Fast Tacrolimus Metabolism Does Not Promote Post-Transplant Diabetes Mellitus after Kidney Transplantation

Post-transplant diabetes mellitus (PTDM) after kidney transplantation induced by tacrolimus is an important issue. Fast tacrolimus metabolism, which can be estimated by concentration-to-dose (C/D) ratio, is associated with increased nephrotoxicity and unfavorable outcomes after kidney transplantation. Herein, we elucidate whether fast tacrolimus metabolism also increases the risk for PTDM. Data from 596 non-diabetic patients treated with tacrolimus-based immunosuppression at the time of kidney transplantation between 2007 and 2015 were retrospectively analyzed. The median follow-up time after kidney transplantation was 4.7 years (IQR 4.2 years). Our analysis was complemented by experimental modeling of fast and slow tacrolimus metabolism kinetics in cultured insulin-producing pancreatic cells (INS-1 cells). During the follow-up period, 117 (19.6%) patients developed PTDM. Of all patients, 210 (35.2%) were classified as fast metabolizers (C/D ratio < 1.05 ng/mL × 1/mg). Fast tacrolimus metabolizers did not have a higher incidence of PTDM than slow tacrolimus metabolizers (p = 0.496). Consistent with this, insulin secretion and the viability of tacrolimus-treated INS-1 cells exposed to 12 h of tacrolimus concentrations analogous to the serum profiles of fast or slow tacrolimus metabolizers or to continuous exposure did not differ (p = 0.286). In conclusion, fast tacrolimus metabolism is not associated with increased incidence of PTDM after kidney transplantation, either in vitro or in vivo. A short period of incubation of INS-1 cells with tacrolimus using different concentration profiles led to comparable effects on cell viability and insulin secretion in vitro. Consistent with this, in our patient, collective fast Tac metabolizers did not show a higher PTDM incidence compared to slow metabolizers.

Unfortunately, Tac has a narrow therapeutic range. Due to the high intra-and interindividual pharmacokinetics and pharmacodynamics of Tac, correct dose adjustment can be difficult [3]. Dependent on the concentration-to-dose (C/D) ratio, a slow (C/D ratio ≥ 1.05 ng/mL × 1/mg) or fast (C/D ratio < 1.05 ng/mL × 1/mg) Tac metabolism of treated individuals can be differentiated. This could be important because it has been shown by us and others that fast Tac metabolism is significantly associated with unfavorable renal allograft outcomes [4][5][6]. Although the exact mechanisms are still being researched, it is clear that Tac has direct toxic effects on pancreatic ß-cells and is therefore diabetogenic [4]. Nevertheless, these toxic effects are linked to pre-existing ß-cell-dysfunction, which might 2 of 9 be induced by glucolipotoxicity [7]. Tac might act as an accelerator of pre-existing β-cell damage [8]. In rats, high-dose Tac treatment (10 mg/kg/day) decreased insulin mRNA transcription and reduced insulin production due to high levels of FK506-binding protein-12 in pancreatic β-cells [9]. Tong et al. found that high Tac concentrations inhibited insulin release and induced apoptosis in ß-cells [10]. Furthermore, CNI-induced hypomagnesemia is associated with post-transplant diabetes mellitus (PTDM), although the underlying mechanisms are still unknown [11]. Tac is therefore one important driver for PTDM. PTDM is frequently observed after KTx and is also associated with inferior outcomes such as infections and cardiovascular events [8]. The cellular mechanisms that lead to PTDM resemble those leading to type 2 diabetes mellitus to a considerable extent [12]. Risk factors for PTDM include obesity, dyslipidemia and insulin resistance [8].
We hypothesized that fast Tac metabolizers are at higher risk for PTDM due to the higher CNI toxicity of pancreatic ß-cells, since this is also true for renal vascular endothelial and tubular cells [13]. To investigate the effects of fast Tac metabolism on the development of PTDM, we performed a retrospective data analysis of 679 patients on the one hand. On the other hand, we modeled different tacrolimus exposure kinetics in slow and fast metabolizers via cell culture experiments in insulin-secreting pancreatic cells (INS-1 rat pancreatic insulinoma cell line).

Retrospective Analysis of Patient Cohort
Of the 679 patients, 83 (12.2%) already had diabetes prior to KTx and were therefore excluded.
A binary logistic regression confirmed BMI (p = 0.015) and recipients' age (p < 0.001) as significantly associated with PTDM. The C/D ratio, which indicates Tac metabolism type, was not significantly associated with PTDM (p = 0.859) ( Table 3).

