses. This lack of compliance may partly explain why the Westlake interval did not meet the bioequivalence criteria for narrow therapeutic index drugs in LTR. Still, the correlation between AUC 0-24 h at V4 and at V2 in LTR who did not need and did not have dose adjustment after the conversion was poorer than that observed in KTR in the same situation ( Figure 6). This observation clearly suggests that the 1: 0.7 dose conversion ratio is adequate for KTR patients overall but may need to be slightly decreased and followed by dose adjustment in LTR patients. Still, the Westlake interval fell within the larger acceptance interval [0.8-1.25] recommended by the FDA for bioequivalence studies [29].
The poor correlation between AUC 0-24 h at V2 and at the subsequent visits confirms the wide intra-individual variability in tacrolimus exposure [7], which was unfortunately not compensated for by individual dose adjustment. Given the short time of participation in the study, this variability cannot be attributed to long-term tacrolimus clearance variation observed mainly in patients in the late vs early period after transplantation (after M12 vs. before M1). Correlations between C 0 values at V2 and the subsequent visits were even poorer. Although a decrease in the intra-individual variability of tacrolimus exposure on LCP-tacrolimus vs. IR-tacrolimus could be expected, studies in solid organ transplantation have reported comparable intra-patient variability on C 0 [30,31]. Only one study has reported a significantly lower intra-patient variability of the AUC on LCP-tacrolimus (10.9%) vs. IR-tacrolimus (14.1%) [32]. In any case, the poorer correlation between C 0 values is in favour of considering the AUC 0-24 h rather than C 0 , at least when patients are converted from IR-to LCP-tacrolimus, then at regular time points during follow-up.
The poor correlation between AUC 0-24 h values before the switch and afterwards may also be due to the relatively short time period between the switch and the subsequent AUC 0-24 h measurement. It was ≤3 days in 85/101 patients, so that V3 AUC 0-24 h may not reflect steady-state. This may have led to imprecise or even wrong dose recommendations. Furthermore, steady state may not even have been reached at V4 in all patients, as suggested by the poor correlation between AUC 0-24 h at V2 and V4 in LTR who did not need and did not have tacrolimus dose adjustment ( Figure 6).
Variability may have also come from the use of DBS collected using non-volumetric devices and from the study design, where nurses collected the morning AUC at V2 while the other AUCs were collected by the patients. A comparison of AUC at V3 vs. V4, all sampled by the patients themselves, confirmed the high intraindividual variability (data not shown), dwarfing inter-operator differences as a source of variability. At the time the study was launched, analytical validation data for the measurement of tacrolimus concentrations were available only for the Whatman ™ 903 protein saver cards [14,17,33]. In the meantime, experience has shown that the insufficient standardization of the volume of blood drops contributes to a relative imprecision of concentration measurements [22]. Another potential source of imprecision could have been the hematocrit (varying between 26.2% and 47.0% among patients at V2), as no correction of the analytical results was performed based on the hematocrit. Various patient-centered volumetric micro-sampling devices are now favored for the TDM of immunosuppressants [15,16,22,34]. Twenty-two patients (17%) were withdrawn from the FAS because of unexploitable AUCs, mostly due to non-compliance with sampling times or poor quality of the DBS samples. Yet, training and a user manual had been provided to the healthcare teams and patients. This suggests that using home-based collection of microsamples requires may require even more training for certain patients, in order for them to understand the importance of respecting the sampling schedule, rigorously collect sampling information, and proceed to proper sample collection.
Interestingly, significantly lower exposure was observed in LTR compared to KTR. This may be related to the large C 0 target window at inclusion, allowing liver transplant physicians to target lower C 0 than kidney transplant doctors. This hypothesis is confirmed by the lower daily doses received by LTR compared to KTR ( Table 2; p < 0.001). As the individual AUC 0-24 h at V2 was used as a target for LCP-tacrolimus dose adjustment after V3, the lower exposure in LTR compared to KTR was carried forward throughout the study. Of note, no AUC target has been validated so far for either kidney or liver transplant patients in late periods after transplantation; the only proposed AUC 0-12 h target of 150 h.μg/L [7], was derived from a study performed in 100 kidney transplant patients in the early post-transplantation period [35].
The ENVARSWITCH study used an original approach, where the Westlake interval served to compare the mean exposure obtained with the twice daily IR-tacrolimus vs the once daily LCP-tacrolimus formulation at a 0.7 dose ratio and after dose adjustment. The Westlake interval is generally used in bioavailability studies comparing generic to reference formulations, or newer to reference formulations of brand name drugs for instance. The results obtained here allow for the conclusion that LCPtacrolimus and IR-tacrolimus had bioequivalent AUCs (since C max and T max were not studied), despite the above-mentioned sources of intra-individual variability. This is consistent with previous studies showing that respecting the 1:0.7 dose ratio obviated the need for dose adjustment in the majority of patients [30]. This means that the IR-tacrolimus to LCP-tacrolimus 1: 0.7 conversion dose ratio is appropriate on average, in particular in KTR, but may deserve to be refined in LTR. We calculated that the dose ratio that would zero-in the Westlake interval within the acceptance limits for LTR should be 5% lower, i.e., 1:0.665. However, given the small difference and the impossibility of giving each patient a very precise dose, this option was not considered. The best recommendation would therefore be that tacrolimus exposure should be closely monitored in LTR, preferably based on the AUC 0-24 h , in order to adjust their dose individually, i.e., to compensate for the largest individual exposure differences.
Finally, a relatively low incidence of adverse events was reported in the ENVARSWITCH study. This is related to the short duration of patient participation in the study, especially while on IR-tacrolimus (mean of 6 days between inclusion at V1 and the switch to LCP-tacrolimus).
In conclusion, while the design of the ENVARSWITCH study does not allow comparing therapeutic drug monitoring strategies (AUC or C 0 monitoring from venous blood or DBS samples), its results suggest that the combination of DBS and Bayesian estimation for tacrolimus dose adjustment elicits reaching rapidly appropriate exposure to tacrolimus after the switch from IR-tacrolimus to LCP-tacrolimus. The use of volumetric microsampling devices should further improve the reliability of AUC 0-24 h estimation and individual dose adjustment.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by

AUTHOR CONTRIBUTIONS
CMo, J-BW, SC, and PMa participated in research design, the writing of the paper, the performance of the research, and data analysis. All authors contributed to the article and approved the submitted version.

FUNDING
The authors declare that this study received funding from Chiesi SAS France, Bois-Colombes, France. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.