Experimental Results
To determine whether INS-1 cell insulin secretion is affected by the Tac concentration profile used, INS-1 cells were exposed to a continuous Tac concentration and concentration profiles representing slow and fast Tac metabolism over 12 h (see Materials and Methods section for details).
The insulin secretion of INS-1 cells exposed to Tac in different concentration profiles (fast vs. slow vs. continuous serum profile) was comparable in all three groups (Continuous vs. Fast p = 0.323, Continuous vs. Slow p = 1.000, Slow vs. Fast p = 0.279, Figure 2).
The MTT assay was performed to ensure that cell viability did not differ between the Tac groups. It showed the comparable viability/metabolic activity of treated INS-1 cells between the three groups ( Figure 3). As a control, we used INS-1 cells exposed to standard medium without Tac. Parallel to the Tac-treated groups, we changed medium every hour.
The incidence of PTDM was not statistically different between slow and metabolizers (Log-rank p = 0.496, Figure 1).  A binary logistic regression confirmed BMI (p = 0.015) and recipients' age (p < 0.001) as significantly associated with PTDM. The C/D ratio, which indicates Tac metabolism type, was not significantly associated with PTDM (p = 0.859) ( Table 3).

Experimental Results
To determine whether INS-1 cell insulin secretion is affected by the Tac concentration profile used, INS-1 cells were exposed to a continuous Tac concentration and concentration profiles representing slow and fast Tac metabolism over 12 h (see Materials and Methods section for details).
The insulin secretion of INS-1 cells exposed to Tac in different concentration profiles (fast vs. slow vs. continuous serum profile) was comparable in all three groups (Continuous vs. Fast p = 0.323, Continuous vs. Slow p = 1.000, Slow vs. Fast p = 0.279, Figure 2). The MTT assay was performed to ensure that cell viability did not differ between the Tac groups. It showed the comparable viability/metabolic activity of treated INS-1 cells between the three groups ( Figure 3). As a control, we used INS-1 cells exposed to standard medium without Tac. Parallel to the Tac-treated groups, we changed medium every hour.

Discussion
We found no evidence in patient and cell culture data that a fast Tac metabolism, depicted by C/D ratio, is associated with increased rates of PTDM. Fitting to that, neith Tac trough levels nor daily Tac dose was statistically associated with the development PTDM.
Based on the knowledge that the underlying mechanism of Tac-associated PTDM supposed to directly affect insulin secretion in pancreatic β-cells rather than inducing i sulin resistance as an underlying mechanism for Tac-associated PTDM [9,10], we modele the effects of slow and fast Tac metabolism on insulin-producing pancreatic cells (INS cells) in vitro. The kinetics of Tac concentration used in cell culture experiments was d termined in a former study showing that fast Tac metabolism is associated with increase nephrotoxicity in vitro and in vivo, respectively, and affected allograft function and all graft survival [4,14]. Because CNI toxicity is strongly related to the administered Tac do relative to the trough level (C/D ratio) and thus to higher peak Tac levels, a higher risk Tac renal toxicity in fast metabolizers is understandable [15,16]. However, the toxic actio of Tac is not limited to renal cells and function, but, amongst others, is also apparent pancreatic β-cells [10,12]. One mechanism for the diabetogenic effect of Tac might be th inhibition of the calcineurin-nuclear factor of activated T-cells (NFAT) pathway. This i hibition leads to the dysregulation of the insulin gene and alters β-cell mass, which w demonstrated in mice [17]. In contrast to the CNI cyclosporine A, Tac augments diabet genic effects in patients with metabolic syndrome and present glucolipotoxicity [18]. Co sidering that, we treated INS-1 cells for our experiments with glucose and palmitate befo exposure to Tac according to Trinanes et al. [19].
Because insulin release and the viability of INS-1 cells treated with different Tac pr tocols did not differ, peak Tac concentration does not seem to be a driving force for PTDM

Discussion
We found no evidence in patient and cell culture data that a fast Tac metabolism, as depicted by C/D ratio, is associated with increased rates of PTDM. Fitting to that, neither Tac trough levels nor daily Tac dose was statistically associated with the development of PTDM.
Based on the knowledge that the underlying mechanism of Tac-associated PTDM is supposed to directly affect insulin secretion in pancreatic β-cells rather than inducing insulin resistance as an underlying mechanism for Tac-associated PTDM [9,10], we modeled the effects of slow and fast Tac metabolism on insulin-producing pancreatic cells (INS-1 cells) in vitro. The kinetics of Tac concentration used in cell culture experiments was determined in a former study showing that fast Tac metabolism is associated with increased nephrotoxicity in vitro and in vivo, respectively, and affected allograft function and allograft survival [4,14]. Because CNI toxicity is strongly related to the administered Tac dose relative to the trough level (C/D ratio) and thus to higher peak Tac levels, a higher risk of Tac renal toxicity in fast metabolizers is understandable [15,16]. However, the toxic action of Tac is not limited to renal cells and function, but, amongst others, is also apparent in pancreatic β-cells [10,12]. One mechanism for the diabetogenic effect of Tac might be the inhibition of the calcineurin-nuclear factor of activated T-cells (NFAT) pathway. This inhibition leads to the dysregulation of the insulin gene and alters β-cell mass, which was demonstrated in mice [17]. In contrast to the CNI cyclosporine A, Tac augments diabetogenic effects in patients with metabolic syndrome and present glucolipotoxicity [18]. Considering that, we treated INS-1 cells for our experiments with glucose and palmitate before exposure to Tac according to Trinanes et al. [19].
Because insulin release and the viability of INS-1 cells treated with different Tac protocols did not differ, peak Tac concentration does not seem to be a driving force for PTDM, unlike in nephrotoxicity. Notably, trough levels, daily Tac dose and C/D ratio were not different between patients who developed PTDM and those who did not (Table 2).
Rather, the development of PTDM in the clinical setting is determined by older age and higher BMI, as is known from other studies in KTx patients [20] and is also reproducible in our binary logistic regression model (Table 3). All patients in this study were medicated with steroids as one component of the triple immunosuppressive regimen in a standard maintenance dosage of 5 mg per day after 100 mg prednisolone for 3 days initially and 20 mg per day within the first month after KTx. Nevertheless, possible steroid pulses as therapy for allograft rejection were not taken into account. Since it has been demonstrated that KTx recipients with fast Tac metabolism are at higher risk of allograft rejection [4], it can be assumed that fast metabolizers even tended to be exposed to higher steroid dosages than slow metabolizers.
This study entails several further limitations. First, it cannot completely be excluded that some of the patients with PTDM already had undiagnosed diabetes before KTx. Because we used routine clinical data, we do not have data on the Tac peak level of patients. However, we know from a former study that fast metabolizers develop higher peak concentrations [14]. Furthermore, a retrospective data analysis on PTDM can only be hypothesis-generating. Regarding the experimental part of this study, the experiments did not model long-term effects of Tac exposure on pancreatic cells.

Study Cohort
We retrospectively analyzed a total of 679 KTx adult patients receiving an immunosuppressive regimen containing immediate-release (IR) or extended-release (ER) Tac with respect to the C/D ratio and classified them into slow versus fast Tac metabolizers based on the 3-month C/D ratio, as investigated in former studies (slow: C/D ratio ≥ 1.05 ng/mL × 1/mg, fast: C/D ratio < 1.05 ng/mL × 1/mg) [4,21]. Six hundred and forty-seven (95.3%) of the patients received IR Tac and thirty-one (4.7%) ER Tac. As well as Tac, the immunosuppressive triple regimen contained mycophenolate and steroids. The induction therapy was chosen according to the immunological risks of the patients.
From this cohort, 117 (17.2%) patients developed PTDM within the follow-up period (median 4.7 years (IQR 4.2 years)); a further 83 (12.2%) patients already suffered from diabetes before transplantation and were therefore excluded from further analyses. Moreover, we excluded from the study recipients < 18 years of age, pregnant women and patients with uncontrolled infection or active malignancy. PTDM was defined as persistent hyperglycemia that did not resolve within the first year after kidney transplantation in patients who did not suffer from diabetes before transplantation [22]. PTDM was defined based on the criteria for defining diabetes mellitus type 2 in the general population (ADA, WHO).
Written informed consent was obtained from all patients to record their data at the time of KTx. Data was taken from patients' clinical files and personal information was anonymized prior to the analysis. This study was performed in accordance with the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines and approved by the local ethics committee (EthikKommission der Ärztekammer Westfalen-Lippe und der MedizinischenFakultät der WestfälischenWilhelms-Universität, 2014-381-f-N).
INS-1 cells were grown in 12-well or 96-well plates until 80% confluence followed by treatment with Tac (Prograf ® i.v., Astellas, Munich, Germany) diluted in medium or with medium only as a control over 12 h. Tac working solutions were freshly prepared by appropriate dilution of stock solution in the culture medium [18]. According to the work of Rodriguez-Rodriguez et al., who showed that the toxicity of TAC depends on the pre-existence of β-cell dysfunction, we modeled glucolipotoxicity to the INS cells with glucose (22 mmol/L) and palmitate (100 µmol/L) for 5 days before we started the 12 h Tac time trials [5,8].
For 12 h time trials with exposure to the different Tac kinetic profiles, we applied Krebs-Ringer buffer (KRB) medium (own completion).
The culture medium of 12-well plates was changed every hour using the indicated Tac concentrations between 6 and 15 µg/mL that were based on our previous study [14], as shown in Table 4.
Binary logistic regression analysis was applied to identify potential independent predictors of PTDM development.

Conclusions
In conclusion, fast Tac metabolism, defined by a C/D ratio of <1.05 ng/mL × 1/mg, in contrast to kidney function, does not impair pancreatic β-cell function to a greater extent than Tac slow metabolism, neither in a patient collective of KTx patients nor in an experimental setting.
Author Contributions: U.J. conducted research and wrote the paper; N.W. conducted research; G.R.B. collected data and revised the paper; G.T. conducted research and revised the paper; H.P. revised the paper; S.R. designed the research and wrote the paper. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.